)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":7715,"name":"Hirofumi Ichihara","email":"ichihara.hirofumi@gmail.com","username":"Hirofumi-Ichihara"},"change_message_id":"3a16e1417ceea928a0088f82a3cee3633514ad14","unresolved":false,"context_lines":[{"line_number":7,"context_line":"Add instrumentation devref"},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"Presents what instrumentation is available from VIFs in Nova,"},{"line_number":10,"context_line":"Metering Lables and Rules, Linux Bridge, and OVS. Proposes"},{"line_number":11,"context_line":"mappings for structures defined in RFC 2863 and RFC 4293 as"},{"line_number":12,"context_line":"well as how to communicate and aggregate harvested counters"},{"line_number":13,"context_line":"from agents to Ceilometer."}],"source_content_type":"text/x-gerrit-commit-message","patch_set":19,"id":"da20952f_8acb8571","line":10,"updated":"2015-09-08 00:36:14.000000000","message":"Lables -\u003e Labels?","commit_id":"054506810cc066f984333703ba753f060080b96e"},{"author":{"_account_id":6854,"name":"YAMAMOTO Takashi","email":"yamamoto@midokura.com","username":"yamamoto"},"change_message_id":"fdba8029f7abaaa59c6d78ac4a19b8c2161d7770","unresolved":false,"context_lines":[{"line_number":7,"context_line":"Add instrumentation devref"},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"Presents what instrumentation is available from VIFs in Nova,"},{"line_number":10,"context_line":"Metering Lables and Rules, Linux Bridge, and OVS. Proposes"},{"line_number":11,"context_line":"mappings for structures defined in RFC 2863 and RFC 4293 as"},{"line_number":12,"context_line":"well as how to communicate and aggregate harvested counters"},{"line_number":13,"context_line":"from agents to Ceilometer."}],"source_content_type":"text/x-gerrit-commit-message","patch_set":24,"id":"ba15a1d1_1fe059f9","line":10,"range":{"start_line":10,"start_character":9,"end_line":10,"end_character":15},"updated":"2015-09-17 09:52:50.000000000","message":"typo: Labels","commit_id":"f8dcbac1b1320a61b0a8f36707784e98f4cce655"}],"doc/source/devref/index.rst":[{"author":{"_account_id":7776,"name":"Paul Carver","email":"pcarver@att.com","username":"pcarver"},"change_message_id":"1676920ae6aaab55eec4563a7f87c0c0b4cc00c7","unresolved":false,"context_lines":[{"line_number":15,"context_line":"      License for the specific language governing permissions and limitations"},{"line_number":16,"context_line":"      under the License."},{"line_number":17,"context_line":""},{"line_number":18,"context_line":"      Convention for heading levels in Neutron devref:"},{"line_number":19,"context_line":"      \u003d\u003d\u003d\u003d\u003d\u003d\u003d  Heading 0 (reserved for the title in a document)"},{"line_number":20,"context_line":"      -------  Heading 1"},{"line_number":21,"context_line":"      ~~~~~~~  Heading 2"}],"source_content_type":"text/x-rst","patch_set":14,"id":"1a4dcd0f_89cff856","line":18,"updated":"2015-08-13 21:25:15.000000000","message":"I don\u0027t have an objection to this formatting tip, but it seems unrelated to the commit message. Did this slip in by accident or did you intend to piggyback it in?","commit_id":"5f35e3c9294a3e1a34102209d79704ad47882e24"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"3ca19c2982fbe8a71533c346f912144b32b52c64","unresolved":false,"context_lines":[{"line_number":15,"context_line":"      License for the specific language governing permissions and limitations"},{"line_number":16,"context_line":"      under the License."},{"line_number":17,"context_line":""},{"line_number":18,"context_line":"      Convention for heading levels in Neutron devref:"},{"line_number":19,"context_line":"      \u003d\u003d\u003d\u003d\u003d\u003d\u003d  Heading 0 (reserved for the title in a document)"},{"line_number":20,"context_line":"      -------  Heading 1"},{"line_number":21,"context_line":"      ~~~~~~~  Heading 2"}],"source_content_type":"text/x-rst","patch_set":14,"id":"1a4dcd0f_d19906d2","line":18,"in_reply_to":"1a4dcd0f_89cff856","updated":"2015-08-14 12:56:23.000000000","message":"This came in via a rebase","commit_id":"5f35e3c9294a3e1a34102209d79704ad47882e24"}],"doc/source/devref/instrumentation.rst":[{"author":{"_account_id":9970,"name":"John Kasperski","email":"jckasper@linux.vnet.ibm.com","username":"jckasper"},"change_message_id":"dedcc1f2bf1c94b83ab7faca83afedbfc821b1b1","unresolved":false,"context_lines":[{"line_number":3,"context_line":""},{"line_number":4,"context_line":"This section discusses instrumentation of neutron.  For the first pass of"},{"line_number":5,"context_line":"adding instrumentation, cataloguing information that can be used to fill"},{"line_number":6,"context_line":"in the data structures defined by [RFC2863]_, [RFC4293]_, [RFC4022]_ and [RFC4113]_.  "},{"line_number":7,"context_line":""},{"line_number":8,"context_line":".. [RFC2863] https://tools.ietf.org/html/rfc2863"},{"line_number":9,"context_line":".. [RFC4293] https://tools.ietf.org/html/rfc4293"}],"source_content_type":"text/x-rst","patch_set":2,"id":"3a50d1a3_3024455c","line":6,"updated":"2015-07-23 05:09:04.000000000","message":"Remove trailing white space.   :-)","commit_id":"447149cdeabfdaf1c70b0a8b061c10457217d69d"},{"author":{"_account_id":4656,"name":"Sean M. Collins","email":"sean@coreitpro.com","username":"scollins"},"change_message_id":"9ed8f314c8650d112c7ed8f63ada51ba179ee559","unresolved":false,"context_lines":[{"line_number":154,"context_line":"To allow data collection to be leveraged by vendors that have implemented"},{"line_number":155,"context_line":"their own ML2 and L3 agents, data collection of items will be handled via"},{"line_number":156,"context_line":"RPC calls to respective ML2 and L3 agents.  The specific RPC calls that"},{"line_number":157,"context_line":"will be used are out of scope of this section."},{"line_number":158,"context_line":""},{"line_number":159,"context_line":"``To Do: flesh the following out``"},{"line_number":160,"context_line":""}],"source_content_type":"text/x-rst","patch_set":6,"id":"3a50d1a3_33a8178c","line":157,"updated":"2015-07-27 21:53:11.000000000","message":"Instead of RPC calls, I think this doc should discuss pushing the data into ceilometer.","commit_id":"f51b962661e10cdb601bd513413adaacf3b53e12"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"afddf30d757ad654a7511708af6bab09f4fd1394","unresolved":false,"context_lines":[{"line_number":154,"context_line":"To allow data collection to be leveraged by vendors that have implemented"},{"line_number":155,"context_line":"their own ML2 and L3 agents, data collection of items will be handled via"},{"line_number":156,"context_line":"RPC calls to respective ML2 and L3 agents.  The specific RPC calls that"},{"line_number":157,"context_line":"will be used are out of scope of this section."},{"line_number":158,"context_line":""},{"line_number":159,"context_line":"``To Do: flesh the following out``"},{"line_number":160,"context_line":""}],"source_content_type":"text/x-rst","patch_set":6,"id":"3a50d1a3_bd7e30ad","line":157,"in_reply_to":"3a50d1a3_33a8178c","updated":"2015-07-28 01:55:57.000000000","message":"The question of push/pull for getting data over to ceilometer is part of what I think of as phase 2 of this effort (and I\u0027m still hip deep in phase 1 at this point).  That being said, you are correct that this document should state that both options are to be considered.\n\nThere is also the question of whether the summations for DVR discussed above would take place in ceilometer or neutron, and I\u0027m nowhere close to being decided on which is better, so I\u0027m trying to push that off to phase 2 which I currently see as coming under a separate devref/spec...","commit_id":"f51b962661e10cdb601bd513413adaacf3b53e12"},{"author":{"_account_id":4656,"name":"Sean M. Collins","email":"sean@coreitpro.com","username":"scollins"},"change_message_id":"988e5e8bf900b873a0a1fcee10b4c8b50a0076b8","unresolved":false,"context_lines":[{"line_number":241,"context_line":""},{"line_number":242,"context_line":"This section considers the different models for getting data from the"},{"line_number":243,"context_line":"agents where it has been harvested back to the upstream collection point"},{"line_number":244,"context_line":"(Ceilometer)."},{"line_number":245,"context_line":""},{"line_number":246,"context_line":"``To Do: flesh the following out``"},{"line_number":247,"context_line":""}],"source_content_type":"text/x-rst","patch_set":7,"id":"3a50d1a3_dca19795","line":244,"updated":"2015-07-28 17:21:54.000000000","message":"Cool, I\u0027m happy with this.","commit_id":"79a57316fe822eab1a06e5a42bdc8f39f665feff"},{"author":{"_account_id":7776,"name":"Paul Carver","email":"pcarver@att.com","username":"pcarver"},"change_message_id":"da5754ecd922adee4f4f468a8d535e988975a2e7","unresolved":false,"context_lines":[{"line_number":22,"context_line":""},{"line_number":23,"context_line":"Instrumentation is **not** about having Neutron act as a bitbucket"},{"line_number":24,"context_line":"for storing historical data, but rather cache the information for"},{"line_number":25,"context_line":"retrieveal by Ceilometer.  The details of how this retrieval will occur"},{"line_number":26,"context_line":"is out of the scope of this devref."},{"line_number":27,"context_line":""},{"line_number":28,"context_line":"What is Available"}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_6a6850f6","line":25,"updated":"2015-08-04 01:16:06.000000000","message":"retrieveal -\u003e retrieval","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"c8642c342ea5332bffe4e672d7ef89fc43905824","unresolved":false,"context_lines":[{"line_number":22,"context_line":""},{"line_number":23,"context_line":"Instrumentation is **not** about having Neutron act as a bitbucket"},{"line_number":24,"context_line":"for storing historical data, but rather cache the information for"},{"line_number":25,"context_line":"retrieveal by Ceilometer.  The details of how this retrieval will occur"},{"line_number":26,"context_line":"is out of the scope of this devref."},{"line_number":27,"context_line":""},{"line_number":28,"context_line":"What is Available"}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_915fd540","line":25,"in_reply_to":"1a4dcd0f_6a6850f6","updated":"2015-08-04 21:35:06.000000000","message":"Done","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":7776,"name":"Paul Carver","email":"pcarver@att.com","username":"pcarver"},"change_message_id":"da5754ecd922adee4f4f468a8d535e988975a2e7","unresolved":false,"context_lines":[{"line_number":50,"context_line":"into these directly [#]_, so collecting these statistics will"},{"line_number":51,"context_line":"be considered out of scope of this devref."},{"line_number":52,"context_line":""},{"line_number":53,"context_line":".. [#] https://wiki.openstack.org/wiki/EfficientMetering/FurtureNovaInteractionModel"},{"line_number":54,"context_line":""},{"line_number":55,"context_line":"Metering Labels and Rules"},{"line_number":56,"context_line":"~~~~~~~~~~~~~~~~~~~~~~~~~"}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_4a31d40a","line":53,"updated":"2015-08-04 01:16:06.000000000","message":"Is this a link to a WIP wiki page? There\u0027s currently no information at this URL.","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"c8642c342ea5332bffe4e672d7ef89fc43905824","unresolved":false,"context_lines":[{"line_number":50,"context_line":"into these directly [#]_, so collecting these statistics will"},{"line_number":51,"context_line":"be considered out of scope of this devref."},{"line_number":52,"context_line":""},{"line_number":53,"context_line":".. [#] https://wiki.openstack.org/wiki/EfficientMetering/FurtureNovaInteractionModel"},{"line_number":54,"context_line":""},{"line_number":55,"context_line":"Metering Labels and Rules"},{"line_number":56,"context_line":"~~~~~~~~~~~~~~~~~~~~~~~~~"}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_a43651e5","line":53,"in_reply_to":"1a4dcd0f_4a31d40a","updated":"2015-08-04 21:35:06.000000000","message":"Sigh, I had a typo in the URL - will be fixed in the next draft","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":7776,"name":"Paul Carver","email":"pcarver@att.com","username":"pcarver"},"change_message_id":"da5754ecd922adee4f4f468a8d535e988975a2e7","unresolved":false,"context_lines":[{"line_number":74,"context_line":"be mapped to appropriate places, as discussed below."},{"line_number":75,"context_line":""},{"line_number":76,"context_line":"Note: the examples below talk in terms of mapping counters"},{"line_number":77,"context_line":"available from the linux operating system:"},{"line_number":78,"context_line":""},{"line_number":79,"context_line":"* receive bytes, packets, errors, dropped, overrun and multicast"},{"line_number":80,"context_line":"* transmit bytes, packets, errors, dropped, carrier, and collisions"}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_45d0e753","line":77,"updated":"2015-08-04 01:16:06.000000000","message":"linux -\u003e Linux","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"c8642c342ea5332bffe4e672d7ef89fc43905824","unresolved":false,"context_lines":[{"line_number":74,"context_line":"be mapped to appropriate places, as discussed below."},{"line_number":75,"context_line":""},{"line_number":76,"context_line":"Note: the examples below talk in terms of mapping counters"},{"line_number":77,"context_line":"available from the linux operating system:"},{"line_number":78,"context_line":""},{"line_number":79,"context_line":"* receive bytes, packets, errors, dropped, overrun and multicast"},{"line_number":80,"context_line":"* transmit bytes, packets, errors, dropped, carrier, and collisions"}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_e45a192a","line":77,"in_reply_to":"1a4dcd0f_45d0e753","updated":"2015-08-04 21:35:06.000000000","message":"Changed everywhere but the URL","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":7776,"name":"Paul Carver","email":"pcarver@att.com","username":"pcarver"},"change_message_id":"da5754ecd922adee4f4f468a8d535e988975a2e7","unresolved":false,"context_lines":[{"line_number":119,"context_line":"~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~"},{"line_number":120,"context_line":""},{"line_number":121,"context_line":"For each compute host, each network will be represented with a"},{"line_number":122,"context_line":"\"switch\", modelled by instances of ifTable and ifXTable. This"},{"line_number":123,"context_line":"mapping has the advantage that for a particular network, the"},{"line_number":124,"context_line":"view to the tenant or the operator is identical - the only"},{"line_number":125,"context_line":"difference is that the operator can see all networks, while a"}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_8a06fc1e","line":122,"updated":"2015-08-04 01:16:06.000000000","message":"modelled -\u003e modeled","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"c8642c342ea5332bffe4e672d7ef89fc43905824","unresolved":false,"context_lines":[{"line_number":119,"context_line":"~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~"},{"line_number":120,"context_line":""},{"line_number":121,"context_line":"For each compute host, each network will be represented with a"},{"line_number":122,"context_line":"\"switch\", modelled by instances of ifTable and ifXTable. This"},{"line_number":123,"context_line":"mapping has the advantage that for a particular network, the"},{"line_number":124,"context_line":"view to the tenant or the operator is identical - the only"},{"line_number":125,"context_line":"difference is that the operator can see all networks, while a"}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_8442353d","line":122,"in_reply_to":"1a4dcd0f_8a06fc1e","updated":"2015-08-04 21:35:06.000000000","message":"Done","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":7776,"name":"Paul Carver","email":"pcarver@att.com","username":"pcarver"},"change_message_id":"da5754ecd922adee4f4f468a8d535e988975a2e7","unresolved":false,"context_lines":[{"line_number":125,"context_line":"difference is that the operator can see all networks, while a"},{"line_number":126,"context_line":"tenant will only see the networks under their tenant id."},{"line_number":127,"context_line":""},{"line_number":128,"context_line":"As tap interfaces names identify the Neutron port they are"},{"line_number":129,"context_line":"associated with and the Neutron port identifies the Neutron"},{"line_number":130,"context_line":"network.  It is possible to take counters from each tap"},{"line_number":131,"context_line":"interface and map them into an entry in the appropriate tables,"}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_45ff8701","line":128,"updated":"2015-08-04 01:16:06.000000000","message":"interfaces -\u003e interface","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"c8642c342ea5332bffe4e672d7ef89fc43905824","unresolved":false,"context_lines":[{"line_number":125,"context_line":"difference is that the operator can see all networks, while a"},{"line_number":126,"context_line":"tenant will only see the networks under their tenant id."},{"line_number":127,"context_line":""},{"line_number":128,"context_line":"As tap interfaces names identify the Neutron port they are"},{"line_number":129,"context_line":"associated with and the Neutron port identifies the Neutron"},{"line_number":130,"context_line":"network.  It is possible to take counters from each tap"},{"line_number":131,"context_line":"interface and map them into an entry in the appropriate tables,"}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_444c2d70","line":128,"in_reply_to":"1a4dcd0f_45ff8701","updated":"2015-08-04 21:35:06.000000000","message":"Done","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":7776,"name":"Paul Carver","email":"pcarver@att.com","username":"pcarver"},"change_message_id":"da5754ecd922adee4f4f468a8d535e988975a2e7","unresolved":false,"context_lines":[{"line_number":127,"context_line":""},{"line_number":128,"context_line":"As tap interfaces names identify the Neutron port they are"},{"line_number":129,"context_line":"associated with and the Neutron port identifies the Neutron"},{"line_number":130,"context_line":"network.  It is possible to take counters from each tap"},{"line_number":131,"context_line":"interface and map them into an entry in the appropriate tables,"},{"line_number":132,"context_line":"using the following proposed assignments:"},{"line_number":133,"context_line":""}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_851d8f68","line":130,"updated":"2015-08-04 01:16:06.000000000","message":"The first sentence is a fragment:\n\n\"network.  It\" -\u003e \"network, it\"","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"c8642c342ea5332bffe4e672d7ef89fc43905824","unresolved":false,"context_lines":[{"line_number":127,"context_line":""},{"line_number":128,"context_line":"As tap interfaces names identify the Neutron port they are"},{"line_number":129,"context_line":"associated with and the Neutron port identifies the Neutron"},{"line_number":130,"context_line":"network.  It is possible to take counters from each tap"},{"line_number":131,"context_line":"interface and map them into an entry in the appropriate tables,"},{"line_number":132,"context_line":"using the following proposed assignments:"},{"line_number":133,"context_line":""}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_a4abb1d5","line":130,"in_reply_to":"1a4dcd0f_851d8f68","updated":"2015-08-04 21:35:06.000000000","message":"I made a slightly different change to remove the fragment","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":7776,"name":"Paul Carver","email":"pcarver@att.com","username":"pcarver"},"change_message_id":"da5754ecd922adee4f4f468a8d535e988975a2e7","unresolved":false,"context_lines":[{"line_number":228,"context_line":"Counters tracked by RFC 4293 come in two flavors: ones that are"},{"line_number":229,"context_line":"inherited from the interface, and those that track L3 events,"},{"line_number":230,"context_line":"such as fragmentation, re-assembly, truncations, etc.  Given"},{"line_number":231,"context_line":"the existing instrumentation of the systems that underlie neutron,"},{"line_number":232,"context_line":"the following counters are declared out of scope for this devref:"},{"line_number":233,"context_line":""},{"line_number":234,"context_line":"* ipSystemStatsInHdrErrors, ipIfStatsInHdrErrors"}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_25cfdbb1","line":231,"updated":"2015-08-04 01:16:06.000000000","message":"neutron -\u003e Neutron","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"c8642c342ea5332bffe4e672d7ef89fc43905824","unresolved":false,"context_lines":[{"line_number":228,"context_line":"Counters tracked by RFC 4293 come in two flavors: ones that are"},{"line_number":229,"context_line":"inherited from the interface, and those that track L3 events,"},{"line_number":230,"context_line":"such as fragmentation, re-assembly, truncations, etc.  Given"},{"line_number":231,"context_line":"the existing instrumentation of the systems that underlie neutron,"},{"line_number":232,"context_line":"the following counters are declared out of scope for this devref:"},{"line_number":233,"context_line":""},{"line_number":234,"context_line":"* ipSystemStatsInHdrErrors, ipIfStatsInHdrErrors"}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_a492d19d","line":231,"in_reply_to":"1a4dcd0f_25cfdbb1","updated":"2015-08-04 21:35:06.000000000","message":"Changed everywhere","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":7776,"name":"Paul Carver","email":"pcarver@att.com","username":"pcarver"},"change_message_id":"da5754ecd922adee4f4f468a8d535e988975a2e7","unresolved":false,"context_lines":[{"line_number":295,"context_line":""},{"line_number":296,"context_line":"This section considers the different models for getting data from the"},{"line_number":297,"context_line":"agents where it has been harvested back to the upstream collection point"},{"line_number":298,"context_line":"(Ceilometer).  There basic models to choose from are:"},{"line_number":299,"context_line":""},{"line_number":300,"context_line":"* Communicate directly between the data harvesters and Ceilometer, and"},{"line_number":301,"context_line":"* Use a neutron agent/thread to aggregate data from the harvesters and"}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_05a5ff02","line":298,"updated":"2015-08-04 01:16:06.000000000","message":"There -\u003e The","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"c8642c342ea5332bffe4e672d7ef89fc43905824","unresolved":false,"context_lines":[{"line_number":295,"context_line":""},{"line_number":296,"context_line":"This section considers the different models for getting data from the"},{"line_number":297,"context_line":"agents where it has been harvested back to the upstream collection point"},{"line_number":298,"context_line":"(Ceilometer).  There basic models to choose from are:"},{"line_number":299,"context_line":""},{"line_number":300,"context_line":"* Communicate directly between the data harvesters and Ceilometer, and"},{"line_number":301,"context_line":"* Use a neutron agent/thread to aggregate data from the harvesters and"}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_04eaa50f","line":298,"in_reply_to":"1a4dcd0f_05a5ff02","updated":"2015-08-04 21:35:06.000000000","message":"Done","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":7776,"name":"Paul Carver","email":"pcarver@att.com","username":"pcarver"},"change_message_id":"da5754ecd922adee4f4f468a8d535e988975a2e7","unresolved":false,"context_lines":[{"line_number":301,"context_line":"* Use a neutron agent/thread to aggregate data from the harvesters and"},{"line_number":302,"context_line":"  communicate the aggregation to Ceilometer."},{"line_number":303,"context_line":""},{"line_number":304,"context_line":"One of the driving questions in choose between the two models is where"},{"line_number":305,"context_line":"should harvested data be aggregated.  The first option requires all"},{"line_number":306,"context_line":"aggregation to occur in Ceilometer, while the second (obviously) aggregates"},{"line_number":307,"context_line":"the data within neutron.  Because the aggregation is driven by the"}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_85b90f31","line":304,"updated":"2015-08-04 01:16:06.000000000","message":"choose -\u003e choosing","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"c8642c342ea5332bffe4e672d7ef89fc43905824","unresolved":false,"context_lines":[{"line_number":301,"context_line":"* Use a neutron agent/thread to aggregate data from the harvesters and"},{"line_number":302,"context_line":"  communicate the aggregation to Ceilometer."},{"line_number":303,"context_line":""},{"line_number":304,"context_line":"One of the driving questions in choose between the two models is where"},{"line_number":305,"context_line":"should harvested data be aggregated.  The first option requires all"},{"line_number":306,"context_line":"aggregation to occur in Ceilometer, while the second (obviously) aggregates"},{"line_number":307,"context_line":"the data within neutron.  Because the aggregation is driven by the"}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_84deb56b","line":304,"in_reply_to":"1a4dcd0f_85b90f31","updated":"2015-08-04 21:35:06.000000000","message":"Done","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":7776,"name":"Paul Carver","email":"pcarver@att.com","username":"pcarver"},"change_message_id":"da5754ecd922adee4f4f468a8d535e988975a2e7","unresolved":false,"context_lines":[{"line_number":305,"context_line":"should harvested data be aggregated.  The first option requires all"},{"line_number":306,"context_line":"aggregation to occur in Ceilometer, while the second (obviously) aggregates"},{"line_number":307,"context_line":"the data within neutron.  Because the aggregation is driven by the"},{"line_number":308,"context_line":"provisioned artifacts within neutron, it may make more sense to use the"},{"line_number":309,"context_line":"second approach."},{"line_number":310,"context_line":""},{"line_number":311,"context_line":"In addition to where aggregation takes place, there is the choice between"}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_c54b972a","line":308,"updated":"2015-08-04 01:16:06.000000000","message":"neutron -\u003e Neutron","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"c8642c342ea5332bffe4e672d7ef89fc43905824","unresolved":false,"context_lines":[{"line_number":305,"context_line":"should harvested data be aggregated.  The first option requires all"},{"line_number":306,"context_line":"aggregation to occur in Ceilometer, while the second (obviously) aggregates"},{"line_number":307,"context_line":"the data within neutron.  Because the aggregation is driven by the"},{"line_number":308,"context_line":"provisioned artifacts within neutron, it may make more sense to use the"},{"line_number":309,"context_line":"second approach."},{"line_number":310,"context_line":""},{"line_number":311,"context_line":"In addition to where aggregation takes place, there is the choice between"}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_e4f699f8","line":308,"in_reply_to":"1a4dcd0f_c54b972a","updated":"2015-08-04 21:35:06.000000000","message":"Done","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":7776,"name":"Paul Carver","email":"pcarver@att.com","username":"pcarver"},"change_message_id":"da5754ecd922adee4f4f468a8d535e988975a2e7","unresolved":false,"context_lines":[{"line_number":318,"context_line":"second approach makes more sense."},{"line_number":319,"context_line":""},{"line_number":320,"context_line":"A corollary to the above paragraph is whether or not the counters"},{"line_number":321,"context_line":"collected by the neutron server need to be persisted.  Three choices"},{"line_number":322,"context_line":"are persist in a database (overkill), persist via a disk file (slow)"},{"line_number":323,"context_line":"or only hold in memory (potential scaling issues).  For this iteration,"},{"line_number":324,"context_line":"the \"persist via a disk file\" approach will be taken."}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_650ac3d4","line":321,"updated":"2015-08-04 01:16:06.000000000","message":"neutron -\u003e Neutron","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":4656,"name":"Sean M. Collins","email":"sean@coreitpro.com","username":"scollins"},"change_message_id":"3630283ad8fdb381cf6269a5ee330fc696a81298","unresolved":false,"context_lines":[{"line_number":321,"context_line":"collected by the neutron server need to be persisted.  Three choices"},{"line_number":322,"context_line":"are persist in a database (overkill), persist via a disk file (slow)"},{"line_number":323,"context_line":"or only hold in memory (potential scaling issues).  For this iteration,"},{"line_number":324,"context_line":"the \"persist via a disk file\" approach will be taken."},{"line_number":325,"context_line":""},{"line_number":326,"context_line":"The result is each agent would have a harvester thread that periodically"},{"line_number":327,"context_line":"notify the neutron server of updated counter information, which would be"}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_e7a4a009","line":324,"updated":"2015-08-03 21:05:40.000000000","message":"I think this is not a valid conclusion, and would strongly research how data is currently collected via ceilometer\u0027s compute agent. We should not be duplicating functionality in the Neutron project for items that are out of scope, and for which we have a dedicated project already for.","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"c8642c342ea5332bffe4e672d7ef89fc43905824","unresolved":false,"context_lines":[{"line_number":321,"context_line":"collected by the neutron server need to be persisted.  Three choices"},{"line_number":322,"context_line":"are persist in a database (overkill), persist via a disk file (slow)"},{"line_number":323,"context_line":"or only hold in memory (potential scaling issues).  For this iteration,"},{"line_number":324,"context_line":"the \"persist via a disk file\" approach will be taken."},{"line_number":325,"context_line":""},{"line_number":326,"context_line":"The result is each agent would have a harvester thread that periodically"},{"line_number":327,"context_line":"notify the neutron server of updated counter information, which would be"}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_c4fefd6a","line":324,"in_reply_to":"1a4dcd0f_e7a4a009","updated":"2015-08-04 21:35:06.000000000","message":"the head of the Ceilometer code looks like it polls elements for all network information, so I\u0027m not sure I agree that this is out of scope.  Given that the aggregation rules are known to Neutron (tenant network definitions) I\u0027m still dubious about asking Ceilometer to pull those each time it aggregates","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":7776,"name":"Paul Carver","email":"pcarver@att.com","username":"pcarver"},"change_message_id":"da5754ecd922adee4f4f468a8d535e988975a2e7","unresolved":false,"context_lines":[{"line_number":324,"context_line":"the \"persist via a disk file\" approach will be taken."},{"line_number":325,"context_line":""},{"line_number":326,"context_line":"The result is each agent would have a harvester thread that periodically"},{"line_number":327,"context_line":"notify the neutron server of updated counter information, which would be"},{"line_number":328,"context_line":"persisted on the neutron server. The neutron server would also have it own"},{"line_number":329,"context_line":"reporting thread that periodically aggregates the current persisted"},{"line_number":330,"context_line":"counters and notifies Ceilometer of the results."}],"source_content_type":"text/x-rst","patch_set":12,"id":"1a4dcd0f_85034fb8","line":327,"updated":"2015-08-04 01:16:06.000000000","message":"neutron -\u003e Neutron","commit_id":"32cb98855989db18cc32602dcb0055a9cb863a22"},{"author":{"_account_id":2874,"name":"yong sheng gong","email":"gong.yongsheng@99cloud.net","username":"gongysh"},"change_message_id":"d3b1686217c341276fd3ed2239661d5ed7e59537","unresolved":false,"context_lines":[{"line_number":21,"context_line":"-----"},{"line_number":22,"context_line":""},{"line_number":23,"context_line":"Instrumentation is **not** about having Neutron act as a bitbucket"},{"line_number":24,"context_line":"for storing historical data, but rather cache the information for"},{"line_number":25,"context_line":"retrieval by Ceilometer.  The details of how this retrieval will occur"},{"line_number":26,"context_line":"is out of the scope of this devref."},{"line_number":27,"context_line":""}],"source_content_type":"text/x-rst","patch_set":17,"id":"da20952f_46a11893","line":24,"updated":"2015-09-06 03:52:15.000000000","message":"cache -\u003e caching?","commit_id":"836bac9a0a3ef8a5040a31a56795daddd18dc448"},{"author":{"_account_id":2874,"name":"yong sheng gong","email":"gong.yongsheng@99cloud.net","username":"gongysh"},"change_message_id":"d3b1686217c341276fd3ed2239661d5ed7e59537","unresolved":false,"context_lines":[{"line_number":105,"context_line":"necessary to consider how information from the OVS gets mapped into the"},{"line_number":106,"context_line":"defined data structures"},{"line_number":107,"context_line":""},{"line_number":108,"context_line":"1. VIF and port statistics to switch interfaces information"},{"line_number":109,"context_line":"2. Host Switch information to switch interface information for"},{"line_number":110,"context_line":"   classic routers"},{"line_number":111,"context_line":"3. Host OVS interface statistics to router interface information for"}],"source_content_type":"text/x-rst","patch_set":17,"id":"da20952f_c6e208d5","line":108,"updated":"2015-09-06 03:52:15.000000000","message":"can we use rst struct to generate the list?","commit_id":"836bac9a0a3ef8a5040a31a56795daddd18dc448"},{"author":{"_account_id":16237,"name":"venkatamahesh","email":"venkatamaheshkotha@gmail.com","username":"venkatamahesh"},"change_message_id":"20173d9826425d8742fde643656a8b42bc457540","unresolved":false,"context_lines":[{"line_number":14,"context_line":""},{"line_number":15,"context_line":"* Identify what data is already available from the components used by"},{"line_number":16,"context_line":"  Neutron and how it can be mapped into the structures defined by the RFC."},{"line_number":17,"context_line":"* Idenitfy what code needs to be added for data that isn\u0027t already"},{"line_number":18,"context_line":"  available."},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"Scope"}],"source_content_type":"text/x-rst","patch_set":19,"id":"da20952f_0431a9a1","line":17,"updated":"2015-09-08 04:32:53.000000000","message":"+1","commit_id":"054506810cc066f984333703ba753f060080b96e"},{"author":{"_account_id":7715,"name":"Hirofumi Ichihara","email":"ichihara.hirofumi@gmail.com","username":"Hirofumi-Ichihara"},"change_message_id":"3a16e1417ceea928a0088f82a3cee3633514ad14","unresolved":false,"context_lines":[{"line_number":14,"context_line":""},{"line_number":15,"context_line":"* Identify what data is already available from the components used by"},{"line_number":16,"context_line":"  Neutron and how it can be mapped into the structures defined by the RFC."},{"line_number":17,"context_line":"* Idenitfy what code needs to be added for data that isn\u0027t already"},{"line_number":18,"context_line":"  available."},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"Scope"}],"source_content_type":"text/x-rst","patch_set":19,"id":"da20952f_2ac6f967","line":17,"updated":"2015-09-08 00:36:14.000000000","message":"Idenitfy -\u003e Identify","commit_id":"054506810cc066f984333703ba753f060080b96e"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"bdd265429f6b62b4d9fc452dabaa940b3360c712","unresolved":false,"context_lines":[{"line_number":14,"context_line":""},{"line_number":15,"context_line":"* Identify what data is already available from the components used by"},{"line_number":16,"context_line":"  Neutron and how it can be mapped into the structures defined by the RFC."},{"line_number":17,"context_line":"* Idenitfy what code needs to be added for data that isn\u0027t already"},{"line_number":18,"context_line":"  available."},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"Scope"}],"source_content_type":"text/x-rst","patch_set":19,"id":"da20952f_2efa7b44","line":17,"in_reply_to":"da20952f_2ac6f967","updated":"2015-09-08 11:58:24.000000000","message":"Done","commit_id":"054506810cc066f984333703ba753f060080b96e"},{"author":{"_account_id":16237,"name":"venkatamahesh","email":"venkatamaheshkotha@gmail.com","username":"venkatamahesh"},"change_message_id":"20173d9826425d8742fde643656a8b42bc457540","unresolved":false,"context_lines":[{"line_number":22,"context_line":""},{"line_number":23,"context_line":"Instrumentation is **not** about having Neutron act as a bitbucket"},{"line_number":24,"context_line":"for storing historical data, but rather cache the information for"},{"line_number":25,"context_line":"retrieval by Ceilometer.  The details of how this retrieval will occur"},{"line_number":26,"context_line":"is out of the scope of this devref."},{"line_number":27,"context_line":""},{"line_number":28,"context_line":"What is Available"}],"source_content_type":"text/x-rst","patch_set":19,"id":"da20952f_841db910","line":25,"updated":"2015-09-08 04:32:53.000000000","message":"Before \"The\" two bit spaces are there. we can remove a bit space.","commit_id":"054506810cc066f984333703ba753f060080b96e"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"bdd265429f6b62b4d9fc452dabaa940b3360c712","unresolved":false,"context_lines":[{"line_number":22,"context_line":""},{"line_number":23,"context_line":"Instrumentation is **not** about having Neutron act as a bitbucket"},{"line_number":24,"context_line":"for storing historical data, but rather cache the information for"},{"line_number":25,"context_line":"retrieval by Ceilometer.  The details of how this retrieval will occur"},{"line_number":26,"context_line":"is out of the scope of this devref."},{"line_number":27,"context_line":""},{"line_number":28,"context_line":"What is Available"}],"source_content_type":"text/x-rst","patch_set":19,"id":"da20952f_aeb3eb99","line":25,"in_reply_to":"da20952f_841db910","updated":"2015-09-08 11:58:24.000000000","message":"And once again, I show my age, because that\u0027s *still* what I consider to be proper.  Having briefly groused, so done.","commit_id":"054506810cc066f984333703ba753f060080b96e"},{"author":{"_account_id":16237,"name":"venkatamahesh","email":"venkatamaheshkotha@gmail.com","username":"venkatamahesh"},"change_message_id":"20173d9826425d8742fde643656a8b42bc457540","unresolved":false,"context_lines":[{"line_number":77,"context_line":"available from the Linux operating system:"},{"line_number":78,"context_line":""},{"line_number":79,"context_line":"* receive bytes, packets, errors, dropped, overrun and multicast"},{"line_number":80,"context_line":"* transmit bytes, packets, errors, dropped, carrier, and collisions"},{"line_number":81,"context_line":""},{"line_number":82,"context_line":"Available counters for interfaces on other operating systems"},{"line_number":83,"context_line":"can be mapped in a similar fashion."}],"source_content_type":"text/x-rst","patch_set":19,"id":"da20952f_64974548","line":80,"updated":"2015-09-08 04:32:53.000000000","message":"After carrier can remove \",\"","commit_id":"054506810cc066f984333703ba753f060080b96e"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"bdd265429f6b62b4d9fc452dabaa940b3360c712","unresolved":false,"context_lines":[{"line_number":77,"context_line":"available from the Linux operating system:"},{"line_number":78,"context_line":""},{"line_number":79,"context_line":"* receive bytes, packets, errors, dropped, overrun and multicast"},{"line_number":80,"context_line":"* transmit bytes, packets, errors, dropped, carrier, and collisions"},{"line_number":81,"context_line":""},{"line_number":82,"context_line":"Available counters for interfaces on other operating systems"},{"line_number":83,"context_line":"can be mapped in a similar fashion."}],"source_content_type":"text/x-rst","patch_set":19,"id":"da20952f_2e7fdbac","line":80,"in_reply_to":"da20952f_64974548","updated":"2015-09-08 11:58:24.000000000","message":"Done","commit_id":"054506810cc066f984333703ba753f060080b96e"},{"author":{"_account_id":16237,"name":"venkatamahesh","email":"venkatamaheshkotha@gmail.com","username":"venkatamahesh"},"change_message_id":"20173d9826425d8742fde643656a8b42bc457540","unresolved":false,"context_lines":[{"line_number":112,"context_line":"   DVR"},{"line_number":113,"context_line":"4. Network node OVS statistics to router interface information for"},{"line_number":114,"context_line":"   classic routers"},{"line_number":115,"context_line":"5. Network node OVS statistics to router interface infromation for"},{"line_number":116,"context_line":"   sNAT addresses for DVR."},{"line_number":117,"context_line":""},{"line_number":118,"context_line":"RFC 2863 Structures from OVS - Switches"}],"source_content_type":"text/x-rst","patch_set":19,"id":"da20952f_e4891527","line":115,"updated":"2015-09-08 04:32:53.000000000","message":"+1","commit_id":"054506810cc066f984333703ba753f060080b96e"},{"author":{"_account_id":7715,"name":"Hirofumi Ichihara","email":"ichihara.hirofumi@gmail.com","username":"Hirofumi-Ichihara"},"change_message_id":"3a16e1417ceea928a0088f82a3cee3633514ad14","unresolved":false,"context_lines":[{"line_number":112,"context_line":"   DVR"},{"line_number":113,"context_line":"4. Network node OVS statistics to router interface information for"},{"line_number":114,"context_line":"   classic routers"},{"line_number":115,"context_line":"5. Network node OVS statistics to router interface infromation for"},{"line_number":116,"context_line":"   sNAT addresses for DVR."},{"line_number":117,"context_line":""},{"line_number":118,"context_line":"RFC 2863 Structures from OVS - Switches"}],"source_content_type":"text/x-rst","patch_set":19,"id":"da20952f_eaa241ad","line":115,"updated":"2015-09-08 00:36:14.000000000","message":"infromation -\u003e information","commit_id":"054506810cc066f984333703ba753f060080b96e"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"bdd265429f6b62b4d9fc452dabaa940b3360c712","unresolved":false,"context_lines":[{"line_number":112,"context_line":"   DVR"},{"line_number":113,"context_line":"4. Network node OVS statistics to router interface information for"},{"line_number":114,"context_line":"   classic routers"},{"line_number":115,"context_line":"5. Network node OVS statistics to router interface infromation for"},{"line_number":116,"context_line":"   sNAT addresses for DVR."},{"line_number":117,"context_line":""},{"line_number":118,"context_line":"RFC 2863 Structures from OVS - Switches"}],"source_content_type":"text/x-rst","patch_set":19,"id":"da20952f_8e59475e","line":115,"in_reply_to":"da20952f_eaa241ad","updated":"2015-09-08 11:58:24.000000000","message":"Done","commit_id":"054506810cc066f984333703ba753f060080b96e"},{"author":{"_account_id":16237,"name":"venkatamahesh","email":"venkatamaheshkotha@gmail.com","username":"venkatamahesh"},"change_message_id":"20173d9826425d8742fde643656a8b42bc457540","unresolved":false,"context_lines":[{"line_number":174,"context_line":"with the router will be mapped to rows in the tables, using the"},{"line_number":175,"context_line":"same counter assignments as above."},{"line_number":176,"context_line":""},{"line_number":177,"context_line":"For DVR routers, counters fom each qr and qg interfaces associated"},{"line_number":178,"context_line":"with the router (whether they are on the network node or compute"},{"line_number":179,"context_line":"nodes) need to be summed together (on a per interface basis) before"},{"line_number":180,"context_line":"they are inserted into the table (each interface will be a separate"}],"source_content_type":"text/x-rst","patch_set":19,"id":"da20952f_a4ad5d85","line":177,"updated":"2015-09-08 04:32:53.000000000","message":"+1","commit_id":"054506810cc066f984333703ba753f060080b96e"},{"author":{"_account_id":7715,"name":"Hirofumi Ichihara","email":"ichihara.hirofumi@gmail.com","username":"Hirofumi-Ichihara"},"change_message_id":"3a16e1417ceea928a0088f82a3cee3633514ad14","unresolved":false,"context_lines":[{"line_number":174,"context_line":"with the router will be mapped to rows in the tables, using the"},{"line_number":175,"context_line":"same counter assignments as above."},{"line_number":176,"context_line":""},{"line_number":177,"context_line":"For DVR routers, counters fom each qr and qg interfaces associated"},{"line_number":178,"context_line":"with the router (whether they are on the network node or compute"},{"line_number":179,"context_line":"nodes) need to be summed together (on a per interface basis) before"},{"line_number":180,"context_line":"they are inserted into the table (each interface will be a separate"}],"source_content_type":"text/x-rst","patch_set":19,"id":"da20952f_0aa695c0","line":177,"updated":"2015-09-08 00:36:14.000000000","message":"fom -\u003e from","commit_id":"054506810cc066f984333703ba753f060080b96e"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"bdd265429f6b62b4d9fc452dabaa940b3360c712","unresolved":false,"context_lines":[{"line_number":174,"context_line":"with the router will be mapped to rows in the tables, using the"},{"line_number":175,"context_line":"same counter assignments as above."},{"line_number":176,"context_line":""},{"line_number":177,"context_line":"For DVR routers, counters fom each qr and qg interfaces associated"},{"line_number":178,"context_line":"with the router (whether they are on the network node or compute"},{"line_number":179,"context_line":"nodes) need to be summed together (on a per interface basis) before"},{"line_number":180,"context_line":"they are inserted into the table (each interface will be a separate"}],"source_content_type":"text/x-rst","patch_set":19,"id":"da20952f_2e56fb2a","line":177,"in_reply_to":"da20952f_0aa695c0","updated":"2015-09-08 11:58:24.000000000","message":"Done","commit_id":"054506810cc066f984333703ba753f060080b96e"},{"author":{"_account_id":7805,"name":"Li Ma","email":"skywalker.nick@gmail.com","username":"Nick"},"change_message_id":"882cdbaceb345e768652fb7d0f9016338ff98f28","unresolved":false,"context_lines":[{"line_number":174,"context_line":"with the router will be mapped to rows in the tables, using the"},{"line_number":175,"context_line":"same counter assignments as above."},{"line_number":176,"context_line":""},{"line_number":177,"context_line":"For DVR routers, counters fom each qr and qg interfaces associated"},{"line_number":178,"context_line":"with the router (whether they are on the network node or compute"},{"line_number":179,"context_line":"nodes) need to be summed together (on a per interface basis) before"},{"line_number":180,"context_line":"they are inserted into the table (each interface will be a separate"}],"source_content_type":"text/x-rst","patch_set":22,"id":"da20952f_a7fc1b40","line":177,"updated":"2015-09-10 06:24:12.000000000","message":"fom --\u003e for?","commit_id":"7e83183e7f845b1e8f7214c8406031f64bb99f2a"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"ce4568c09d4d8e896b01a61ab94355abec51c094","unresolved":false,"context_lines":[{"line_number":174,"context_line":"with the router will be mapped to rows in the tables, using the"},{"line_number":175,"context_line":"same counter assignments as above."},{"line_number":176,"context_line":""},{"line_number":177,"context_line":"For DVR routers, counters fom each qr and qg interfaces associated"},{"line_number":178,"context_line":"with the router (whether they are on the network node or compute"},{"line_number":179,"context_line":"nodes) need to be summed together (on a per interface basis) before"},{"line_number":180,"context_line":"they are inserted into the table (each interface will be a separate"}],"source_content_type":"text/x-rst","patch_set":22,"id":"da20952f_83b2ef94","line":177,"in_reply_to":"da20952f_a7fc1b40","updated":"2015-09-11 12:55:12.000000000","message":"sigh - done","commit_id":"7e83183e7f845b1e8f7214c8406031f64bb99f2a"},{"author":{"_account_id":7805,"name":"Li Ma","email":"skywalker.nick@gmail.com","username":"Nick"},"change_message_id":"882cdbaceb345e768652fb7d0f9016338ff98f28","unresolved":false,"context_lines":[{"line_number":321,"context_line":"collected by the Neutron server need to be persisted.  Three choices"},{"line_number":322,"context_line":"are persist in a database (overkill), persist via a disk file (slow)"},{"line_number":323,"context_line":"or only hold in memory (potential scaling issues).  For this iteration,"},{"line_number":324,"context_line":"the \"persist via a disk file\" approach will be taken."},{"line_number":325,"context_line":""},{"line_number":326,"context_line":"The result is each agent would have a harvester thread that periodically"},{"line_number":327,"context_line":"notify the Neutron server of updated counter information, which would be"}],"source_content_type":"text/x-rst","patch_set":22,"id":"da20952f_a7717be0","line":324,"updated":"2015-09-10 06:24:12.000000000","message":"What\u0027s the format of this disk file?\nbinary-obj, json, rrd, sqlite?","commit_id":"7e83183e7f845b1e8f7214c8406031f64bb99f2a"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"ce4568c09d4d8e896b01a61ab94355abec51c094","unresolved":false,"context_lines":[{"line_number":321,"context_line":"collected by the Neutron server need to be persisted.  Three choices"},{"line_number":322,"context_line":"are persist in a database (overkill), persist via a disk file (slow)"},{"line_number":323,"context_line":"or only hold in memory (potential scaling issues).  For this iteration,"},{"line_number":324,"context_line":"the \"persist via a disk file\" approach will be taken."},{"line_number":325,"context_line":""},{"line_number":326,"context_line":"The result is each agent would have a harvester thread that periodically"},{"line_number":327,"context_line":"notify the Neutron server of updated counter information, which would be"}],"source_content_type":"text/x-rst","patch_set":22,"id":"da20952f_c3630719","line":324,"in_reply_to":"da20952f_a7717be0","updated":"2015-09-11 12:55:12.000000000","message":"I\u0027m open to suggestions.  Remember that this is a cache of single data points, not a big data collection and while the statement of \"cacheness\" is made at the begining, I\u0027ll spin an updated devref that calls out that point in more detail in this section.","commit_id":"7e83183e7f845b1e8f7214c8406031f64bb99f2a"},{"author":{"_account_id":14957,"name":"Brad Behle","email":"behle@us.ibm.com","username":"behle"},"change_message_id":"91cc7c46b6721a19576f62e979352a86c82fe898","unresolved":false,"context_lines":[{"line_number":317,"context_line":"On the other hand, asynchronous notification requires that the aggregator"},{"line_number":318,"context_line":"have the capability to hold the current results from each harvester."},{"line_number":319,"context_line":"Given the current tradeoff between processing power and storage, the"},{"line_number":320,"context_line":"second approach makes more sense."},{"line_number":321,"context_line":""},{"line_number":322,"context_line":"A corollary to the above paragraph is whether or not the cached counters"},{"line_number":323,"context_line":"collected by the Neutron server need to be persisted. Our choices"}],"source_content_type":"text/x-rst","patch_set":24,"id":"da20952f_8b799d31","line":320,"updated":"2015-09-11 15:35:50.000000000","message":"In the above paragraph, you list requiring a polling thread as the disadvantage of polling vs. async notification.  But then two paragraphs later you state \"The Neutron server would also have it own reporting thread that periodically aggregates the cached counters and notifies Ceilometer of the results.\"\n\nSince either way you need a thread to do the work of aggregating and notifying Ceilometer, doesn\u0027t it make more sense to poll on a set interval and eliminate the complexity of getting async notifications from various agents and storing that data until it is aggregated and sent to Ceilomter?","commit_id":"f8dcbac1b1320a61b0a8f36707784e98f4cce655"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"083d4137a1c32e289c7109e30c3d4550c5bc038c","unresolved":false,"context_lines":[{"line_number":317,"context_line":"On the other hand, asynchronous notification requires that the aggregator"},{"line_number":318,"context_line":"have the capability to hold the current results from each harvester."},{"line_number":319,"context_line":"Given the current tradeoff between processing power and storage, the"},{"line_number":320,"context_line":"second approach makes more sense."},{"line_number":321,"context_line":""},{"line_number":322,"context_line":"A corollary to the above paragraph is whether or not the cached counters"},{"line_number":323,"context_line":"collected by the Neutron server need to be persisted. Our choices"}],"source_content_type":"text/x-rst","patch_set":24,"id":"da20952f_79f983c7","line":320,"in_reply_to":"da20952f_8b799d31","updated":"2015-09-11 16:26:07.000000000","message":"Hmm, let me spin a new patchset and correct that (as well as the red thingy), because that is *not* what I meant to say.","commit_id":"f8dcbac1b1320a61b0a8f36707784e98f4cce655"},{"author":{"_account_id":6854,"name":"YAMAMOTO Takashi","email":"yamamoto@midokura.com","username":"yamamoto"},"change_message_id":"ed301c106952eb6fbe3d052e2249060732ea4631","unresolved":false,"context_lines":[{"line_number":18,"context_line":"  available."},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"Scope"},{"line_number":21,"context_line":"-----"},{"line_number":22,"context_line":""},{"line_number":23,"context_line":"Instrumentation is **not** about having Neutron act as a bitbucket"},{"line_number":24,"context_line":"for storing historical data, but rather cache the information for"}],"source_content_type":"text/x-rst","patch_set":25,"id":"ba15a1d1_aa156d40","line":21,"updated":"2015-09-17 10:12:19.000000000","message":"this will only cover LB ans OVS, and have no effect to other plugins?\nor will this provides a way for other plugins to implement the counters?\nif the latter, how?","commit_id":"681351e0628e571ca3802a90b6d6db034c000179"},{"author":{"_account_id":261,"name":"Salvatore Orlando","email":"salv.orlando@gmail.com","username":"salvatore-orlando"},"change_message_id":"c5e3aedfbdf52c526bf5d3e0e6cc4f6f182a4ca4","unresolved":false,"context_lines":[{"line_number":18,"context_line":"  available."},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"Scope"},{"line_number":21,"context_line":"-----"},{"line_number":22,"context_line":""},{"line_number":23,"context_line":"Instrumentation is **not** about having Neutron act as a bitbucket"},{"line_number":24,"context_line":"for storing historical data, but rather cache the information for"}],"source_content_type":"text/x-rst","patch_set":25,"id":"9a1a9d01_c7f54f57","line":21,"in_reply_to":"ba15a1d1_9e1f2b09","updated":"2015-09-29 14:02:27.000000000","message":"technically it\u0027s not covering any specific plugin, but components of the neutron control plane.\n\nI think the document clarifies this very well - the work, including any public API that should result from it is specific to the reference control plane agents. Other plugins might or might not fit in the models introduced here, but that\u0027s out of scope.","commit_id":"681351e0628e571ca3802a90b6d6db034c000179"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"d93cf2f08153abb008a0b4899016b6ceefcdc19a","unresolved":false,"context_lines":[{"line_number":18,"context_line":"  available."},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"Scope"},{"line_number":21,"context_line":"-----"},{"line_number":22,"context_line":""},{"line_number":23,"context_line":"Instrumentation is **not** about having Neutron act as a bitbucket"},{"line_number":24,"context_line":"for storing historical data, but rather cache the information for"}],"source_content_type":"text/x-rst","patch_set":25,"id":"ba15a1d1_9e1f2b09","line":21,"in_reply_to":"ba15a1d1_aa156d40","updated":"2015-09-20 15:09:03.000000000","message":"Yes, this will cover LB and OVS - It is up to other plugins to implement the necessary counters.","commit_id":"681351e0628e571ca3802a90b6d6db034c000179"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"8999a39fef7cececba697596deab1b00f96adc3a","unresolved":false,"context_lines":[{"line_number":22,"context_line":""},{"line_number":23,"context_line":"Instrumentation is **not** about having Neutron act as a bitbucket"},{"line_number":24,"context_line":"for storing historical data, but rather cache the information for"},{"line_number":25,"context_line":"retrieval by Ceilometer. The details of how this retrieval will occur"},{"line_number":26,"context_line":"is out of the scope of this devref."},{"line_number":27,"context_line":""},{"line_number":28,"context_line":"What is Available"}],"source_content_type":"text/x-rst","patch_set":25,"id":"9a1a9d01_75189a1e","line":25,"updated":"2015-09-29 19:24:04.000000000","message":"I am wondering if it is better not to mandate here who is going to be the consumer of this data. Some may want to put an adapter in front of these faucets so that Ceilometer or Monasca can tap into, or any other recipient for that matter.","commit_id":"681351e0628e571ca3802a90b6d6db034c000179"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"8999a39fef7cececba697596deab1b00f96adc3a","unresolved":false,"context_lines":[{"line_number":25,"context_line":"retrieval by Ceilometer. The details of how this retrieval will occur"},{"line_number":26,"context_line":"is out of the scope of this devref."},{"line_number":27,"context_line":""},{"line_number":28,"context_line":"What is Available"},{"line_number":29,"context_line":"-----------------"},{"line_number":30,"context_line":""},{"line_number":31,"context_line":"The survey of what data is available and how it can be mapped will cover"}],"source_content_type":"text/x-rst","patch_set":25,"id":"9a1a9d01_15a27e4b","line":28,"updated":"2015-09-29 19:24:04.000000000","message":"there have been a number of attempts at improving visibility into Neutron systems in the past (I recall one proposing sflow at one point), some have crushed and burned. Did your research yield no results?","commit_id":"681351e0628e571ca3802a90b6d6db034c000179"},{"author":{"_account_id":261,"name":"Salvatore Orlando","email":"salv.orlando@gmail.com","username":"salvatore-orlando"},"change_message_id":"c5e3aedfbdf52c526bf5d3e0e6cc4f6f182a4ca4","unresolved":false,"context_lines":[{"line_number":68,"context_line":""},{"line_number":69,"context_line":"For the Linux bridge, a check of [#]_ shows that IEEE 802.1d"},{"line_number":70,"context_line":"mandated statistics are only a \"wishlist\" item. The alternative"},{"line_number":71,"context_line":"is to use NETLINK/shell to list the interfaces attached to"},{"line_number":72,"context_line":"a particular bridge and then to collect statistics for each"},{"line_number":73,"context_line":"interface attached to the bridge. These statistics could then"},{"line_number":74,"context_line":"be mapped to appropriate places, as discussed below."}],"source_content_type":"text/x-rst","patch_set":25,"id":"9a1a9d01_0389fa74","line":71,"updated":"2015-09-29 14:02:27.000000000","message":"you can also opt to parse relevant files in /proc/net, if that\u0027s a valid alternative","commit_id":"681351e0628e571ca3802a90b6d6db034c000179"},{"author":{"_account_id":261,"name":"Salvatore Orlando","email":"salv.orlando@gmail.com","username":"salvatore-orlando"},"change_message_id":"c5e3aedfbdf52c526bf5d3e0e6cc4f6f182a4ca4","unresolved":false,"context_lines":[{"line_number":288,"context_line":"Since core Neutron constructs handle up through layer three,"},{"line_number":289,"context_line":"these two RFCs are declared out of scope for this devref as they"},{"line_number":290,"context_line":"are more appropriate when instrumenting higher protocol services"},{"line_number":291,"context_line":"(such as the \\*aaS family)."},{"line_number":292,"context_line":""},{"line_number":293,"context_line":"Data Collection"},{"line_number":294,"context_line":"---------------"}],"source_content_type":"text/x-rst","patch_set":25,"id":"9a1a9d01_033eda8b","line":291,"updated":"2015-09-29 14:02:27.000000000","message":"I agree, but allow me to ask this question... why do you felt compelled to reference them?","commit_id":"681351e0628e571ca3802a90b6d6db034c000179"},{"author":{"_account_id":261,"name":"Salvatore Orlando","email":"salv.orlando@gmail.com","username":"salvatore-orlando"},"change_message_id":"c5e3aedfbdf52c526bf5d3e0e6cc4f6f182a4ca4","unresolved":false,"context_lines":[{"line_number":318,"context_line":"asynchronous notification requires that the aggregator have the capability"},{"line_number":319,"context_line":"to hold the current results from each harvester.  Given the current"},{"line_number":320,"context_line":"tradeoff between processing power and storage, the second approach makes"},{"line_number":321,"context_line":"more sense in this interation."},{"line_number":322,"context_line":""},{"line_number":323,"context_line":"A corollary to the above paragraph is whether or not the cached counters"},{"line_number":324,"context_line":"collected by the Neutron server need to be persisted. Our choices"}],"source_content_type":"text/x-rst","patch_set":25,"id":"9a1a9d01_e32c0604","line":321,"updated":"2015-09-29 14:02:27.000000000","message":"I am you considered also the impact of notification frequency and the load it will impose on the harvester but most importantly on the collector, for not mentioning the message bus.\n\nAs I have been not able to infer from the document when these notifications should be generated (e.g.: every time a counter changes by more than 0.1%) I am not able to formulate an opinion regarding whether notifications, and their asynchronous processing is going to scale better than polling.\n\nAlso, where the collector lives has quite an impact from a scalability perspective, and therefore this might also impact the choice of the communication paradigm.","commit_id":"681351e0628e571ca3802a90b6d6db034c000179"},{"author":{"_account_id":6854,"name":"YAMAMOTO Takashi","email":"yamamoto@midokura.com","username":"yamamoto"},"change_message_id":"ed301c106952eb6fbe3d052e2249060732ea4631","unresolved":false,"context_lines":[{"line_number":325,"context_line":"are to persist in a database (which seems like overkill), persist via a"},{"line_number":326,"context_line":"disk file (which can be slow) or not persist and only hold in memory"},{"line_number":327,"context_line":"(which has restart and potential scaling issues). For this PoC/iteration,"},{"line_number":328,"context_line":"the \"persist via a disk file\" approach will be taken, which leads in"},{"line_number":329,"context_line":"turn to what format will the disk file use. Since the data being passed"},{"line_number":330,"context_line":"via the asynchronous RPC notification will be in json format, it makes"},{"line_number":331,"context_line":"sense to use a json format for the persistence file."}],"source_content_type":"text/x-rst","patch_set":25,"id":"ba15a1d1_aac6cd97","line":328,"updated":"2015-09-17 10:12:19.000000000","message":"does it imply a single neutron server instance?","commit_id":"681351e0628e571ca3802a90b6d6db034c000179"},{"author":{"_account_id":261,"name":"Salvatore Orlando","email":"salv.orlando@gmail.com","username":"salvatore-orlando"},"change_message_id":"c5e3aedfbdf52c526bf5d3e0e6cc4f6f182a4ca4","unresolved":false,"context_lines":[{"line_number":325,"context_line":"are to persist in a database (which seems like overkill), persist via a"},{"line_number":326,"context_line":"disk file (which can be slow) or not persist and only hold in memory"},{"line_number":327,"context_line":"(which has restart and potential scaling issues). For this PoC/iteration,"},{"line_number":328,"context_line":"the \"persist via a disk file\" approach will be taken, which leads in"},{"line_number":329,"context_line":"turn to what format will the disk file use. Since the data being passed"},{"line_number":330,"context_line":"via the asynchronous RPC notification will be in json format, it makes"},{"line_number":331,"context_line":"sense to use a json format for the persistence file."}],"source_content_type":"text/x-rst","patch_set":25,"id":"9a1a9d01_c30ea28b","line":328,"in_reply_to":"ba15a1d1_1ed23b35","updated":"2015-09-29 14:02:27.000000000","message":"A database table is simple for me. Even simpler than a file, honestly.","commit_id":"681351e0628e571ca3802a90b6d6db034c000179"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"d93cf2f08153abb008a0b4899016b6ceefcdc19a","unresolved":false,"context_lines":[{"line_number":325,"context_line":"are to persist in a database (which seems like overkill), persist via a"},{"line_number":326,"context_line":"disk file (which can be slow) or not persist and only hold in memory"},{"line_number":327,"context_line":"(which has restart and potential scaling issues). For this PoC/iteration,"},{"line_number":328,"context_line":"the \"persist via a disk file\" approach will be taken, which leads in"},{"line_number":329,"context_line":"turn to what format will the disk file use. Since the data being passed"},{"line_number":330,"context_line":"via the asynchronous RPC notification will be in json format, it makes"},{"line_number":331,"context_line":"sense to use a json format for the persistence file."}],"source_content_type":"text/x-rst","patch_set":25,"id":"ba15a1d1_1ed23b35","line":328,"in_reply_to":"ba15a1d1_aac6cd97","updated":"2015-09-20 15:09:03.000000000","message":"Unfortunately, it does and that is an excellent point.  My preference would be to separate the rest of the design from the persistence method so that we can start with something simple and replace it with something more robust.","commit_id":"681351e0628e571ca3802a90b6d6db034c000179"},{"author":{"_account_id":6854,"name":"YAMAMOTO Takashi","email":"yamamoto@midokura.com","username":"yamamoto"},"change_message_id":"ed301c106952eb6fbe3d052e2249060732ea4631","unresolved":false,"context_lines":[{"line_number":328,"context_line":"the \"persist via a disk file\" approach will be taken, which leads in"},{"line_number":329,"context_line":"turn to what format will the disk file use. Since the data being passed"},{"line_number":330,"context_line":"via the asynchronous RPC notification will be in json format, it makes"},{"line_number":331,"context_line":"sense to use a json format for the persistence file."},{"line_number":332,"context_line":""},{"line_number":333,"context_line":"The result is each agent would have a harvester thread that periodically"},{"line_number":334,"context_line":"notify the Neutron server of updated counter information. When the Neutron"}],"source_content_type":"text/x-rst","patch_set":25,"id":"ba15a1d1_aa980dc1","line":331,"updated":"2015-09-17 10:12:19.000000000","message":"can you estimate how much data it will have?\n\ni tend to think a data file with home-grown format is almost always a bad idea.","commit_id":"681351e0628e571ca3802a90b6d6db034c000179"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"d93cf2f08153abb008a0b4899016b6ceefcdc19a","unresolved":false,"context_lines":[{"line_number":328,"context_line":"the \"persist via a disk file\" approach will be taken, which leads in"},{"line_number":329,"context_line":"turn to what format will the disk file use. Since the data being passed"},{"line_number":330,"context_line":"via the asynchronous RPC notification will be in json format, it makes"},{"line_number":331,"context_line":"sense to use a json format for the persistence file."},{"line_number":332,"context_line":""},{"line_number":333,"context_line":"The result is each agent would have a harvester thread that periodically"},{"line_number":334,"context_line":"notify the Neutron server of updated counter information. When the Neutron"}],"source_content_type":"text/x-rst","patch_set":25,"id":"ba15a1d1_dedb634e","line":331,"in_reply_to":"ba15a1d1_aa980dc1","updated":"2015-09-20 15:09:03.000000000","message":"As implied above, I\u0027ll see about spinning an update that pulls persistence out into a separate section where issues like that can be discussed.","commit_id":"681351e0628e571ca3802a90b6d6db034c000179"},{"author":{"_account_id":6854,"name":"YAMAMOTO Takashi","email":"yamamoto@midokura.com","username":"yamamoto"},"change_message_id":"1c7839574530bbd431d79c6cd1d0aab93a7b3735","unresolved":false,"context_lines":[{"line_number":117,"context_line":"* for each switch:"},{"line_number":118,"context_line":""},{"line_number":119,"context_line":"  * lookup and matched for each flow table"},{"line_number":120,"context_line":""},{"line_number":121,"context_line":"Mapping from Available Information to MIB Data Set"},{"line_number":122,"context_line":"~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~"},{"line_number":123,"context_line":""}],"source_content_type":"text/x-rst","patch_set":26,"id":"7a2fa921_f540b909","line":120,"updated":"2015-10-06 05:34:32.000000000","message":"have you considered NetFlow/sFlow?","commit_id":"4ad02c2bf16f059b646acff5b54a4f7d95598470"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"b2d8a61766cd473a53e12a0128da717fa640cac2","unresolved":false,"context_lines":[{"line_number":117,"context_line":"* for each switch:"},{"line_number":118,"context_line":""},{"line_number":119,"context_line":"  * lookup and matched for each flow table"},{"line_number":120,"context_line":""},{"line_number":121,"context_line":"Mapping from Available Information to MIB Data Set"},{"line_number":122,"context_line":"~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~"},{"line_number":123,"context_line":""}],"source_content_type":"text/x-rst","patch_set":26,"id":"3a29b11f_e14ae6ab","line":120,"in_reply_to":"7a2fa921_f540b909","updated":"2015-10-20 17:36:35.000000000","message":"Netflow/sflow could certainly be considered at a later date - this devref is more focused on what is there now, without adding additional packages","commit_id":"4ad02c2bf16f059b646acff5b54a4f7d95598470"},{"author":{"_account_id":6854,"name":"YAMAMOTO Takashi","email":"yamamoto@midokura.com","username":"yamamoto"},"change_message_id":"1c7839574530bbd431d79c6cd1d0aab93a7b3735","unresolved":false,"context_lines":[{"line_number":358,"context_line":"While it would be preferrable to use the same model for communication to"},{"line_number":359,"context_line":"the upstream consumer, that creates an a priori dependency within Neutron to"},{"line_number":360,"context_line":"know the upstream consumer\u0027s internals. Instead, the neutron server will"},{"line_number":361,"context_line":"provide a set of RPC calls that can be accessed by the upstream consumer"},{"line_number":362,"context_line":"to retrieve aggregated instrumentation data."},{"line_number":363,"context_line":""},{"line_number":364,"context_line":"Data Storage"}],"source_content_type":"text/x-rst","patch_set":26,"id":"7a2fa921_b5d1e1fd","line":361,"range":{"start_line":361,"start_character":17,"end_line":361,"end_character":20},"updated":"2015-10-06 05:34:32.000000000","message":"amqp rpc is meant here?\n\nthe consumer calls neutron?\n\nwhy not rest api?","commit_id":"4ad02c2bf16f059b646acff5b54a4f7d95598470"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"b2d8a61766cd473a53e12a0128da717fa640cac2","unresolved":false,"context_lines":[{"line_number":358,"context_line":"While it would be preferrable to use the same model for communication to"},{"line_number":359,"context_line":"the upstream consumer, that creates an a priori dependency within Neutron to"},{"line_number":360,"context_line":"know the upstream consumer\u0027s internals. Instead, the neutron server will"},{"line_number":361,"context_line":"provide a set of RPC calls that can be accessed by the upstream consumer"},{"line_number":362,"context_line":"to retrieve aggregated instrumentation data."},{"line_number":363,"context_line":""},{"line_number":364,"context_line":"Data Storage"}],"source_content_type":"text/x-rst","patch_set":26,"id":"3a29b11f_01604200","line":361,"in_reply_to":"7a2fa921_b5d1e1fd","updated":"2015-10-20 17:36:35.000000000","message":"Yes, AMQP rpc is mean here.\n\nYes, the consumer would call neutron.\n\nBy, \"REST API\", I assume you mean HTTP?  I wasn\u0027t looking to add to the NB API","commit_id":"4ad02c2bf16f059b646acff5b54a4f7d95598470"},{"author":{"_account_id":6854,"name":"YAMAMOTO Takashi","email":"yamamoto@midokura.com","username":"yamamoto"},"change_message_id":"1c7839574530bbd431d79c6cd1d0aab93a7b3735","unresolved":false,"context_lines":[{"line_number":359,"context_line":"the upstream consumer, that creates an a priori dependency within Neutron to"},{"line_number":360,"context_line":"know the upstream consumer\u0027s internals. Instead, the neutron server will"},{"line_number":361,"context_line":"provide a set of RPC calls that can be accessed by the upstream consumer"},{"line_number":362,"context_line":"to retrieve aggregated instrumentation data."},{"line_number":363,"context_line":""},{"line_number":364,"context_line":"Data Storage"},{"line_number":365,"context_line":"~~~~~~~~~~~~"}],"source_content_type":"text/x-rst","patch_set":26,"id":"7a2fa921_f5fd196f","line":362,"updated":"2015-10-06 05:34:32.000000000","message":"the aggregated data is about logical entities\nand thus (mostly) independent from specific plugins/drivers?","commit_id":"4ad02c2bf16f059b646acff5b54a4f7d95598470"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"b2d8a61766cd473a53e12a0128da717fa640cac2","unresolved":false,"context_lines":[{"line_number":359,"context_line":"the upstream consumer, that creates an a priori dependency within Neutron to"},{"line_number":360,"context_line":"know the upstream consumer\u0027s internals. Instead, the neutron server will"},{"line_number":361,"context_line":"provide a set of RPC calls that can be accessed by the upstream consumer"},{"line_number":362,"context_line":"to retrieve aggregated instrumentation data."},{"line_number":363,"context_line":""},{"line_number":364,"context_line":"Data Storage"},{"line_number":365,"context_line":"~~~~~~~~~~~~"}],"source_content_type":"text/x-rst","patch_set":26,"id":"3a29b11f_41139a66","line":362,"in_reply_to":"7a2fa921_f5fd196f","updated":"2015-10-20 17:36:35.000000000","message":"yes","commit_id":"4ad02c2bf16f059b646acff5b54a4f7d95598470"},{"author":{"_account_id":6854,"name":"YAMAMOTO Takashi","email":"yamamoto@midokura.com","username":"yamamoto"},"change_message_id":"1c7839574530bbd431d79c6cd1d0aab93a7b3735","unresolved":false,"context_lines":[{"line_number":366,"context_line":""},{"line_number":367,"context_line":"As discussed above under the heading of architecture, once an aggregation"},{"line_number":368,"context_line":"point is introduced, the question of data storage comes into play."},{"line_number":369,"context_line":"Because instrumentation is not aboue making Neutron into a bitbucket, data"},{"line_number":370,"context_line":"storage is limited to only a single data point for each collected counter at"},{"line_number":371,"context_line":"the Neutron server."},{"line_number":372,"context_line":""}],"source_content_type":"text/x-rst","patch_set":26,"id":"7a2fa921_95ee25eb","line":369,"range":{"start_line":369,"start_character":31,"end_line":369,"end_character":36},"updated":"2015-10-06 05:34:32.000000000","message":"typo: about","commit_id":"4ad02c2bf16f059b646acff5b54a4f7d95598470"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"b2d8a61766cd473a53e12a0128da717fa640cac2","unresolved":false,"context_lines":[{"line_number":366,"context_line":""},{"line_number":367,"context_line":"As discussed above under the heading of architecture, once an aggregation"},{"line_number":368,"context_line":"point is introduced, the question of data storage comes into play."},{"line_number":369,"context_line":"Because instrumentation is not aboue making Neutron into a bitbucket, data"},{"line_number":370,"context_line":"storage is limited to only a single data point for each collected counter at"},{"line_number":371,"context_line":"the Neutron server."},{"line_number":372,"context_line":""}],"source_content_type":"text/x-rst","patch_set":26,"id":"3a29b11f_010e023b","line":369,"in_reply_to":"7a2fa921_95ee25eb","updated":"2015-10-20 17:36:35.000000000","message":"Done in next draft","commit_id":"4ad02c2bf16f059b646acff5b54a4f7d95598470"},{"author":{"_account_id":6854,"name":"YAMAMOTO Takashi","email":"yamamoto@midokura.com","username":"yamamoto"},"change_message_id":"1c7839574530bbd431d79c6cd1d0aab93a7b3735","unresolved":false,"context_lines":[{"line_number":387,"context_line":"option will be persued."},{"line_number":388,"context_line":""},{"line_number":389,"context_line":"The proposed database table will have a row for each tap, qr-, or bridge"},{"line_number":390,"context_line":"interface, and columns for each of the collected counters listed above."},{"line_number":391,"context_line":"In addition, there will be a column storing of the interface identifier"},{"line_number":392,"context_line":"(to act as a key) as well as a column storing the network or router uuid"},{"line_number":393,"context_line":"(to aid the aggregation query). Further details are left to the"}],"source_content_type":"text/x-rst","patch_set":26,"id":"7a2fa921_d5925d57","line":390,"updated":"2015-10-06 05:34:32.000000000","message":"does this imply that this table is specific to plugin/driver?  eg. ml2/ovs","commit_id":"4ad02c2bf16f059b646acff5b54a4f7d95598470"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"b2d8a61766cd473a53e12a0128da717fa640cac2","unresolved":false,"context_lines":[{"line_number":387,"context_line":"option will be persued."},{"line_number":388,"context_line":""},{"line_number":389,"context_line":"The proposed database table will have a row for each tap, qr-, or bridge"},{"line_number":390,"context_line":"interface, and columns for each of the collected counters listed above."},{"line_number":391,"context_line":"In addition, there will be a column storing of the interface identifier"},{"line_number":392,"context_line":"(to act as a key) as well as a column storing the network or router uuid"},{"line_number":393,"context_line":"(to aid the aggregation query). Further details are left to the"}],"source_content_type":"text/x-rst","patch_set":26,"id":"3a29b11f_8c4ec96d","line":390,"in_reply_to":"7a2fa921_d5925d57","updated":"2015-10-20 17:36:35.000000000","message":"I don\u0027t see another option to the alternative is using the DB table to hold a bunch of NVPs and that\u0027s not really a good use of a DB.","commit_id":"4ad02c2bf16f059b646acff5b54a4f7d95598470"},{"author":{"_account_id":7787,"name":"Kevin Benton","email":"kevin@benton.pub","username":"blak111"},"change_message_id":"0c9cc4283112e8f085f175226e5d769449f6bd50","unresolved":false,"context_lines":[{"line_number":172,"context_line":"  * ifHCInOctets \u003d 64 bits of port received byte count"},{"line_number":173,"context_line":"  * ifHCInUcastPkts \u003d 64 bits of port received packet count"},{"line_number":174,"context_line":"  * ifHCOctOctets \u003d 64 bits of port transmit byte count"},{"line_number":175,"context_line":"  * ifHCOctUcastPkts \u003d 64 bits of port transmit packet count"},{"line_number":176,"context_line":""},{"line_number":177,"context_line":"For Neuton Networks that do not have DVRs, the counter from"},{"line_number":178,"context_line":"the qr ports associated with that particular network on br-int"}],"source_content_type":"text/x-rst","patch_set":27,"id":"3a29b11f_817b79bd","line":175,"updated":"2015-10-23 03:26:35.000000000","message":"can you add a small explanation of the discrepancy between the bit counts between ifTable and ifXTable? We need to make sure that\u0027s well-understood by operators since the 32 bit one will overflow quite frequently.","commit_id":"0ab6e7e71168e04284ed3bbeb168c152caaf7b01"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"e065de533f89370a3dbb2240dba461834c9690ec","unresolved":false,"context_lines":[{"line_number":172,"context_line":"  * ifHCInOctets \u003d 64 bits of port received byte count"},{"line_number":173,"context_line":"  * ifHCInUcastPkts \u003d 64 bits of port received packet count"},{"line_number":174,"context_line":"  * ifHCOctOctets \u003d 64 bits of port transmit byte count"},{"line_number":175,"context_line":"  * ifHCOctUcastPkts \u003d 64 bits of port transmit packet count"},{"line_number":176,"context_line":""},{"line_number":177,"context_line":"For Neuton Networks that do not have DVRs, the counter from"},{"line_number":178,"context_line":"the qr ports associated with that particular network on br-int"}],"source_content_type":"text/x-rst","patch_set":27,"id":"3a29b11f_371652f2","line":175,"in_reply_to":"3a29b11f_817b79bd","updated":"2015-10-23 14:01:34.000000000","message":"That\u0027s an excellent suggestion, I\u0027ll add it into the next draft","commit_id":"0ab6e7e71168e04284ed3bbeb168c152caaf7b01"},{"author":{"_account_id":7787,"name":"Kevin Benton","email":"kevin@benton.pub","username":"blak111"},"change_message_id":"0c9cc4283112e8f085f175226e5d769449f6bd50","unresolved":false,"context_lines":[{"line_number":353,"context_line":"a whole separate polling cycle just for collection instrumentation data."},{"line_number":354,"context_line":"The result is each agent\u0027s status thread would also return updated counter"},{"line_number":355,"context_line":"information to the Neutron server. When the Neutron server receives"},{"line_number":356,"context_line":"these counter updates, it replaces the corresponding values in the"},{"line_number":357,"context_line":"database table."},{"line_number":358,"context_line":""},{"line_number":359,"context_line":"An alternative would be to implement a new thread for pushing instrumentation"}],"source_content_type":"text/x-rst","patch_set":27,"id":"3a29b11f_214dc599","line":356,"updated":"2015-10-23 03:26:35.000000000","message":"I\u0027m for this idea. We are already moving the report_state messages onto a new topic that will ultimately have it\u0027s own dedicated consumer. report_state messages need to be processed very quickly by the neutron server due to their loss triggering rescheduling in busy environments.","commit_id":"0ab6e7e71168e04284ed3bbeb168c152caaf7b01"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"e065de533f89370a3dbb2240dba461834c9690ec","unresolved":false,"context_lines":[{"line_number":353,"context_line":"a whole separate polling cycle just for collection instrumentation data."},{"line_number":354,"context_line":"The result is each agent\u0027s status thread would also return updated counter"},{"line_number":355,"context_line":"information to the Neutron server. When the Neutron server receives"},{"line_number":356,"context_line":"these counter updates, it replaces the corresponding values in the"},{"line_number":357,"context_line":"database table."},{"line_number":358,"context_line":""},{"line_number":359,"context_line":"An alternative would be to implement a new thread for pushing instrumentation"}],"source_content_type":"text/x-rst","patch_set":27,"id":"3a29b11f_17755631","line":356,"in_reply_to":"3a29b11f_214dc599","updated":"2015-10-23 14:01:34.000000000","message":"Kevin, while this isn\u0027t really part of this devref, I\u0027ve gone back and forth on whether this collection should be part of the report_state message or in its own rpc call.  \n\nOn the one hand, report_state is already there, but on the other hand, I don\u0027t want to bog report_state message processing down with this type of update.","commit_id":"0ab6e7e71168e04284ed3bbeb168c152caaf7b01"},{"author":{"_account_id":7787,"name":"Kevin Benton","email":"kevin@benton.pub","username":"blak111"},"change_message_id":"0c9cc4283112e8f085f175226e5d769449f6bd50","unresolved":false,"context_lines":[{"line_number":360,"context_line":"data. This could be used if it is determined that the status thread is not"},{"line_number":361,"context_line":"providing fresh enough instrumentation data."},{"line_number":362,"context_line":""},{"line_number":363,"context_line":"While it would be preferrable to use the same model for communication to"},{"line_number":364,"context_line":"the upstream consumer, that creates an a priori dependency within Neutron to"},{"line_number":365,"context_line":"know the upstream consumer\u0027s internals. Instead, the neutron server will"},{"line_number":366,"context_line":"provide a set of AMQP RPC calls that can be accessed by the upstream"}],"source_content_type":"text/x-rst","patch_set":27,"id":"3a29b11f_41500132","line":363,"updated":"2015-10-23 03:26:35.000000000","message":"I\u0027m -1 to exposing this via AMQP. The only stuff that calls the server via the AMQP bus is internal stuff. Our AMQP APIs are only guarunteed be backward compatible N-1 releases so I don\u0027t want to have a one-off one that needs to be handled as a public API.\n\nAMQP is also an implementation detail of our reference implementation. Other plugins may avoid having AMQP consumers entirely.\n\nHTTP API should be fine for this one.","commit_id":"0ab6e7e71168e04284ed3bbeb168c152caaf7b01"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"e065de533f89370a3dbb2240dba461834c9690ec","unresolved":false,"context_lines":[{"line_number":360,"context_line":"data. This could be used if it is determined that the status thread is not"},{"line_number":361,"context_line":"providing fresh enough instrumentation data."},{"line_number":362,"context_line":""},{"line_number":363,"context_line":"While it would be preferrable to use the same model for communication to"},{"line_number":364,"context_line":"the upstream consumer, that creates an a priori dependency within Neutron to"},{"line_number":365,"context_line":"know the upstream consumer\u0027s internals. Instead, the neutron server will"},{"line_number":366,"context_line":"provide a set of AMQP RPC calls that can be accessed by the upstream"}],"source_content_type":"text/x-rst","patch_set":27,"id":"3a29b11f_97fc0674","line":363,"in_reply_to":"3a29b11f_41500132","updated":"2015-10-23 14:01:34.000000000","message":"Ok, that was the conversation I was looking to have with you in the IRC channel esterday :) , so I will change this in the next draft and add a section that documents what the HTTP API could look like for folks to review.\n\nOne thing that I haven\u0027t quite worked out yet is how to get the plugin specific additional data (see below) out through the HTTP API - I\u0027m open to suggestions...","commit_id":"0ab6e7e71168e04284ed3bbeb168c152caaf7b01"},{"author":{"_account_id":7787,"name":"Kevin Benton","email":"kevin@benton.pub","username":"blak111"},"change_message_id":"0c9cc4283112e8f085f175226e5d769449f6bd50","unresolved":false,"context_lines":[{"line_number":385,"context_line":""},{"line_number":386,"context_line":"Using memory to hold the data has the shortcoming of creating a point of"},{"line_number":387,"context_line":"discontinuity if the server process is restarted. Storing in memory or"},{"line_number":388,"context_line":"via a disk file share the shortcoming of requiring the aggregation"},{"line_number":389,"context_line":"calculations to be done in python code, which could create scale issues."},{"line_number":390,"context_line":"On the other hand, using a table in the neutron database would allow the"},{"line_number":391,"context_line":"necessary aggregations to be done in the database itself, an operation that"}],"source_content_type":"text/x-rst","patch_set":27,"id":"3a29b11f_e1426d67","line":388,"updated":"2015-10-23 03:26:35.000000000","message":"definitely. the other two options also fail too offer up-to-date information in multi-server deployments unless we have every server consume every message, which is also a scalability nightmare.","commit_id":"0ab6e7e71168e04284ed3bbeb168c152caaf7b01"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"e065de533f89370a3dbb2240dba461834c9690ec","unresolved":false,"context_lines":[{"line_number":385,"context_line":""},{"line_number":386,"context_line":"Using memory to hold the data has the shortcoming of creating a point of"},{"line_number":387,"context_line":"discontinuity if the server process is restarted. Storing in memory or"},{"line_number":388,"context_line":"via a disk file share the shortcoming of requiring the aggregation"},{"line_number":389,"context_line":"calculations to be done in python code, which could create scale issues."},{"line_number":390,"context_line":"On the other hand, using a table in the neutron database would allow the"},{"line_number":391,"context_line":"necessary aggregations to be done in the database itself, an operation that"}],"source_content_type":"text/x-rst","patch_set":27,"id":"3a29b11f_f7153a9b","line":388,"in_reply_to":"3a29b11f_e1426d67","updated":"2015-10-23 14:01:34.000000000","message":"I had that one written down as well and then forgot to add it - since I\u0027m respinning anyway, I\u0027ll add that limitation to the list...","commit_id":"0ab6e7e71168e04284ed3bbeb168c152caaf7b01"},{"author":{"_account_id":7787,"name":"Kevin Benton","email":"kevin@benton.pub","username":"blak111"},"change_message_id":"0c9cc4283112e8f085f175226e5d769449f6bd50","unresolved":false,"context_lines":[{"line_number":396,"context_line":"each tap, qr-, or bridge interface, and columns for each of the collected"},{"line_number":397,"context_line":"counters listed above. In addition, there will be a column storing of"},{"line_number":398,"context_line":"the interface identifier (to act as a key) as well as a column storing"},{"line_number":399,"context_line":"the network or router uuid (to aid the aggregation query). Further details"},{"line_number":400,"context_line":"are left to the implementation. Note: while having specific tables for each"},{"line_number":401,"context_line":"plugin/driver may not seem optimal, trying to collect information from"},{"line_number":402,"context_line":"different plugin/drivers into a single table quickly leads to the table"}],"source_content_type":"text/x-rst","patch_set":27,"id":"3a29b11f_0146897a","line":399,"updated":"2015-10-23 03:26:35.000000000","message":"Can we not agree on some lowest common denominator that can be expected of all plugins? Then the plugins can just each add an additional table for plugin-specific items.","commit_id":"0ab6e7e71168e04284ed3bbeb168c152caaf7b01"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"e065de533f89370a3dbb2240dba461834c9690ec","unresolved":false,"context_lines":[{"line_number":396,"context_line":"each tap, qr-, or bridge interface, and columns for each of the collected"},{"line_number":397,"context_line":"counters listed above. In addition, there will be a column storing of"},{"line_number":398,"context_line":"the interface identifier (to act as a key) as well as a column storing"},{"line_number":399,"context_line":"the network or router uuid (to aid the aggregation query). Further details"},{"line_number":400,"context_line":"are left to the implementation. Note: while having specific tables for each"},{"line_number":401,"context_line":"plugin/driver may not seem optimal, trying to collect information from"},{"line_number":402,"context_line":"different plugin/drivers into a single table quickly leads to the table"}],"source_content_type":"text/x-rst","patch_set":27,"id":"3a29b11f_378f32ce","line":399,"in_reply_to":"3a29b11f_0146897a","updated":"2015-10-23 14:01:34.000000000","message":"Yes, since we will be specifying a draft HTTP API, it makes sense to me to do that as well.","commit_id":"0ab6e7e71168e04284ed3bbeb168c152caaf7b01"},{"author":{"_account_id":14611,"name":"dasm","display_name":"dasm","email":"dsmigiel@redhat.com","username":"dasm"},"change_message_id":"abf8cb199655705a68013ba08542282063664cf8","unresolved":false,"context_lines":[{"line_number":65,"context_line":"to define either their own data sets and/or their own mappings"},{"line_number":66,"context_line":"to the data sets defined in the referenced RFCs."},{"line_number":67,"context_line":""},{"line_number":68,"context_line":"Focus here is on what is available from \"stock\" linux and openstack."},{"line_number":69,"context_line":"Additional statistics may become available if other items like NetFlow or"},{"line_number":70,"context_line":"sFlow are added to the mix, but those should be covered as an addition to"},{"line_number":71,"context_line":"the basic information discussed here."}],"source_content_type":"text/x-rst","patch_set":28,"id":"da85f559_804bed2a","line":68,"updated":"2015-11-15 18:05:32.000000000","message":"nit: s/linux/Linux and s/openstack/OpenStack","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"97a0f84c8573789d4c93d1df67d43d321e0c6c43","unresolved":false,"context_lines":[{"line_number":65,"context_line":"to define either their own data sets and/or their own mappings"},{"line_number":66,"context_line":"to the data sets defined in the referenced RFCs."},{"line_number":67,"context_line":""},{"line_number":68,"context_line":"Focus here is on what is available from \"stock\" linux and openstack."},{"line_number":69,"context_line":"Additional statistics may become available if other items like NetFlow or"},{"line_number":70,"context_line":"sFlow are added to the mix, but those should be covered as an addition to"},{"line_number":71,"context_line":"the basic information discussed here."}],"source_content_type":"text/x-rst","patch_set":28,"id":"ba8a016a_a53a0acc","line":68,"in_reply_to":"da85f559_804bed2a","updated":"2015-11-16 14:51:56.000000000","message":"Done","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":14611,"name":"dasm","display_name":"dasm","email":"dsmigiel@redhat.com","username":"dasm"},"change_message_id":"abf8cb199655705a68013ba08542282063664cf8","unresolved":false,"context_lines":[{"line_number":73,"context_line":"What is Available from Nova"},{"line_number":74,"context_line":"~~~~~~~~~~~~~~~~~~~~~~~~~~~"},{"line_number":75,"context_line":""},{"line_number":76,"context_line":"Within nova, the libvirt driver makes the following host"},{"line_number":77,"context_line":"traffic statistics available under the get_diagnostics()"},{"line_number":78,"context_line":"and get_instance_diagnostics() calls:"},{"line_number":79,"context_line":""}],"source_content_type":"text/x-rst","patch_set":28,"id":"da85f559_a046311f","line":76,"updated":"2015-11-15 18:05:32.000000000","message":"s/nova/Nova","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"97a0f84c8573789d4c93d1df67d43d321e0c6c43","unresolved":false,"context_lines":[{"line_number":73,"context_line":"What is Available from Nova"},{"line_number":74,"context_line":"~~~~~~~~~~~~~~~~~~~~~~~~~~~"},{"line_number":75,"context_line":""},{"line_number":76,"context_line":"Within nova, the libvirt driver makes the following host"},{"line_number":77,"context_line":"traffic statistics available under the get_diagnostics()"},{"line_number":78,"context_line":"and get_instance_diagnostics() calls:"},{"line_number":79,"context_line":""}],"source_content_type":"text/x-rst","patch_set":28,"id":"ba8a016a_254dba72","line":76,"in_reply_to":"da85f559_a046311f","updated":"2015-11-16 14:51:56.000000000","message":"Done","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":14611,"name":"dasm","display_name":"dasm","email":"dsmigiel@redhat.com","username":"dasm"},"change_message_id":"abf8cb199655705a68013ba08542282063664cf8","unresolved":false,"context_lines":[{"line_number":101,"context_line":"Note: the examples below talk in terms of mapping counters"},{"line_number":102,"context_line":"available from the Linux operating system:"},{"line_number":103,"context_line":""},{"line_number":104,"context_line":"* receive bytes, packets, errors, dropped, overrun and multicast"},{"line_number":105,"context_line":"* transmit bytes, packets, errors, dropped, carrier and collisions"},{"line_number":106,"context_line":""},{"line_number":107,"context_line":"Available counters for interfaces on other operating systems"}],"source_content_type":"text/x-rst","patch_set":28,"id":"da85f559_005e1de9","line":104,"updated":"2015-11-15 18:05:32.000000000","message":"nit: for consistency in document: s/receive/Receive","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"97a0f84c8573789d4c93d1df67d43d321e0c6c43","unresolved":false,"context_lines":[{"line_number":101,"context_line":"Note: the examples below talk in terms of mapping counters"},{"line_number":102,"context_line":"available from the Linux operating system:"},{"line_number":103,"context_line":""},{"line_number":104,"context_line":"* receive bytes, packets, errors, dropped, overrun and multicast"},{"line_number":105,"context_line":"* transmit bytes, packets, errors, dropped, carrier and collisions"},{"line_number":106,"context_line":""},{"line_number":107,"context_line":"Available counters for interfaces on other operating systems"}],"source_content_type":"text/x-rst","patch_set":28,"id":"ba8a016a_055c3624","line":104,"in_reply_to":"da85f559_005e1de9","updated":"2015-11-16 14:51:56.000000000","message":"Done","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":14611,"name":"dasm","display_name":"dasm","email":"dsmigiel@redhat.com","username":"dasm"},"change_message_id":"abf8cb199655705a68013ba08542282063664cf8","unresolved":false,"context_lines":[{"line_number":102,"context_line":"available from the Linux operating system:"},{"line_number":103,"context_line":""},{"line_number":104,"context_line":"* receive bytes, packets, errors, dropped, overrun and multicast"},{"line_number":105,"context_line":"* transmit bytes, packets, errors, dropped, carrier and collisions"},{"line_number":106,"context_line":""},{"line_number":107,"context_line":"Available counters for interfaces on other operating systems"},{"line_number":108,"context_line":"can be mapped in a similar fashion."}],"source_content_type":"text/x-rst","patch_set":28,"id":"da85f559_2061e1a8","line":105,"updated":"2015-11-15 18:05:32.000000000","message":"ditto","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"97a0f84c8573789d4c93d1df67d43d321e0c6c43","unresolved":false,"context_lines":[{"line_number":102,"context_line":"available from the Linux operating system:"},{"line_number":103,"context_line":""},{"line_number":104,"context_line":"* receive bytes, packets, errors, dropped, overrun and multicast"},{"line_number":105,"context_line":"* transmit bytes, packets, errors, dropped, carrier and collisions"},{"line_number":106,"context_line":""},{"line_number":107,"context_line":"Available counters for interfaces on other operating systems"},{"line_number":108,"context_line":"can be mapped in a similar fashion."}],"source_content_type":"text/x-rst","patch_set":28,"id":"ba8a016a_a541aa3b","line":105,"in_reply_to":"da85f559_2061e1a8","updated":"2015-11-16 14:51:56.000000000","message":"Done","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":14611,"name":"dasm","display_name":"dasm","email":"dsmigiel@redhat.com","username":"dasm"},"change_message_id":"abf8cb199655705a68013ba08542282063664cf8","unresolved":false,"context_lines":[{"line_number":116,"context_line":""},{"line_number":117,"context_line":"* for each port:"},{"line_number":118,"context_line":""},{"line_number":119,"context_line":"  * receive packets, bytes, errors, dropped, overruns, frames"},{"line_number":120,"context_line":"  * transmit packets, bytes, errors, dropped, aborted, carrier"},{"line_number":121,"context_line":""},{"line_number":122,"context_line":"* for each switch:"}],"source_content_type":"text/x-rst","patch_set":28,"id":"da85f559_c057b5c8","line":119,"updated":"2015-11-15 18:05:32.000000000","message":"ditto","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"97a0f84c8573789d4c93d1df67d43d321e0c6c43","unresolved":false,"context_lines":[{"line_number":116,"context_line":""},{"line_number":117,"context_line":"* for each port:"},{"line_number":118,"context_line":""},{"line_number":119,"context_line":"  * receive packets, bytes, errors, dropped, overruns, frames"},{"line_number":120,"context_line":"  * transmit packets, bytes, errors, dropped, aborted, carrier"},{"line_number":121,"context_line":""},{"line_number":122,"context_line":"* for each switch:"}],"source_content_type":"text/x-rst","patch_set":28,"id":"ba8a016a_e56d72cd","line":119,"in_reply_to":"da85f559_c057b5c8","updated":"2015-11-16 14:51:56.000000000","message":"Done","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":14611,"name":"dasm","display_name":"dasm","email":"dsmigiel@redhat.com","username":"dasm"},"change_message_id":"abf8cb199655705a68013ba08542282063664cf8","unresolved":false,"context_lines":[{"line_number":117,"context_line":"* for each port:"},{"line_number":118,"context_line":""},{"line_number":119,"context_line":"  * receive packets, bytes, errors, dropped, overruns, frames"},{"line_number":120,"context_line":"  * transmit packets, bytes, errors, dropped, aborted, carrier"},{"line_number":121,"context_line":""},{"line_number":122,"context_line":"* for each switch:"},{"line_number":123,"context_line":""}],"source_content_type":"text/x-rst","patch_set":28,"id":"da85f559_e05af9f1","line":120,"updated":"2015-11-15 18:05:32.000000000","message":"ditto","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"97a0f84c8573789d4c93d1df67d43d321e0c6c43","unresolved":false,"context_lines":[{"line_number":117,"context_line":"* for each port:"},{"line_number":118,"context_line":""},{"line_number":119,"context_line":"  * receive packets, bytes, errors, dropped, overruns, frames"},{"line_number":120,"context_line":"  * transmit packets, bytes, errors, dropped, aborted, carrier"},{"line_number":121,"context_line":""},{"line_number":122,"context_line":"* for each switch:"},{"line_number":123,"context_line":""}],"source_content_type":"text/x-rst","patch_set":28,"id":"ba8a016a_a568cabb","line":120,"in_reply_to":"da85f559_e05af9f1","updated":"2015-11-16 14:51:56.000000000","message":"Done","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":14611,"name":"dasm","display_name":"dasm","email":"dsmigiel@redhat.com","username":"dasm"},"change_message_id":"abf8cb199655705a68013ba08542282063664cf8","unresolved":false,"context_lines":[{"line_number":176,"context_line":""},{"line_number":177,"context_line":"Section 3.1.6 of [RFC2863]_ provides the details of why 64-bit sized"},{"line_number":178,"context_line":"counters need to be supported. The summary is that with increasing"},{"line_number":179,"context_line":"transimission bandwidth use of 32 bit counters would require a"},{"line_number":180,"context_line":"problematic increase in counter polling frequency (a 1Gbs stream of"},{"line_number":181,"context_line":"full-sized packets will cause a 32 bit counter to wrap in 34 seconds)."},{"line_number":182,"context_line":""}],"source_content_type":"text/x-rst","patch_set":28,"id":"da85f559_60fe09b0","line":179,"updated":"2015-11-15 18:05:32.000000000","message":"s/32 bit/32-bit","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"97a0f84c8573789d4c93d1df67d43d321e0c6c43","unresolved":false,"context_lines":[{"line_number":176,"context_line":""},{"line_number":177,"context_line":"Section 3.1.6 of [RFC2863]_ provides the details of why 64-bit sized"},{"line_number":178,"context_line":"counters need to be supported. The summary is that with increasing"},{"line_number":179,"context_line":"transimission bandwidth use of 32 bit counters would require a"},{"line_number":180,"context_line":"problematic increase in counter polling frequency (a 1Gbs stream of"},{"line_number":181,"context_line":"full-sized packets will cause a 32 bit counter to wrap in 34 seconds)."},{"line_number":182,"context_line":""}],"source_content_type":"text/x-rst","patch_set":28,"id":"ba8a016a_05aaf6d4","line":179,"in_reply_to":"da85f559_60fe09b0","updated":"2015-11-16 14:51:56.000000000","message":"Done","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":14611,"name":"dasm","display_name":"dasm","email":"dsmigiel@redhat.com","username":"dasm"},"change_message_id":"abf8cb199655705a68013ba08542282063664cf8","unresolved":false,"context_lines":[{"line_number":178,"context_line":"counters need to be supported. The summary is that with increasing"},{"line_number":179,"context_line":"transimission bandwidth use of 32 bit counters would require a"},{"line_number":180,"context_line":"problematic increase in counter polling frequency (a 1Gbs stream of"},{"line_number":181,"context_line":"full-sized packets will cause a 32 bit counter to wrap in 34 seconds)."},{"line_number":182,"context_line":""},{"line_number":183,"context_line":"For Neuton Networks that do not have DVRs, the counter from"},{"line_number":184,"context_line":"the qr ports associated with that particular network on br-int"}],"source_content_type":"text/x-rst","patch_set":28,"id":"da85f559_202fa13f","line":181,"updated":"2015-11-15 18:05:32.000000000","message":"ditto","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"97a0f84c8573789d4c93d1df67d43d321e0c6c43","unresolved":false,"context_lines":[{"line_number":178,"context_line":"counters need to be supported. The summary is that with increasing"},{"line_number":179,"context_line":"transimission bandwidth use of 32 bit counters would require a"},{"line_number":180,"context_line":"problematic increase in counter polling frequency (a 1Gbs stream of"},{"line_number":181,"context_line":"full-sized packets will cause a 32 bit counter to wrap in 34 seconds)."},{"line_number":182,"context_line":""},{"line_number":183,"context_line":"For Neuton Networks that do not have DVRs, the counter from"},{"line_number":184,"context_line":"the qr ports associated with that particular network on br-int"}],"source_content_type":"text/x-rst","patch_set":28,"id":"ba8a016a_a5af6ae2","line":181,"in_reply_to":"da85f559_202fa13f","updated":"2015-11-16 14:51:56.000000000","message":"Done","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":14611,"name":"dasm","display_name":"dasm","email":"dsmigiel@redhat.com","username":"dasm"},"change_message_id":"abf8cb199655705a68013ba08542282063664cf8","unresolved":false,"context_lines":[{"line_number":180,"context_line":"problematic increase in counter polling frequency (a 1Gbs stream of"},{"line_number":181,"context_line":"full-sized packets will cause a 32 bit counter to wrap in 34 seconds)."},{"line_number":182,"context_line":""},{"line_number":183,"context_line":"For Neuton Networks that do not have DVRs, the counter from"},{"line_number":184,"context_line":"the qr ports associated with that particular network on br-int"},{"line_number":185,"context_line":"in the network node will also be mapped to rows in ifTable and"},{"line_number":186,"context_line":"ifXTable for network \"switch,\" using the above mappings."}],"source_content_type":"text/x-rst","patch_set":28,"id":"da85f559_c025755c","line":183,"updated":"2015-11-15 18:05:32.000000000","message":"typo: s/Neuton/Neutron","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"97a0f84c8573789d4c93d1df67d43d321e0c6c43","unresolved":false,"context_lines":[{"line_number":180,"context_line":"problematic increase in counter polling frequency (a 1Gbs stream of"},{"line_number":181,"context_line":"full-sized packets will cause a 32 bit counter to wrap in 34 seconds)."},{"line_number":182,"context_line":""},{"line_number":183,"context_line":"For Neuton Networks that do not have DVRs, the counter from"},{"line_number":184,"context_line":"the qr ports associated with that particular network on br-int"},{"line_number":185,"context_line":"in the network node will also be mapped to rows in ifTable and"},{"line_number":186,"context_line":"ifXTable for network \"switch,\" using the above mappings."}],"source_content_type":"text/x-rst","patch_set":28,"id":"ba8a016a_85b6e63b","line":183,"in_reply_to":"da85f559_c025755c","updated":"2015-11-16 14:51:56.000000000","message":"Done","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":14611,"name":"dasm","display_name":"dasm","email":"dsmigiel@redhat.com","username":"dasm"},"change_message_id":"abf8cb199655705a68013ba08542282063664cf8","unresolved":false,"context_lines":[{"line_number":183,"context_line":"For Neuton Networks that do not have DVRs, the counter from"},{"line_number":184,"context_line":"the qr ports associated with that particular network on br-int"},{"line_number":185,"context_line":"in the network node will also be mapped to rows in ifTable and"},{"line_number":186,"context_line":"ifXTable for network \"switch,\" using the above mappings."},{"line_number":187,"context_line":""},{"line_number":188,"context_line":"Lastly, for Neutron Networks that do have DVRs, then the counters"},{"line_number":189,"context_line":"from the qr ports associated with each network on br-int on"}],"source_content_type":"text/x-rst","patch_set":28,"id":"da85f559_e028b935","line":186,"updated":"2015-11-15 18:05:32.000000000","message":"s/switch,/switch","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"97a0f84c8573789d4c93d1df67d43d321e0c6c43","unresolved":false,"context_lines":[{"line_number":183,"context_line":"For Neuton Networks that do not have DVRs, the counter from"},{"line_number":184,"context_line":"the qr ports associated with that particular network on br-int"},{"line_number":185,"context_line":"in the network node will also be mapped to rows in ifTable and"},{"line_number":186,"context_line":"ifXTable for network \"switch,\" using the above mappings."},{"line_number":187,"context_line":""},{"line_number":188,"context_line":"Lastly, for Neutron Networks that do have DVRs, then the counters"},{"line_number":189,"context_line":"from the qr ports associated with each network on br-int on"}],"source_content_type":"text/x-rst","patch_set":28,"id":"ba8a016a_2505bac1","line":186,"in_reply_to":"da85f559_e028b935","updated":"2015-11-16 14:51:56.000000000","message":"This punctuation is correct - the comma sets off the parenthetical phrase  and when a quoted parenthetical ends in punctuation, the punctuation is included inside the quotes","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":14611,"name":"dasm","display_name":"dasm","email":"dsmigiel@redhat.com","username":"dasm"},"change_message_id":"abf8cb199655705a68013ba08542282063664cf8","unresolved":false,"context_lines":[{"line_number":225,"context_line":"particular bridge will be mapped to rows in these tables using"},{"line_number":226,"context_line":"the following counter assignments:"},{"line_number":227,"context_line":""},{"line_number":228,"context_line":"* receive bytes, packets, errors, dropped, overrun and multicast"},{"line_number":229,"context_line":"* transmit bytes, packets, errors, dropped, carrier and collisions"},{"line_number":230,"context_line":""},{"line_number":231,"context_line":"* ifTable"}],"source_content_type":"text/x-rst","patch_set":28,"id":"da85f559_20d6c127","line":228,"updated":"2015-11-15 18:05:32.000000000","message":"s/receive/Receive","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"97a0f84c8573789d4c93d1df67d43d321e0c6c43","unresolved":false,"context_lines":[{"line_number":225,"context_line":"particular bridge will be mapped to rows in these tables using"},{"line_number":226,"context_line":"the following counter assignments:"},{"line_number":227,"context_line":""},{"line_number":228,"context_line":"* receive bytes, packets, errors, dropped, overrun and multicast"},{"line_number":229,"context_line":"* transmit bytes, packets, errors, dropped, carrier and collisions"},{"line_number":230,"context_line":""},{"line_number":231,"context_line":"* ifTable"}],"source_content_type":"text/x-rst","patch_set":28,"id":"ba8a016a_6503c2b9","line":228,"in_reply_to":"da85f559_20d6c127","updated":"2015-11-16 14:51:56.000000000","message":"Done","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":14611,"name":"dasm","display_name":"dasm","email":"dsmigiel@redhat.com","username":"dasm"},"change_message_id":"abf8cb199655705a68013ba08542282063664cf8","unresolved":false,"context_lines":[{"line_number":226,"context_line":"the following counter assignments:"},{"line_number":227,"context_line":""},{"line_number":228,"context_line":"* receive bytes, packets, errors, dropped, overrun and multicast"},{"line_number":229,"context_line":"* transmit bytes, packets, errors, dropped, carrier and collisions"},{"line_number":230,"context_line":""},{"line_number":231,"context_line":"* ifTable"},{"line_number":232,"context_line":""}],"source_content_type":"text/x-rst","patch_set":28,"id":"da85f559_c0dc9548","line":229,"updated":"2015-11-15 18:05:32.000000000","message":"s/transmit/Transmit","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"97a0f84c8573789d4c93d1df67d43d321e0c6c43","unresolved":false,"context_lines":[{"line_number":226,"context_line":"the following counter assignments:"},{"line_number":227,"context_line":""},{"line_number":228,"context_line":"* receive bytes, packets, errors, dropped, overrun and multicast"},{"line_number":229,"context_line":"* transmit bytes, packets, errors, dropped, carrier and collisions"},{"line_number":230,"context_line":""},{"line_number":231,"context_line":"* ifTable"},{"line_number":232,"context_line":""}],"source_content_type":"text/x-rst","patch_set":28,"id":"ba8a016a_850026ac","line":229,"in_reply_to":"da85f559_c0dc9548","updated":"2015-11-16 14:51:56.000000000","message":"Done","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":14611,"name":"dasm","display_name":"dasm","email":"dsmigiel@redhat.com","username":"dasm"},"change_message_id":"abf8cb199655705a68013ba08542282063664cf8","unresolved":false,"context_lines":[{"line_number":323,"context_line":"~~~~~~~~~~~~"},{"line_number":324,"context_line":""},{"line_number":325,"context_line":"Obviously, the Neutron agents act as \"data harvesters\" (in that they collect"},{"line_number":326,"context_line":"the raw counts) and the upstream system is the ultimate \"data consumer.\" In"},{"line_number":327,"context_line":"between however, one is immediately presented with the choice of either"},{"line_number":328,"context_line":""},{"line_number":329,"context_line":"* Direct communication between the data harvesters and the data consumer, or"}],"source_content_type":"text/x-rst","patch_set":28,"id":"da85f559_60c36958","line":326,"updated":"2015-11-15 18:05:32.000000000","message":"s/consumer.\"/consumer\".","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"97a0f84c8573789d4c93d1df67d43d321e0c6c43","unresolved":false,"context_lines":[{"line_number":323,"context_line":"~~~~~~~~~~~~"},{"line_number":324,"context_line":""},{"line_number":325,"context_line":"Obviously, the Neutron agents act as \"data harvesters\" (in that they collect"},{"line_number":326,"context_line":"the raw counts) and the upstream system is the ultimate \"data consumer.\" In"},{"line_number":327,"context_line":"between however, one is immediately presented with the choice of either"},{"line_number":328,"context_line":""},{"line_number":329,"context_line":"* Direct communication between the data harvesters and the data consumer, or"}],"source_content_type":"text/x-rst","patch_set":28,"id":"ba8a016a_85eee6ea","line":326,"in_reply_to":"da85f559_60c36958","updated":"2015-11-16 14:51:56.000000000","message":"See above punctuation comment - this is correct as is","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":14611,"name":"dasm","display_name":"dasm","email":"dsmigiel@redhat.com","username":"dasm"},"change_message_id":"abf8cb199655705a68013ba08542282063664cf8","unresolved":false,"context_lines":[{"line_number":324,"context_line":""},{"line_number":325,"context_line":"Obviously, the Neutron agents act as \"data harvesters\" (in that they collect"},{"line_number":326,"context_line":"the raw counts) and the upstream system is the ultimate \"data consumer.\" In"},{"line_number":327,"context_line":"between however, one is immediately presented with the choice of either"},{"line_number":328,"context_line":""},{"line_number":329,"context_line":"* Direct communication between the data harvesters and the data consumer, or"},{"line_number":330,"context_line":"* Add a \"data aggregator\" that collects the information from the harvesters"}],"source_content_type":"text/x-rst","patch_set":28,"id":"da85f559_00a13da9","line":327,"updated":"2015-11-15 18:05:32.000000000","message":"s/either/either:","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"97a0f84c8573789d4c93d1df67d43d321e0c6c43","unresolved":false,"context_lines":[{"line_number":324,"context_line":""},{"line_number":325,"context_line":"Obviously, the Neutron agents act as \"data harvesters\" (in that they collect"},{"line_number":326,"context_line":"the raw counts) and the upstream system is the ultimate \"data consumer.\" In"},{"line_number":327,"context_line":"between however, one is immediately presented with the choice of either"},{"line_number":328,"context_line":""},{"line_number":329,"context_line":"* Direct communication between the data harvesters and the data consumer, or"},{"line_number":330,"context_line":"* Add a \"data aggregator\" that collects the information from the harvesters"}],"source_content_type":"text/x-rst","patch_set":28,"id":"ba8a016a_85c5865c","line":327,"in_reply_to":"da85f559_00a13da9","updated":"2015-11-16 14:51:56.000000000","message":"Done","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":14611,"name":"dasm","display_name":"dasm","email":"dsmigiel@redhat.com","username":"dasm"},"change_message_id":"abf8cb199655705a68013ba08542282063664cf8","unresolved":false,"context_lines":[{"line_number":353,"context_line":"the results from the current polling cycle, but this comes at the cost"},{"line_number":354,"context_line":"of dealing with the stale data issues that scaling a polling cycle will"},{"line_number":355,"context_line":"entail. On the other side, asynchronous notification requires that the"},{"line_number":356,"context_line":"aggregator have the capability to hold the current results from each"},{"line_number":357,"context_line":"harvester. Neutron agents already have a thread that provides status"},{"line_number":358,"context_line":"information back to the neutron server, so extending this thread to"},{"line_number":359,"context_line":"include instrumentation data would be a smaller change than implementing"}],"source_content_type":"text/x-rst","patch_set":28,"id":"da85f559_a090714f","line":356,"updated":"2015-11-15 18:05:32.000000000","message":"s/have/has","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"97a0f84c8573789d4c93d1df67d43d321e0c6c43","unresolved":false,"context_lines":[{"line_number":353,"context_line":"the results from the current polling cycle, but this comes at the cost"},{"line_number":354,"context_line":"of dealing with the stale data issues that scaling a polling cycle will"},{"line_number":355,"context_line":"entail. On the other side, asynchronous notification requires that the"},{"line_number":356,"context_line":"aggregator have the capability to hold the current results from each"},{"line_number":357,"context_line":"harvester. Neutron agents already have a thread that provides status"},{"line_number":358,"context_line":"information back to the neutron server, so extending this thread to"},{"line_number":359,"context_line":"include instrumentation data would be a smaller change than implementing"}],"source_content_type":"text/x-rst","patch_set":28,"id":"ba8a016a_c8108551","line":356,"in_reply_to":"da85f559_a090714f","updated":"2015-11-16 14:51:56.000000000","message":"Done","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":14611,"name":"dasm","display_name":"dasm","email":"dsmigiel@redhat.com","username":"dasm"},"change_message_id":"abf8cb199655705a68013ba08542282063664cf8","unresolved":false,"context_lines":[{"line_number":369,"context_line":""},{"line_number":370,"context_line":"While it would be preferrable to use the same model for communication to"},{"line_number":371,"context_line":"the upstream consumer, that creates an a priori dependency within Neutron to"},{"line_number":372,"context_line":"know the upstream consumer\u0027s internals. Instead, the neutron server will"},{"line_number":373,"context_line":"provide a set of resources and HTTP API that can be accessed by the upstream"},{"line_number":374,"context_line":"consumer to retrieve aggregated instrumentation data that ideally should"},{"line_number":375,"context_line":"be independent from the specific plugins/drivers."}],"source_content_type":"text/x-rst","patch_set":28,"id":"da85f559_c081f5fc","line":372,"updated":"2015-11-15 18:05:32.000000000","message":"s/neutron/Neutron","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"97a0f84c8573789d4c93d1df67d43d321e0c6c43","unresolved":false,"context_lines":[{"line_number":369,"context_line":""},{"line_number":370,"context_line":"While it would be preferrable to use the same model for communication to"},{"line_number":371,"context_line":"the upstream consumer, that creates an a priori dependency within Neutron to"},{"line_number":372,"context_line":"know the upstream consumer\u0027s internals. Instead, the neutron server will"},{"line_number":373,"context_line":"provide a set of resources and HTTP API that can be accessed by the upstream"},{"line_number":374,"context_line":"consumer to retrieve aggregated instrumentation data that ideally should"},{"line_number":375,"context_line":"be independent from the specific plugins/drivers."}],"source_content_type":"text/x-rst","patch_set":28,"id":"ba8a016a_684b995a","line":372,"in_reply_to":"da85f559_c081f5fc","updated":"2015-11-16 14:51:56.000000000","message":"Done","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":14611,"name":"dasm","display_name":"dasm","email":"dsmigiel@redhat.com","username":"dasm"},"change_message_id":"abf8cb199655705a68013ba08542282063664cf8","unresolved":false,"context_lines":[{"line_number":393,"context_line":"Using memory to hold the data has the shortcoming of creating a point of"},{"line_number":394,"context_line":"discontinuity if the server process is restarted. Storing in memory or"},{"line_number":395,"context_line":"via a disk file share the shortcoming of requiring the aggregation"},{"line_number":396,"context_line":"calculations to be done in python code, which could create scale issues."},{"line_number":397,"context_line":"In addition, maintaining up-to-date information in multi-server deployments"},{"line_number":398,"context_line":"will require that every server consume every message, which would be its"},{"line_number":399,"context_line":"own scalability nightmare. On the other hand, using a table in the"}],"source_content_type":"text/x-rst","patch_set":28,"id":"da85f559_2092414c","line":396,"updated":"2015-11-15 18:05:32.000000000","message":"I\u0027m not sure about this, but probably:\ns/python/Python","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"97a0f84c8573789d4c93d1df67d43d321e0c6c43","unresolved":false,"context_lines":[{"line_number":393,"context_line":"Using memory to hold the data has the shortcoming of creating a point of"},{"line_number":394,"context_line":"discontinuity if the server process is restarted. Storing in memory or"},{"line_number":395,"context_line":"via a disk file share the shortcoming of requiring the aggregation"},{"line_number":396,"context_line":"calculations to be done in python code, which could create scale issues."},{"line_number":397,"context_line":"In addition, maintaining up-to-date information in multi-server deployments"},{"line_number":398,"context_line":"will require that every server consume every message, which would be its"},{"line_number":399,"context_line":"own scalability nightmare. On the other hand, using a table in the"}],"source_content_type":"text/x-rst","patch_set":28,"id":"ba8a016a_e8a46909","line":396,"in_reply_to":"da85f559_2092414c","updated":"2015-11-16 14:51:56.000000000","message":"Done","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":14611,"name":"dasm","display_name":"dasm","email":"dsmigiel@redhat.com","username":"dasm"},"change_message_id":"abf8cb199655705a68013ba08542282063664cf8","unresolved":false,"context_lines":[{"line_number":397,"context_line":"In addition, maintaining up-to-date information in multi-server deployments"},{"line_number":398,"context_line":"will require that every server consume every message, which would be its"},{"line_number":399,"context_line":"own scalability nightmare. On the other hand, using a table in the"},{"line_number":400,"context_line":"neutron database would allow the necessary aggregations to be done in"},{"line_number":401,"context_line":"the database itself, an operation that databases are designed and"},{"line_number":402,"context_line":"optimized for. Therefore the database table option will be persued."},{"line_number":403,"context_line":""}],"source_content_type":"text/x-rst","patch_set":28,"id":"da85f559_60a84978","line":400,"updated":"2015-11-15 18:05:32.000000000","message":"s/neutron/Neutron","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"97a0f84c8573789d4c93d1df67d43d321e0c6c43","unresolved":false,"context_lines":[{"line_number":397,"context_line":"In addition, maintaining up-to-date information in multi-server deployments"},{"line_number":398,"context_line":"will require that every server consume every message, which would be its"},{"line_number":399,"context_line":"own scalability nightmare. On the other hand, using a table in the"},{"line_number":400,"context_line":"neutron database would allow the necessary aggregations to be done in"},{"line_number":401,"context_line":"the database itself, an operation that databases are designed and"},{"line_number":402,"context_line":"optimized for. Therefore the database table option will be persued."},{"line_number":403,"context_line":""}],"source_content_type":"text/x-rst","patch_set":28,"id":"ba8a016a_4879f592","line":400,"in_reply_to":"da85f559_60a84978","updated":"2015-11-16 14:51:56.000000000","message":"Done","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":14611,"name":"dasm","display_name":"dasm","email":"dsmigiel@redhat.com","username":"dasm"},"change_message_id":"abf8cb199655705a68013ba08542282063664cf8","unresolved":false,"context_lines":[{"line_number":399,"context_line":"own scalability nightmare. On the other hand, using a table in the"},{"line_number":400,"context_line":"neutron database would allow the necessary aggregations to be done in"},{"line_number":401,"context_line":"the database itself, an operation that databases are designed and"},{"line_number":402,"context_line":"optimized for. Therefore the database table option will be persued."},{"line_number":403,"context_line":""},{"line_number":404,"context_line":"The basic set of tablular information will consist of a new"},{"line_number":405,"context_line":"if_counter_data tables that will hold counter data from ports/interfaces"}],"source_content_type":"text/x-rst","patch_set":28,"id":"da85f559_20b9e1cb","line":402,"updated":"2015-11-15 18:05:32.000000000","message":"Probably: s/persued/pursued","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"97a0f84c8573789d4c93d1df67d43d321e0c6c43","unresolved":false,"context_lines":[{"line_number":399,"context_line":"own scalability nightmare. On the other hand, using a table in the"},{"line_number":400,"context_line":"neutron database would allow the necessary aggregations to be done in"},{"line_number":401,"context_line":"the database itself, an operation that databases are designed and"},{"line_number":402,"context_line":"optimized for. Therefore the database table option will be persued."},{"line_number":403,"context_line":""},{"line_number":404,"context_line":"The basic set of tablular information will consist of a new"},{"line_number":405,"context_line":"if_counter_data tables that will hold counter data from ports/interfaces"}],"source_content_type":"text/x-rst","patch_set":28,"id":"ba8a016a_a8af41e2","line":402,"in_reply_to":"da85f559_20b9e1cb","updated":"2015-11-16 14:51:56.000000000","message":"Done","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":14611,"name":"dasm","display_name":"dasm","email":"dsmigiel@redhat.com","username":"dasm"},"change_message_id":"abf8cb199655705a68013ba08542282063664cf8","unresolved":false,"context_lines":[{"line_number":401,"context_line":"the database itself, an operation that databases are designed and"},{"line_number":402,"context_line":"optimized for. Therefore the database table option will be persued."},{"line_number":403,"context_line":""},{"line_number":404,"context_line":"The basic set of tablular information will consist of a new"},{"line_number":405,"context_line":"if_counter_data tables that will hold counter data from ports/interfaces"},{"line_number":406,"context_line":"attached to OVS and LinuxBridge instances. This table will have the"},{"line_number":407,"context_line":"following columns:"}],"source_content_type":"text/x-rst","patch_set":28,"id":"da85f559_e0b2f9e4","line":404,"updated":"2015-11-15 18:05:32.000000000","message":"typo: s/tablular/tabular","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"97a0f84c8573789d4c93d1df67d43d321e0c6c43","unresolved":false,"context_lines":[{"line_number":401,"context_line":"the database itself, an operation that databases are designed and"},{"line_number":402,"context_line":"optimized for. Therefore the database table option will be persued."},{"line_number":403,"context_line":""},{"line_number":404,"context_line":"The basic set of tablular information will consist of a new"},{"line_number":405,"context_line":"if_counter_data tables that will hold counter data from ports/interfaces"},{"line_number":406,"context_line":"attached to OVS and LinuxBridge instances. This table will have the"},{"line_number":407,"context_line":"following columns:"}],"source_content_type":"text/x-rst","patch_set":28,"id":"ba8a016a_28827153","line":404,"in_reply_to":"da85f559_e0b2f9e4","updated":"2015-11-16 14:51:56.000000000","message":"Done","commit_id":"c5efbbd6c4370b3659c8a45aca5b41078146a954"},{"author":{"_account_id":7787,"name":"Kevin Benton","email":"kevin@benton.pub","username":"blak111"},"change_message_id":"7a195ef868dc758612d765604687591b5296ad2f","unresolved":false,"context_lines":[{"line_number":5,"context_line":"friendly information about the status and health of the Neutron system."},{"line_number":6,"context_line":"It is **not** about having Neutron act as a bitbucket for storing"},{"line_number":7,"context_line":"historical data, but rather providing snapshot counter data"},{"line_number":8,"context_line":"to some upstream consumer (e.g. Ceilometer). Therefore, adding"},{"line_number":9,"context_line":"instrumentation to Neutron can be broken down into the following"},{"line_number":10,"context_line":"parts:"},{"line_number":11,"context_line":""}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_ce155140","line":8,"updated":"2015-11-19 02:49:05.000000000","message":"I agree that something mapping a given port UUID to an interface name on a specific host is nice, but that\u0027s a little different and could be covered under a different spec.\n\nThis basically gives operators a giant \u0027ifconfig\u0027 command that dumps out counters for the entire system, which is a nice way to get a snapshot of everything and you can feed it into other tools as well.\n\nIf you are using something like Nagios, you can always have it do a poll of this stuff just to watch for interface errors and whatnot. So I would argue that this isn\u0027t requiring building another monitoring system, it\u0027s giving existing monitoring systems an API that they can poll. Most already have intelligence for delta calculations for these types of counters so I think this is a good start.\n\nWRT to the performance of systems like nagios polling Neutron, we can follow up later and have the neutron server that processes a counter update generate an additional event on AMQP so systems can react without polling. It should be easy to add that logic (and even generate an internal registry event so drivers/plugins could take action) once this base stuff is in place.","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"68285e1a2ae1c9ce27108c98459ca0df8ab501ea","unresolved":false,"context_lines":[{"line_number":5,"context_line":"friendly information about the status and health of the Neutron system."},{"line_number":6,"context_line":"It is **not** about having Neutron act as a bitbucket for storing"},{"line_number":7,"context_line":"historical data, but rather providing snapshot counter data"},{"line_number":8,"context_line":"to some upstream consumer (e.g. Ceilometer). Therefore, adding"},{"line_number":9,"context_line":"instrumentation to Neutron can be broken down into the following"},{"line_number":10,"context_line":"parts:"},{"line_number":11,"context_line":""}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_d73de7dd","line":8,"updated":"2015-11-19 01:15:02.000000000","message":"I think that is very important to understand the type of consumer of this instrumentation effort and I am not sure Ceilometer should be representative.\n\nMoreover, as an operator I may want to use something like Nagios to monitor my infrastructure so that I can get alerts etc. I am not 100% sure this proposal would be that useful to the operator in this form and how it relates to other monitoring tools: I can\u0027t do anything with it, besides polling every X seconds and do post-processing by building yet another monitoring system around it.\n\nIs it really what we are aiming for here? I think the value provided in the long run is minimal, unless you\u0027re planning more that you\u0027re not telling :)\n\nAs a regular tenant I already have my nova diagnostic tool here, I am happy, and I need nothing else. As an operator however, if I used something like Nagios, what\u0027s missing is the ability to map my physical infrastructure performance counters to OpenStack objects, like routers, floating ips, etc.\n\nI\u0027d argue that the value is not so much in exposing an API for consumption of performance counters...but in an API that allows me to map the elements of the physical topology (the various qr, tap, qb, qvo interfaces that live in the various namespaces) to the element of the Neutron logical topology (routers, dhcp servers, etc).","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":7787,"name":"Kevin Benton","email":"kevin@benton.pub","username":"blak111"},"change_message_id":"ec359da7dbaf6a0cbbd7aeaf33d53387fbb9e1d1","unresolved":false,"context_lines":[{"line_number":5,"context_line":"friendly information about the status and health of the Neutron system."},{"line_number":6,"context_line":"It is **not** about having Neutron act as a bitbucket for storing"},{"line_number":7,"context_line":"historical data, but rather providing snapshot counter data"},{"line_number":8,"context_line":"to some upstream consumer (e.g. Ceilometer). Therefore, adding"},{"line_number":9,"context_line":"instrumentation to Neutron can be broken down into the following"},{"line_number":10,"context_line":"parts:"},{"line_number":11,"context_line":""}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_5eac768b","line":8,"updated":"2015-11-19 22:38:42.000000000","message":"Operators do have access to this data already via other means, but this is associating counters with the actual Neutron objects. So an operator might already see that some interface is showing errors, but they currently don\u0027t know which tenant that corresponds to or if it belongs to a certain router, etc without a lot of legwork on their own.\n\nI suspect that\u0027s why you were thinking of just offering the translation of tap devices to Neutron objects. Even with just a translation, that still leaves a lot of work for the operator to figure out how to have their monitoring tool do the correct lookups from a service. Maybe I\u0027m overestimating the difficulty in getting that correct.","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"e6a5a290eb76d9cddbf1c41bb3807c38d6bfe750","unresolved":false,"context_lines":[{"line_number":5,"context_line":"friendly information about the status and health of the Neutron system."},{"line_number":6,"context_line":"It is **not** about having Neutron act as a bitbucket for storing"},{"line_number":7,"context_line":"historical data, but rather providing snapshot counter data"},{"line_number":8,"context_line":"to some upstream consumer (e.g. Ceilometer). Therefore, adding"},{"line_number":9,"context_line":"instrumentation to Neutron can be broken down into the following"},{"line_number":10,"context_line":"parts:"},{"line_number":11,"context_line":""}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_a0421bac","line":8,"in_reply_to":"ba8a016a_5eac768b","updated":"2015-11-20 01:11:39.000000000","message":"There are many use cases out there that are related to monitoring an openstack deployment. Some use Monasca, some use their own, etc.\n\nI am wondering if the proposed approach is the simplest and most effective mechanism that address the widest range of use cases. On the other end, I wonder if we should just pick one and do it well. I see overlap between what is proposed here and what other monitoring tools try to do.\n\nIt would be useful if some Monasca/Ceilometer folk could chime in here. My gut feeling is telling me that we\u0027re setting ourselves for reinventing a monitoring system in Neutron, and that\u0027s the last thing I think we should do.\n\nIf I am mistaken, I\u0027d love to see this clarified, and how this is intended to be used by another monitoring tool (aka integration strategies).","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"3cd485365e0ff6514e9f52b9f674abc176330c19","unresolved":false,"context_lines":[{"line_number":5,"context_line":"friendly information about the status and health of the Neutron system."},{"line_number":6,"context_line":"It is **not** about having Neutron act as a bitbucket for storing"},{"line_number":7,"context_line":"historical data, but rather providing snapshot counter data"},{"line_number":8,"context_line":"to some upstream consumer (e.g. Ceilometer). Therefore, adding"},{"line_number":9,"context_line":"instrumentation to Neutron can be broken down into the following"},{"line_number":10,"context_line":"parts:"},{"line_number":11,"context_line":""}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_7102390b","line":8,"in_reply_to":"ba8a016a_82725c48","updated":"2015-11-24 05:38:42.000000000","message":"After talking to Armando and Kyle some more, I\u0027m going to split this devref into three sequential patch sets.  The first will address the data model and collection and reporting, which will allow a collection proof of concept to explore potential scaling issues with this part of the solution.\n\nThe second devref patch will add data storage at the neutron server, which will support creating the necessary tables for caching collected data for aggregation.\n\nThe third devref patch will add data consumption and present guidelines for how to implement plug-ins for reporting aggregated data to \"your favorite monitoring system\" - what the reference code will plug into has yet to be determined.","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"7f3bf6829b6bf37ee105e1b8464f53047cc6d741","unresolved":false,"context_lines":[{"line_number":5,"context_line":"friendly information about the status and health of the Neutron system."},{"line_number":6,"context_line":"It is **not** about having Neutron act as a bitbucket for storing"},{"line_number":7,"context_line":"historical data, but rather providing snapshot counter data"},{"line_number":8,"context_line":"to some upstream consumer (e.g. Ceilometer). Therefore, adding"},{"line_number":9,"context_line":"instrumentation to Neutron can be broken down into the following"},{"line_number":10,"context_line":"parts:"},{"line_number":11,"context_line":""}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_82725c48","line":8,"in_reply_to":"ba8a016a_a0421bac","updated":"2015-11-20 15:08:08.000000000","message":"Armando, let me set some fears to rest - First and foremost, I\u0027m definitely *NOT* interested in re-inventing a monitoring system,\n\nIn my view, (and I\u0027ll likely need to re-write the data consumption portion to make this more clear), this is all about:\n\n1. Collecting the data (i.e. southbound from neutron server to implementation pieces),\n2. Aggregating that data based on what Neutron knows (as Kevin said so well above), and \n3. Setting up the data consumption framework so that people can write northbound drivers to plug this into *pick your favorite monitoring framework*.  The REST implementation was chosen only as a reference implementation.\n\nIf I\u0027ve not done those three things clearly, then I definitely need to re-work this","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"bb52e813975ea1ecde69afe173b4e435b0bb5422","unresolved":false,"context_lines":[{"line_number":5,"context_line":"friendly information about the status and health of the Neutron system."},{"line_number":6,"context_line":"It is **not** about having Neutron act as a bitbucket for storing"},{"line_number":7,"context_line":"historical data, but rather providing snapshot counter data"},{"line_number":8,"context_line":"to some upstream consumer (e.g. Ceilometer). Therefore, adding"},{"line_number":9,"context_line":"instrumentation to Neutron can be broken down into the following"},{"line_number":10,"context_line":"parts:"},{"line_number":11,"context_line":""}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_34122a0a","line":8,"in_reply_to":"ba8a016a_ce155140","updated":"2015-11-19 05:42:15.000000000","message":"I don\u0027t disagree that the giant ifconfig is handy for human consumption and can still be used to feed data to an external tool, but isn\u0027t this prone to misuse? Polling a REST interface for fetching highly volatile data would expose us to a usage pattern where the server/s will be pounded to keep the data current. Furthermore the data being collected will have to be kept in sync with the actual state of the system upon life cycle events of neutron resources. Is this really the behavior that we want to enable/promote? I am not 100% sure.\n\nAs an operator, would I want to integrate my monitoring tools like this? As a user, it\u0027s nice to have access to my workload performance counters, but, as an operator, I have access to my infrastructure and have full access to the data sources that are on the compute nodes and/or edge nodes (l3, dhcp, etc).\n\nStrictly speaking, if aggregation is not required to build data series (the proposal aims at keeping only the last snapshot), why would I choose to aggregate data and use a REST presentation layer?\n\nThese are the kinds of questions, I would like to see addressed before we push the button on this one.","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"bb52e813975ea1ecde69afe173b4e435b0bb5422","unresolved":false,"context_lines":[{"line_number":31,"context_line":"   already support this (even incompletely) it is considered to be within"},{"line_number":32,"context_line":"   the scope of the upstream consumer and not Neutron itself."},{"line_number":33,"context_line":""},{"line_number":34,"context_line":"How does Instrumentation differ from Metering Labels and Rules"},{"line_number":35,"context_line":"--------------------------------------------------------------"},{"line_number":36,"context_line":""},{"line_number":37,"context_line":"The existing metering label and rule extension provides the ability to"}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_178048f6","line":34,"updated":"2015-11-19 05:42:15.000000000","message":"Is it worth spending a few words on how this compare/differs from Monasca?","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"3cd485365e0ff6514e9f52b9f674abc176330c19","unresolved":false,"context_lines":[{"line_number":31,"context_line":"   already support this (even incompletely) it is considered to be within"},{"line_number":32,"context_line":"   the scope of the upstream consumer and not Neutron itself."},{"line_number":33,"context_line":""},{"line_number":34,"context_line":"How does Instrumentation differ from Metering Labels and Rules"},{"line_number":35,"context_line":"--------------------------------------------------------------"},{"line_number":36,"context_line":""},{"line_number":37,"context_line":"The existing metering label and rule extension provides the ability to"}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_11208d9a","line":34,"in_reply_to":"ba8a016a_178048f6","updated":"2015-11-24 05:38:42.000000000","message":"I view this as plugging into things like Celiometer or Monasca.  Since I\u0027m breaking the devref apart, let\u0027s revisit this on the next draft.","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":11255,"name":"Dustin Lundquist","username":"dustin.lundquist","inactive":true},"change_message_id":"ae15be991377b6e81fdfc673728ef3c7d4cbb93a","unresolved":false,"context_lines":[{"line_number":128,"context_line":""},{"line_number":129,"context_line":"Because of how classic and distributed routers are implemented, it is"},{"line_number":130,"context_line":"necessary to consider how information from the OVS gets mapped into the"},{"line_number":131,"context_line":"defined data structures"},{"line_number":132,"context_line":""},{"line_number":133,"context_line":"1. VIF and port statistics to switch interfaces information"},{"line_number":134,"context_line":"2. Host Switch information to switch interface information for"}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_f0591c36","line":131,"updated":"2015-11-18 21:50:04.000000000","message":"This is difficult to follow. Maybe separate the list heading from the explanation of its significance.","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"3cd485365e0ff6514e9f52b9f674abc176330c19","unresolved":false,"context_lines":[{"line_number":128,"context_line":""},{"line_number":129,"context_line":"Because of how classic and distributed routers are implemented, it is"},{"line_number":130,"context_line":"necessary to consider how information from the OVS gets mapped into the"},{"line_number":131,"context_line":"defined data structures"},{"line_number":132,"context_line":""},{"line_number":133,"context_line":"1. VIF and port statistics to switch interfaces information"},{"line_number":134,"context_line":"2. Host Switch information to switch interface information for"}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_514db570","line":131,"in_reply_to":"ba8a016a_f0591c36","updated":"2015-11-24 05:38:42.000000000","message":"This section is being re-written to be more clear","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":11255,"name":"Dustin Lundquist","username":"dustin.lundquist","inactive":true},"change_message_id":"ae15be991377b6e81fdfc673728ef3c7d4cbb93a","unresolved":false,"context_lines":[{"line_number":354,"context_line":"of dealing with the stale data issues that scaling a polling cycle will"},{"line_number":355,"context_line":"entail. On the other side, asynchronous notification requires that the"},{"line_number":356,"context_line":"aggregator has the capability to hold the current results from each"},{"line_number":357,"context_line":"harvester. Neutron agents already have a thread that provides status"},{"line_number":358,"context_line":"information back to the Neutron server, so extending this thread to"},{"line_number":359,"context_line":"include instrumentation data would be a smaller change than implementing"},{"line_number":360,"context_line":"a whole separate polling cycle just for collection instrumentation data."}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_162e1843","line":357,"updated":"2015-11-18 21:50:04.000000000","message":"You might be more explicit in adopting the asynchronous model.\n\nAdditionally, you call out issues with aggregating data during using the polling method, the response can not be returned until all agents respond or time out.","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"3cd485365e0ff6514e9f52b9f674abc176330c19","unresolved":false,"context_lines":[{"line_number":354,"context_line":"of dealing with the stale data issues that scaling a polling cycle will"},{"line_number":355,"context_line":"entail. On the other side, asynchronous notification requires that the"},{"line_number":356,"context_line":"aggregator has the capability to hold the current results from each"},{"line_number":357,"context_line":"harvester. Neutron agents already have a thread that provides status"},{"line_number":358,"context_line":"information back to the Neutron server, so extending this thread to"},{"line_number":359,"context_line":"include instrumentation data would be a smaller change than implementing"},{"line_number":360,"context_line":"a whole separate polling cycle just for collection instrumentation data."}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_11572d03","line":357,"in_reply_to":"ba8a016a_162e1843","updated":"2015-11-24 05:38:42.000000000","message":"Done","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":11255,"name":"Dustin Lundquist","username":"dustin.lundquist","inactive":true},"change_message_id":"ae15be991377b6e81fdfc673728ef3c7d4cbb93a","unresolved":false,"context_lines":[{"line_number":419,"context_line":"* \"xmt_byte_count\" which holds 64 bits of port transmit byte count"},{"line_number":420,"context_line":"* \"xmt_pckt_count\" which holds 64 bits of port transmit packet count"},{"line_number":421,"context_line":"* \"xmt_drop_count\" which holds 32 bits of port transmit dropped count"},{"line_number":422,"context_line":"* \"xmt_err_count\" which holds 32 bits of port transmit error count"},{"line_number":423,"context_line":""},{"line_number":424,"context_line":"HTTP API and Resources"},{"line_number":425,"context_line":"~~~~~~~~~~~~~~~~~~~~~~"}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_b6cdcc86","line":422,"updated":"2015-11-18 21:50:04.000000000","message":"Include an update_at time stamp for diagnosing stale agents and excluding stale data? Monitoring systems differentiate between an unchanged counter, and lack of data for an polling interval.","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"3cd485365e0ff6514e9f52b9f674abc176330c19","unresolved":false,"context_lines":[{"line_number":419,"context_line":"* \"xmt_byte_count\" which holds 64 bits of port transmit byte count"},{"line_number":420,"context_line":"* \"xmt_pckt_count\" which holds 64 bits of port transmit packet count"},{"line_number":421,"context_line":"* \"xmt_drop_count\" which holds 32 bits of port transmit dropped count"},{"line_number":422,"context_line":"* \"xmt_err_count\" which holds 32 bits of port transmit error count"},{"line_number":423,"context_line":""},{"line_number":424,"context_line":"HTTP API and Resources"},{"line_number":425,"context_line":"~~~~~~~~~~~~~~~~~~~~~~"}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_914b1d59","line":422,"in_reply_to":"ba8a016a_b6cdcc86","updated":"2015-11-24 05:38:42.000000000","message":"I will keep that in mind for the follow on patch set that will cover this.","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"68285e1a2ae1c9ce27108c98459ca0df8ab501ea","unresolved":false,"context_lines":[{"line_number":438,"context_line":"    }"},{"line_number":439,"context_line":""},{"line_number":440,"context_line":"A GET operation against the element"},{"line_number":441,"context_line":"resource \"/v2.0/network-counters/\u003cnetwork UUID\u003e\" will return the"},{"line_number":442,"context_line":"following counter structure for the network in question:"},{"line_number":443,"context_line":""},{"line_number":444,"context_line":".."}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_176a8fee","line":441,"updated":"2015-11-19 01:15:02.000000000","message":"I would be concerned if a consumer of this API will start periodically polling in order to build a trend of any of these counters.","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":7787,"name":"Kevin Benton","email":"kevin@benton.pub","username":"blak111"},"change_message_id":"ec359da7dbaf6a0cbbd7aeaf33d53387fbb9e1d1","unresolved":false,"context_lines":[{"line_number":438,"context_line":"    }"},{"line_number":439,"context_line":""},{"line_number":440,"context_line":"A GET operation against the element"},{"line_number":441,"context_line":"resource \"/v2.0/network-counters/\u003cnetwork UUID\u003e\" will return the"},{"line_number":442,"context_line":"following counter structure for the network in question:"},{"line_number":443,"context_line":""},{"line_number":444,"context_line":".."}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_7ea7faa6","line":441,"updated":"2015-11-19 22:38:42.000000000","message":"Since this is an admin-only API, isn\u0027t that sort of the operator\u0027s problem? Or are you concerned that operators will get upset with us when they find out frequent polling slows down the server?","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"3cd485365e0ff6514e9f52b9f674abc176330c19","unresolved":false,"context_lines":[{"line_number":438,"context_line":"    }"},{"line_number":439,"context_line":""},{"line_number":440,"context_line":"A GET operation against the element"},{"line_number":441,"context_line":"resource \"/v2.0/network-counters/\u003cnetwork UUID\u003e\" will return the"},{"line_number":442,"context_line":"following counter structure for the network in question:"},{"line_number":443,"context_line":""},{"line_number":444,"context_line":".."}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_91b9dd30","line":441,"in_reply_to":"ba8a016a_000e878b","updated":"2015-11-24 05:38:42.000000000","message":"I\u0027ll keep this in mind for the following on patch set that talks about data consumption","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"e6a5a290eb76d9cddbf1c41bb3807c38d6bfe750","unresolved":false,"context_lines":[{"line_number":438,"context_line":"    }"},{"line_number":439,"context_line":""},{"line_number":440,"context_line":"A GET operation against the element"},{"line_number":441,"context_line":"resource \"/v2.0/network-counters/\u003cnetwork UUID\u003e\" will return the"},{"line_number":442,"context_line":"following counter structure for the network in question:"},{"line_number":443,"context_line":""},{"line_number":444,"context_line":".."}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_000e878b","line":441,"in_reply_to":"ba8a016a_7ea7faa6","updated":"2015-11-20 01:11:39.000000000","message":"History taught us that the operator\u0027s problem is ours, is it not? :)","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":7787,"name":"Kevin Benton","email":"kevin@benton.pub","username":"blak111"},"change_message_id":"7a195ef868dc758612d765604687591b5296ad2f","unresolved":false,"context_lines":[{"line_number":445,"context_line":"    { \"network-counter\": {"},{"line_number":446,"context_line":"        \"uuid\": \u003cnetwork uuid\u003e,"},{"line_number":447,"context_line":"        \"name\": \u003cnetwork name\u003e,"},{"line_number":448,"context_line":"        \"interface counters\": ["},{"line_number":449,"context_line":"             { \"uuid\": \u003cinterface uuid\u003e,"},{"line_number":450,"context_line":"               \"name\": \u003cinterface name\u003e,"},{"line_number":451,"context_line":"               \"ifInOctets\": \u003cvalue\u003e,"}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_ee129556","line":448,"updated":"2015-11-19 02:49:05.000000000","message":"+1. I would rather have the counters living underneath \u0027network\u0027 and \u0027router\u0027 just be the counters specific to that entire object.\n\nFor individual interface counters, can we just have a separate /v2.0/interface-counters that takes a big list of UUIDs you are interested in and returns the interface counters for those?","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"68285e1a2ae1c9ce27108c98459ca0df8ab501ea","unresolved":false,"context_lines":[{"line_number":445,"context_line":"    { \"network-counter\": {"},{"line_number":446,"context_line":"        \"uuid\": \u003cnetwork uuid\u003e,"},{"line_number":447,"context_line":"        \"name\": \u003cnetwork name\u003e,"},{"line_number":448,"context_line":"        \"interface counters\": ["},{"line_number":449,"context_line":"             { \"uuid\": \u003cinterface uuid\u003e,"},{"line_number":450,"context_line":"               \"name\": \u003cinterface name\u003e,"},{"line_number":451,"context_line":"               \"ifInOctets\": \u003cvalue\u003e,"}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_d743a711","line":448,"updated":"2015-11-19 01:15:02.000000000","message":"this can be potentially a very long list of ports, what if I am interested in knowing the performance counter of a single port?","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"bb52e813975ea1ecde69afe173b4e435b0bb5422","unresolved":false,"context_lines":[{"line_number":445,"context_line":"    { \"network-counter\": {"},{"line_number":446,"context_line":"        \"uuid\": \u003cnetwork uuid\u003e,"},{"line_number":447,"context_line":"        \"name\": \u003cnetwork name\u003e,"},{"line_number":448,"context_line":"        \"interface counters\": ["},{"line_number":449,"context_line":"             { \"uuid\": \u003cinterface uuid\u003e,"},{"line_number":450,"context_line":"               \"name\": \u003cinterface name\u003e,"},{"line_number":451,"context_line":"               \"ifInOctets\": \u003cvalue\u003e,"}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_f176f4d5","line":448,"in_reply_to":"ba8a016a_ee129556","updated":"2015-11-19 05:42:15.000000000","message":"either way...I\u0027d rather not have to find a needle in a haystack :)","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"3cd485365e0ff6514e9f52b9f674abc176330c19","unresolved":false,"context_lines":[{"line_number":445,"context_line":"    { \"network-counter\": {"},{"line_number":446,"context_line":"        \"uuid\": \u003cnetwork uuid\u003e,"},{"line_number":447,"context_line":"        \"name\": \u003cnetwork name\u003e,"},{"line_number":448,"context_line":"        \"interface counters\": ["},{"line_number":449,"context_line":"             { \"uuid\": \u003cinterface uuid\u003e,"},{"line_number":450,"context_line":"               \"name\": \u003cinterface name\u003e,"},{"line_number":451,"context_line":"               \"ifInOctets\": \u003cvalue\u003e,"}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_317fd167","line":448,"in_reply_to":"ba8a016a_f176f4d5","updated":"2015-11-24 05:38:42.000000000","message":"Ditto my L441 comment","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"68285e1a2ae1c9ce27108c98459ca0df8ab501ea","unresolved":false,"context_lines":[{"line_number":496,"context_line":"               \"ifHCInUcastPkts\": \u003cvalue\u003e,"},{"line_number":497,"context_line":"               \"ifHCOctOctets\": \u003cvalue\u003e,"},{"line_number":498,"context_line":"               \"ifHCOctUcastPkts\": \u003cvalue\u003e } ],"},{"line_number":499,"context_line":"        \"router counters\": {"},{"line_number":500,"context_line":"               \"ipSystemStatsInReceives\": \u003cvalue\u003e,"},{"line_number":501,"context_line":"               \"ipSystemStatsHCInReceives\": \u003cvalue\u003e,"},{"line_number":502,"context_line":"               \"ipSystemStatsInOctets\": \u003cvalue\u003e,"}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_d75ac7f1","line":499,"updated":"2015-11-19 01:15:02.000000000","message":"why can\u0027t these simply be at the same level of uuid and name?","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"3cd485365e0ff6514e9f52b9f674abc176330c19","unresolved":false,"context_lines":[{"line_number":496,"context_line":"               \"ifHCInUcastPkts\": \u003cvalue\u003e,"},{"line_number":497,"context_line":"               \"ifHCOctOctets\": \u003cvalue\u003e,"},{"line_number":498,"context_line":"               \"ifHCOctUcastPkts\": \u003cvalue\u003e } ],"},{"line_number":499,"context_line":"        \"router counters\": {"},{"line_number":500,"context_line":"               \"ipSystemStatsInReceives\": \u003cvalue\u003e,"},{"line_number":501,"context_line":"               \"ipSystemStatsHCInReceives\": \u003cvalue\u003e,"},{"line_number":502,"context_line":"               \"ipSystemStatsInOctets\": \u003cvalue\u003e,"}],"source_content_type":"text/x-rst","patch_set":29,"id":"ba8a016a_f191499b","line":499,"in_reply_to":"ba8a016a_d75ac7f1","updated":"2015-11-24 05:38:42.000000000","message":"Ditto my L441 comment","commit_id":"2dd15231deb6c8261fd6cd12cb29313ba78fa863"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"5337d410a238144921781c8cb487e9b25517ecb4","unresolved":false,"context_lines":[{"line_number":13,"context_line":"#. Data Collection (i.e. what data should be collected and how),"},{"line_number":14,"context_line":"#. Data Aggregation (i.e. how and where raw data should be aggregated"},{"line_number":15,"context_line":"   into project information)"},{"line_number":16,"context_line":"#. Data Consumption (i.e. how is aggregated data consumed)"},{"line_number":17,"context_line":""},{"line_number":18,"context_line":"While instrumentation might also be considered to include asynchronous event"},{"line_number":19,"context_line":"notifications, like fault detection, this is considered out of scope"}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_8f822c51","line":16,"updated":"2015-12-02 01:26:38.000000000","message":"I ain\u0027t English native speaker, but shouldn\u0027t this be:\n\nhow aggregated data is consumed? \n\nSorry for being pedantic, I am nostalgic of Salvatore\u0027s reviews.","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"60d12dd2b38885214248bccaf7d984e9a2c5d7d4","unresolved":false,"context_lines":[{"line_number":13,"context_line":"#. Data Collection (i.e. what data should be collected and how),"},{"line_number":14,"context_line":"#. Data Aggregation (i.e. how and where raw data should be aggregated"},{"line_number":15,"context_line":"   into project information)"},{"line_number":16,"context_line":"#. Data Consumption (i.e. how is aggregated data consumed)"},{"line_number":17,"context_line":""},{"line_number":18,"context_line":"While instrumentation might also be considered to include asynchronous event"},{"line_number":19,"context_line":"notifications, like fault detection, this is considered out of scope"}],"source_content_type":"text/x-rst","patch_set":32,"id":"7a740942_0d2bf431","line":16,"in_reply_to":"9a8ffd7b_65b10001","updated":"2015-12-04 00:30:14.000000000","message":"Um, I have mixed feelings..I shouldn\u0027t trust an American English native speaker, but ok :)","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"1bc336f1f0be538f3eae351caebac158e476d68e","unresolved":false,"context_lines":[{"line_number":13,"context_line":"#. Data Collection (i.e. what data should be collected and how),"},{"line_number":14,"context_line":"#. Data Aggregation (i.e. how and where raw data should be aggregated"},{"line_number":15,"context_line":"   into project information)"},{"line_number":16,"context_line":"#. Data Consumption (i.e. how is aggregated data consumed)"},{"line_number":17,"context_line":""},{"line_number":18,"context_line":"While instrumentation might also be considered to include asynchronous event"},{"line_number":19,"context_line":"notifications, like fault detection, this is considered out of scope"}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_65b10001","line":16,"in_reply_to":"9a8ffd7b_8f822c51","updated":"2015-12-02 18:27:19.000000000","message":"It can go either way, so I\u0027m leaving it this way :)","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"5337d410a238144921781c8cb487e9b25517ecb4","unresolved":false,"context_lines":[{"line_number":38,"context_line":""},{"line_number":39,"context_line":"The existing metering label and rule extension provides the ability to"},{"line_number":40,"context_line":"collect traffic information on a per CIDR basis, with the reference"},{"line_number":41,"context_line":"implementation implementing this function via iptables rules."},{"line_number":42,"context_line":""},{"line_number":43,"context_line":"A possible implementation of instrumentation would be to use per-instance"},{"line_number":44,"context_line":"metering rules for all IP addresses in both directions. This suffers two"}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_6f5b48f7","line":41,"updated":"2015-12-02 01:26:38.000000000","message":"any chance you could provide a pointer to the actual code, to help the reader visualize what you mean?","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"d9729c3ae36684abe6504a5db857c35f7d4f63bf","unresolved":false,"context_lines":[{"line_number":38,"context_line":""},{"line_number":39,"context_line":"The existing metering label and rule extension provides the ability to"},{"line_number":40,"context_line":"collect traffic information on a per CIDR basis, with the reference"},{"line_number":41,"context_line":"implementation implementing this function via iptables rules."},{"line_number":42,"context_line":""},{"line_number":43,"context_line":"A possible implementation of instrumentation would be to use per-instance"},{"line_number":44,"context_line":"metering rules for all IP addresses in both directions. This suffers two"}],"source_content_type":"text/x-rst","patch_set":32,"id":"7a740942_bd30b657","line":41,"in_reply_to":"7a740942_6d1ab819","updated":"2015-12-04 13:32:33.000000000","message":"gleep - well, I\u0027m red faced.\n\nSince the PoC is going to have an update to this devref anyway (covering the implementation details) I\u0027ll go back and update this section then.","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"60d12dd2b38885214248bccaf7d984e9a2c5d7d4","unresolved":false,"context_lines":[{"line_number":38,"context_line":""},{"line_number":39,"context_line":"The existing metering label and rule extension provides the ability to"},{"line_number":40,"context_line":"collect traffic information on a per CIDR basis, with the reference"},{"line_number":41,"context_line":"implementation implementing this function via iptables rules."},{"line_number":42,"context_line":""},{"line_number":43,"context_line":"A possible implementation of instrumentation would be to use per-instance"},{"line_number":44,"context_line":"metering rules for all IP addresses in both directions. This suffers two"}],"source_content_type":"text/x-rst","patch_set":32,"id":"7a740942_6d1ab819","line":41,"in_reply_to":"9a8ffd7b_05711444","updated":"2015-12-04 00:30:14.000000000","message":"nothing?","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"1bc336f1f0be538f3eae351caebac158e476d68e","unresolved":false,"context_lines":[{"line_number":38,"context_line":""},{"line_number":39,"context_line":"The existing metering label and rule extension provides the ability to"},{"line_number":40,"context_line":"collect traffic information on a per CIDR basis, with the reference"},{"line_number":41,"context_line":"implementation implementing this function via iptables rules."},{"line_number":42,"context_line":""},{"line_number":43,"context_line":"A possible implementation of instrumentation would be to use per-instance"},{"line_number":44,"context_line":"metering rules for all IP addresses in both directions. This suffers two"}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_05711444","line":41,"in_reply_to":"9a8ffd7b_6f5b48f7","updated":"2015-12-02 18:27:19.000000000","message":"I\u0027ll see what I can find","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"5337d410a238144921781c8cb487e9b25517ecb4","unresolved":false,"context_lines":[{"line_number":44,"context_line":"metering rules for all IP addresses in both directions. This suffers two"},{"line_number":45,"context_line":"issues:"},{"line_number":46,"context_line":""},{"line_number":47,"context_line":"#. The scalability of such an approach is, at best, questionable."},{"line_number":48,"context_line":"#. The information collected by metering rules is focused more on billing"},{"line_number":49,"context_line":"   and so does not have the desired granularity (i.e. it counts transmitted"},{"line_number":50,"context_line":"   packets without keeping track of what caused packets to fail)"}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_2f1c6012","line":47,"updated":"2015-12-02 01:26:38.000000000","message":"I might sound obvious but I tend to refrain from making these statements without giving proper justification...things can bite back you know? So, consider elaborating why you think it\u0027s questionable. In any case against you, it puts you in a stronger position ;)","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"1bc336f1f0be538f3eae351caebac158e476d68e","unresolved":false,"context_lines":[{"line_number":44,"context_line":"metering rules for all IP addresses in both directions. This suffers two"},{"line_number":45,"context_line":"issues:"},{"line_number":46,"context_line":""},{"line_number":47,"context_line":"#. The scalability of such an approach is, at best, questionable."},{"line_number":48,"context_line":"#. The information collected by metering rules is focused more on billing"},{"line_number":49,"context_line":"   and so does not have the desired granularity (i.e. it counts transmitted"},{"line_number":50,"context_line":"   packets without keeping track of what caused packets to fail)"}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_a5c84823","line":47,"in_reply_to":"9a8ffd7b_2f1c6012","updated":"2015-12-02 18:27:19.000000000","message":"actually, I\u0027ll drop this in the next respin","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"5337d410a238144921781c8cb487e9b25517ecb4","unresolved":false,"context_lines":[{"line_number":80,"context_line":"and get_instance_diagnostics() calls:"},{"line_number":81,"context_line":""},{"line_number":82,"context_line":"* Receive bytes, packets, errors and drops"},{"line_number":83,"context_line":"* Transmit bytes, packets, errors and drops"},{"line_number":84,"context_line":""},{"line_number":85,"context_line":"These are available via the CLI command \"nova diagnostics"},{"line_number":86,"context_line":"\u003cserver\u003e\" where server is a running instance. In addition,"}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_6fb7c8f5","line":83,"updated":"2015-12-02 01:26:38.000000000","message":"these are per-nic, right?","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"1bc336f1f0be538f3eae351caebac158e476d68e","unresolved":false,"context_lines":[{"line_number":80,"context_line":"and get_instance_diagnostics() calls:"},{"line_number":81,"context_line":""},{"line_number":82,"context_line":"* Receive bytes, packets, errors and drops"},{"line_number":83,"context_line":"* Transmit bytes, packets, errors and drops"},{"line_number":84,"context_line":""},{"line_number":85,"context_line":"These are available via the CLI command \"nova diagnostics"},{"line_number":86,"context_line":"\u003cserver\u003e\" where server is a running instance. In addition,"}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_856f44fb","line":83,"in_reply_to":"9a8ffd7b_6fb7c8f5","updated":"2015-12-02 18:27:19.000000000","message":"yes, I\u0027ll say that","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"5337d410a238144921781c8cb487e9b25517ecb4","unresolved":false,"context_lines":[{"line_number":84,"context_line":""},{"line_number":85,"context_line":"These are available via the CLI command \"nova diagnostics"},{"line_number":86,"context_line":"\u003cserver\u003e\" where server is a running instance. In addition,"},{"line_number":87,"context_line":"there is already an effort underway to have Ceilometer call"},{"line_number":88,"context_line":"into these directly [#]_, so collecting these statistics will"},{"line_number":89,"context_line":"be considered out of scope of this devref."},{"line_number":90,"context_line":""}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_8fb6acda","line":87,"updated":"2015-12-02 01:26:38.000000000","message":"what\u0027s the latest on this? The wiki last update goes back to 2013 and the ML thread is a year old. Do you know? Can we perhaps share a newer link?","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"d9729c3ae36684abe6504a5db857c35f7d4f63bf","unresolved":false,"context_lines":[{"line_number":84,"context_line":""},{"line_number":85,"context_line":"These are available via the CLI command \"nova diagnostics"},{"line_number":86,"context_line":"\u003cserver\u003e\" where server is a running instance. In addition,"},{"line_number":87,"context_line":"there is already an effort underway to have Ceilometer call"},{"line_number":88,"context_line":"into these directly [#]_, so collecting these statistics will"},{"line_number":89,"context_line":"be considered out of scope of this devref."},{"line_number":90,"context_line":""}],"source_content_type":"text/x-rst","patch_set":32,"id":"7a740942_5d83d2cc","line":87,"in_reply_to":"7a740942_2dba70e2","updated":"2015-12-04 13:32:33.000000000","message":"...very...","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"60d12dd2b38885214248bccaf7d984e9a2c5d7d4","unresolved":false,"context_lines":[{"line_number":84,"context_line":""},{"line_number":85,"context_line":"These are available via the CLI command \"nova diagnostics"},{"line_number":86,"context_line":"\u003cserver\u003e\" where server is a running instance. In addition,"},{"line_number":87,"context_line":"there is already an effort underway to have Ceilometer call"},{"line_number":88,"context_line":"into these directly [#]_, so collecting these statistics will"},{"line_number":89,"context_line":"be considered out of scope of this devref."},{"line_number":90,"context_line":""}],"source_content_type":"text/x-rst","patch_set":32,"id":"7a740942_2dba70e2","line":87,"in_reply_to":"9a8ffd7b_6606e61d","updated":"2015-12-04 00:30:14.000000000","message":"that\u0027s disappointing.","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"1bc336f1f0be538f3eae351caebac158e476d68e","unresolved":false,"context_lines":[{"line_number":84,"context_line":""},{"line_number":85,"context_line":"These are available via the CLI command \"nova diagnostics"},{"line_number":86,"context_line":"\u003cserver\u003e\" where server is a running instance. In addition,"},{"line_number":87,"context_line":"there is already an effort underway to have Ceilometer call"},{"line_number":88,"context_line":"into these directly [#]_, so collecting these statistics will"},{"line_number":89,"context_line":"be considered out of scope of this devref."},{"line_number":90,"context_line":""}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_6606e61d","line":87,"in_reply_to":"9a8ffd7b_8fb6acda","updated":"2015-12-02 18:27:19.000000000","message":"I\u0027ll have to rewrite this as http://lists.openstack.org/pipermail/openstack-dev/2015-June/067589.html indicates this is still an open question","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"5337d410a238144921781c8cb487e9b25517ecb4","unresolved":false,"context_lines":[{"line_number":121,"context_line":""},{"line_number":122,"context_line":"  * DHCP namespace tap interface (if defined)"},{"line_number":123,"context_line":"  * Router namespace qr interface"},{"line_number":124,"context_line":"  * Router namespace qg interface"},{"line_number":125,"context_line":""},{"line_number":126,"context_line":"What is Available from Openvswitch"},{"line_number":127,"context_line":"~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~"}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_2f3b003a","line":124,"updated":"2015-12-02 01:26:38.000000000","message":"might be worth noting here that at the time of writing LB only support centralized routing.","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"1bc336f1f0be538f3eae351caebac158e476d68e","unresolved":false,"context_lines":[{"line_number":121,"context_line":""},{"line_number":122,"context_line":"  * DHCP namespace tap interface (if defined)"},{"line_number":123,"context_line":"  * Router namespace qr interface"},{"line_number":124,"context_line":"  * Router namespace qg interface"},{"line_number":125,"context_line":""},{"line_number":126,"context_line":"What is Available from Openvswitch"},{"line_number":127,"context_line":"~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~"}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_26153e3f","line":124,"in_reply_to":"9a8ffd7b_2f3b003a","updated":"2015-12-02 18:27:19.000000000","message":"sure","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"5337d410a238144921781c8cb487e9b25517ecb4","unresolved":false,"context_lines":[{"line_number":200,"context_line":"difference is that the operator can see all networks, while a"},{"line_number":201,"context_line":"project will only see the networks under their project id."},{"line_number":202,"context_line":""},{"line_number":203,"context_line":"Tap interface names identify the Neutron port they are"},{"line_number":204,"context_line":"associated with and the Neutron port identifies the Neutron"},{"line_number":205,"context_line":"network. Therefore, it is possible to take counters from each tap"},{"line_number":206,"context_line":"interface and map them into an entry in the appropriate tables,"}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_eff73857","line":203,"updated":"2015-12-02 01:26:38.000000000","message":"I don\u0027t think this is subject to change but it\u0027s worth noting it\u0027s implementation specific.","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"1bc336f1f0be538f3eae351caebac158e476d68e","unresolved":false,"context_lines":[{"line_number":200,"context_line":"difference is that the operator can see all networks, while a"},{"line_number":201,"context_line":"project will only see the networks under their project id."},{"line_number":202,"context_line":""},{"line_number":203,"context_line":"Tap interface names identify the Neutron port they are"},{"line_number":204,"context_line":"associated with and the Neutron port identifies the Neutron"},{"line_number":205,"context_line":"network. Therefore, it is possible to take counters from each tap"},{"line_number":206,"context_line":"interface and map them into an entry in the appropriate tables,"}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_463962d6","line":203,"in_reply_to":"9a8ffd7b_eff73857","updated":"2015-12-02 18:27:19.000000000","message":"fair point","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"1bc336f1f0be538f3eae351caebac158e476d68e","unresolved":false,"context_lines":[{"line_number":203,"context_line":"Tap interface names identify the Neutron port they are"},{"line_number":204,"context_line":"associated with and the Neutron port identifies the Neutron"},{"line_number":205,"context_line":"network. Therefore, it is possible to take counters from each tap"},{"line_number":206,"context_line":"interface and map them into an entry in the appropriate tables,"},{"line_number":207,"context_line":"using the following proposed assignments:"},{"line_number":208,"context_line":""},{"line_number":209,"context_line":"* ifTable"}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_0646fa4e","line":206,"updated":"2015-12-02 18:27:19.000000000","message":"note to self: s/an entry/entries/","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"5337d410a238144921781c8cb487e9b25517ecb4","unresolved":false,"context_lines":[{"line_number":226,"context_line":""},{"line_number":227,"context_line":"Section 3.1.6 of [RFC2863]_ provides the details of why 64-bit sized"},{"line_number":228,"context_line":"counters need to be supported. The summary is that with increasing"},{"line_number":229,"context_line":"transimission bandwidth use of 32-bit counters would require a"},{"line_number":230,"context_line":"problematic increase in counter polling frequency (a 1Gbs stream of"},{"line_number":231,"context_line":"full-sized packets will cause a 32-bit counter to wrap in 34 seconds)."},{"line_number":232,"context_line":""}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_cfe594a8","line":229,"updated":"2015-12-02 01:26:38.000000000","message":"typo!!!!","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"1bc336f1f0be538f3eae351caebac158e476d68e","unresolved":false,"context_lines":[{"line_number":226,"context_line":""},{"line_number":227,"context_line":"Section 3.1.6 of [RFC2863]_ provides the details of why 64-bit sized"},{"line_number":228,"context_line":"counters need to be supported. The summary is that with increasing"},{"line_number":229,"context_line":"transimission bandwidth use of 32-bit counters would require a"},{"line_number":230,"context_line":"problematic increase in counter polling frequency (a 1Gbs stream of"},{"line_number":231,"context_line":"full-sized packets will cause a 32-bit counter to wrap in 34 seconds)."},{"line_number":232,"context_line":""}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_6654a6fd","line":229,"in_reply_to":"9a8ffd7b_cfe594a8","updated":"2015-12-02 18:27:19.000000000","message":"Done","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"5337d410a238144921781c8cb487e9b25517ecb4","unresolved":false,"context_lines":[{"line_number":236,"context_line":"Counters tracked by RFC 4293 come in two flavors: ones that are"},{"line_number":237,"context_line":"inherited from the interface, and those that track L3 events,"},{"line_number":238,"context_line":"such as fragmentation, re-assembly, truncations, etc. Given"},{"line_number":239,"context_line":"the existing instrumentation of the systems that underlie Neutron,"},{"line_number":240,"context_line":"the following counters are declared out of scope for this devref:"},{"line_number":241,"context_line":""},{"line_number":242,"context_line":"* ipSystemStatsInHdrErrors, ipIfStatsInHdrErrors"}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_afbff05e","line":239,"updated":"2015-12-02 01:26:38.000000000","message":"not sure I grasp the motivation of this exclusion but I am not against it. I am only curious why.","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"d9729c3ae36684abe6504a5db857c35f7d4f63bf","unresolved":false,"context_lines":[{"line_number":236,"context_line":"Counters tracked by RFC 4293 come in two flavors: ones that are"},{"line_number":237,"context_line":"inherited from the interface, and those that track L3 events,"},{"line_number":238,"context_line":"such as fragmentation, re-assembly, truncations, etc. Given"},{"line_number":239,"context_line":"the existing instrumentation of the systems that underlie Neutron,"},{"line_number":240,"context_line":"the following counters are declared out of scope for this devref:"},{"line_number":241,"context_line":""},{"line_number":242,"context_line":"* ipSystemStatsInHdrErrors, ipIfStatsInHdrErrors"}],"source_content_type":"text/x-rst","patch_set":32,"id":"7a740942_7d5fae72","line":239,"in_reply_to":"7a740942_ad950041","updated":"2015-12-04 13:32:33.000000000","message":"agreed - if somebody knows of a place, I\u0027m all ears!","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"60d12dd2b38885214248bccaf7d984e9a2c5d7d4","unresolved":false,"context_lines":[{"line_number":236,"context_line":"Counters tracked by RFC 4293 come in two flavors: ones that are"},{"line_number":237,"context_line":"inherited from the interface, and those that track L3 events,"},{"line_number":238,"context_line":"such as fragmentation, re-assembly, truncations, etc. Given"},{"line_number":239,"context_line":"the existing instrumentation of the systems that underlie Neutron,"},{"line_number":240,"context_line":"the following counters are declared out of scope for this devref:"},{"line_number":241,"context_line":""},{"line_number":242,"context_line":"* ipSystemStatsInHdrErrors, ipIfStatsInHdrErrors"}],"source_content_type":"text/x-rst","patch_set":32,"id":"7a740942_ad950041","line":239,"in_reply_to":"9a8ffd7b_867e6a68","updated":"2015-12-04 00:30:14.000000000","message":"I suppose we can revisit later, I am sure we can pull them from somewhere.","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"1bc336f1f0be538f3eae351caebac158e476d68e","unresolved":false,"context_lines":[{"line_number":236,"context_line":"Counters tracked by RFC 4293 come in two flavors: ones that are"},{"line_number":237,"context_line":"inherited from the interface, and those that track L3 events,"},{"line_number":238,"context_line":"such as fragmentation, re-assembly, truncations, etc. Given"},{"line_number":239,"context_line":"the existing instrumentation of the systems that underlie Neutron,"},{"line_number":240,"context_line":"the following counters are declared out of scope for this devref:"},{"line_number":241,"context_line":""},{"line_number":242,"context_line":"* ipSystemStatsInHdrErrors, ipIfStatsInHdrErrors"}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_867e6a68","line":239,"in_reply_to":"9a8ffd7b_afbff05e","updated":"2015-12-02 18:27:19.000000000","message":"I\u0027ve not been able to find a place to get these from for the reference implementation.","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"5337d410a238144921781c8cb487e9b25517ecb4","unresolved":false,"context_lines":[{"line_number":307,"context_line":"the initial proof of concept will run a separate agent to isolate the effects"},{"line_number":308,"context_line":"of counter collection and reporting. Once the performance of the collection"},{"line_number":309,"context_line":"agent is understood, then merging the agent into the L3 or ML2 agents will"},{"line_number":310,"context_line":"be considered."},{"line_number":311,"context_line":""},{"line_number":312,"context_line":"In addition, there are two options for how to report counters back to the"},{"line_number":313,"context_line":"Neutron server: push or pull (or asynchronous notification vs polling)."}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_cf06f41f","line":310,"updated":"2015-12-02 01:26:38.000000000","message":"can you say a little about how data collection is going to happen? i.e. what tools you\u0027re going to use to extract the raw data to be mapped?","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"d9729c3ae36684abe6504a5db857c35f7d4f63bf","unresolved":false,"context_lines":[{"line_number":307,"context_line":"the initial proof of concept will run a separate agent to isolate the effects"},{"line_number":308,"context_line":"of counter collection and reporting. Once the performance of the collection"},{"line_number":309,"context_line":"agent is understood, then merging the agent into the L3 or ML2 agents will"},{"line_number":310,"context_line":"be considered."},{"line_number":311,"context_line":""},{"line_number":312,"context_line":"In addition, there are two options for how to report counters back to the"},{"line_number":313,"context_line":"Neutron server: push or pull (or asynchronous notification vs polling)."}],"source_content_type":"text/x-rst","patch_set":32,"id":"7a740942_1d8a2ac0","line":310,"in_reply_to":"7a740942_8db6043c","updated":"2015-12-04 13:32:33.000000000","message":"Yes, I did (that was my original plan) - The PoC commit (213474) will modify this and the next paragraph with a deeper discussion of how the PoC works and what numbers I\u0027ve found in initial load testing.  The only reason it isn\u0027t there yet is that I got tired of trying to keep the two patch sets in sync with each other.","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"60d12dd2b38885214248bccaf7d984e9a2c5d7d4","unresolved":false,"context_lines":[{"line_number":307,"context_line":"the initial proof of concept will run a separate agent to isolate the effects"},{"line_number":308,"context_line":"of counter collection and reporting. Once the performance of the collection"},{"line_number":309,"context_line":"agent is understood, then merging the agent into the L3 or ML2 agents will"},{"line_number":310,"context_line":"be considered."},{"line_number":311,"context_line":""},{"line_number":312,"context_line":"In addition, there are two options for how to report counters back to the"},{"line_number":313,"context_line":"Neutron server: push or pull (or asynchronous notification vs polling)."}],"source_content_type":"text/x-rst","patch_set":32,"id":"7a740942_8db6043c","line":310,"in_reply_to":"9a8ffd7b_8643ca12","updated":"2015-12-04 00:30:14.000000000","message":"have you considered reading this stuff from /proc (assumed that everything you need is already available)?","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"1bc336f1f0be538f3eae351caebac158e476d68e","unresolved":false,"context_lines":[{"line_number":307,"context_line":"the initial proof of concept will run a separate agent to isolate the effects"},{"line_number":308,"context_line":"of counter collection and reporting. Once the performance of the collection"},{"line_number":309,"context_line":"agent is understood, then merging the agent into the L3 or ML2 agents will"},{"line_number":310,"context_line":"be considered."},{"line_number":311,"context_line":""},{"line_number":312,"context_line":"In addition, there are two options for how to report counters back to the"},{"line_number":313,"context_line":"Neutron server: push or pull (or asynchronous notification vs polling)."}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_8643ca12","line":310,"in_reply_to":"9a8ffd7b_cf06f41f","updated":"2015-12-02 18:27:19.000000000","message":"I\u0027m worried that will cause people to complain, but initially it\u0027s shell based ip netns and ip link, when privsep is finished, go to native python (via iproute2)","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"5337d410a238144921781c8cb487e9b25517ecb4","unresolved":false,"context_lines":[{"line_number":319,"context_line":"the current results from each collector. As the L3 and ML2 agents already"},{"line_number":320,"context_line":"have use asynchronous notification to report status back to the Neutron"},{"line_number":321,"context_line":"server, the proof of concept will follow the same model to ease a future"},{"line_number":322,"context_line":"merging of functionality."},{"line_number":323,"context_line":""},{"line_number":324,"context_line":"Data Aggregation"},{"line_number":325,"context_line":"----------------"}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_e2a8af79","line":322,"updated":"2015-12-02 01:26:38.000000000","message":"I wonder if we should mention something about the granularity. You hint that we\u0027re going to the individual packet or single byte. Although there\u0027s an attempt to be in line with the RFCs I wonder if it\u0027d be wiser to introduce a granularity where counters go in a configurable increment, B, KB, MB, GB or 10s, 100s, 1000s of units.\n\nFor instance, you mentioned push vs pull. I am sure you\u0027re not implying to that you want to piggyback exactly on the report mechanism used for agent status management because the time dynamics are different (report interval may be too wide and a shorter period might be required to capture events more granularly), so a word to clarify this would be nice.\n\nNow, say that the agent uses polling to fetch counters, and push to the aggregator (which I think this is what you\u0027re going for, am I right?). I would like to suggest that the two be not in sync. For instance, the agent can collect a counter, but report it only if it has gone above a certain threshold depending on the granularity chosen by the user. This would ease the load in circumstances where less granular counters are desired. Furthermore, I wonder if you intend to collect/report all counters at once, or having separate threads of execution (whichever way they are going to be implemented, greenthread or otherwise) one for each (class of) counter. I assume that dynamics here can also be different, e.g. e counter associated to an instance tap interface may have a different life span from router namespace qr interface.\n\nThoughts? Am I shooting a fly with a bazooka?","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"60d12dd2b38885214248bccaf7d984e9a2c5d7d4","unresolved":false,"context_lines":[{"line_number":319,"context_line":"the current results from each collector. As the L3 and ML2 agents already"},{"line_number":320,"context_line":"have use asynchronous notification to report status back to the Neutron"},{"line_number":321,"context_line":"server, the proof of concept will follow the same model to ease a future"},{"line_number":322,"context_line":"merging of functionality."},{"line_number":323,"context_line":""},{"line_number":324,"context_line":"Data Aggregation"},{"line_number":325,"context_line":"----------------"}],"source_content_type":"text/x-rst","patch_set":32,"id":"7a740942_ad966092","line":322,"in_reply_to":"9a8ffd7b_108e2b79","updated":"2015-12-04 00:30:14.000000000","message":"cool, let\u0027s continue the conversation later on.","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"},{"author":{"_account_id":11682,"name":"Ryan Moats","email":"rmoats@us.ibm.com","username":"regXboi"},"change_message_id":"1bc336f1f0be538f3eae351caebac158e476d68e","unresolved":false,"context_lines":[{"line_number":319,"context_line":"the current results from each collector. As the L3 and ML2 agents already"},{"line_number":320,"context_line":"have use asynchronous notification to report status back to the Neutron"},{"line_number":321,"context_line":"server, the proof of concept will follow the same model to ease a future"},{"line_number":322,"context_line":"merging of functionality."},{"line_number":323,"context_line":""},{"line_number":324,"context_line":"Data Aggregation"},{"line_number":325,"context_line":"----------------"}],"source_content_type":"text/x-rst","patch_set":32,"id":"9a8ffd7b_108e2b79","line":322,"in_reply_to":"9a8ffd7b_e2a8af79","updated":"2015-12-02 18:27:19.000000000","message":"\"Am I shooting a fly with a bazooka?\" -- is this a trick question?\n\nI\u0027m pulling individual packet and bytes because that is what the reference implementation provides.  To the later paragraph on fetching counters, having a settable granularity won\u0027t help things - we\u0027ll still have to poll the underlying counters and we\u0027ll have the additional complexity of not recording an update unless the delta has reached a certain point.  Since the major load I\u0027m worried about is the polling loop, I really don\u0027t see how that helps.\n\nYes on the reason for a different time dynamic.\n\nI think I understand now - you are talking about wanting reporting to be deltas and change the granularity of the delta - if I am correct, then that is certainly a topic for part II and its PoC","commit_id":"1d46a9162a975adef6c5c964f0cd6fce4a762714"}]}
