)]}'
{"specs/amqp10-driver-implementation.rst":[{"author":{"_account_id":2472,"name":"Doug Hellmann","email":"dhellmann@redhat.com","username":"doug-hellmann"},"change_message_id":"02424ae5d6c2fdbef8c810def31c82f51614930c","unresolved":false,"context_lines":[{"line_number":45,"context_line":"prefixes to the addresses specified in the source and target for"},{"line_number":46,"context_line":"receiving and sending links respectively, it should be possible to get"},{"line_number":47,"context_line":"the required functionality working with a variety of implementations"},{"line_number":48,"context_line":"and offer considerable flexibility to deployers."},{"line_number":49,"context_line":""},{"line_number":50,"context_line":"In RPC, there are three patterns for invocations, and a different"},{"line_number":51,"context_line":"address format will be used for each."}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_e30cddef","line":48,"updated":"2014-06-03 19:04:09.000000000","message":"I don\u0027t really follow this paragraph. I thought the need to provision nodes on the fly had to do with brokerless communication, which the previous paragraphs say is out of scope for this blueprint. Am I misunderstanding the need for provisioning nodes on the fly?","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"cd202cb4f5306eb6df165dc90b16f6cd2e914758","unresolved":false,"context_lines":[{"line_number":45,"context_line":"prefixes to the addresses specified in the source and target for"},{"line_number":46,"context_line":"receiving and sending links respectively, it should be possible to get"},{"line_number":47,"context_line":"the required functionality working with a variety of implementations"},{"line_number":48,"context_line":"and offer considerable flexibility to deployers."},{"line_number":49,"context_line":""},{"line_number":50,"context_line":"In RPC, there are three patterns for invocations, and a different"},{"line_number":51,"context_line":"address format will be used for each."}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_bedf480c","line":48,"in_reply_to":"1ae5cdf2_e30cddef","updated":"2014-06-03 20:26:59.000000000","message":"Sorry, I\u0027m using AMQP terminology (which isn\u0027t always intuitive!). In AMQP 1.0 a node is something like a queue or exchange or topic in other specifications. I.t. it is not a broker or a machine, but rather a communication \u0027pathway\u0027 within the broker. I\u0027ll clarify that in the text.","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":2472,"name":"Doug Hellmann","email":"dhellmann@redhat.com","username":"doug-hellmann"},"change_message_id":"02424ae5d6c2fdbef8c810def31c82f51614930c","unresolved":false,"context_lines":[{"line_number":63,"context_line":"topic name with \u0027/broadcast/\u0027 prepended (to distinguish it from the"},{"line_number":64,"context_line":"address used for invocations on only one server in the group). It may"},{"line_number":65,"context_line":"also be prefixed by the exchange if specified and/or by a deployment"},{"line_number":66,"context_line":"specific \u0027broadcast_prefix\u0027 string."},{"line_number":67,"context_line":""},{"line_number":68,"context_line":"For each send request, the appropriate address will be deduced from"},{"line_number":69,"context_line":"the specified target. If the server value on the target is specified,"}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_8318791f","line":66,"updated":"2014-06-03 19:04:09.000000000","message":"How do these mappings differ from what the existing AMQP drivers do? Does it matter if they are inconsistent, or does the protocol change encourage us to use different addressing for some reason?","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"28ff0e03cde130340b177841eaa990201617ee05","unresolved":false,"context_lines":[{"line_number":63,"context_line":"topic name with \u0027/broadcast/\u0027 prepended (to distinguish it from the"},{"line_number":64,"context_line":"address used for invocations on only one server in the group). It may"},{"line_number":65,"context_line":"also be prefixed by the exchange if specified and/or by a deployment"},{"line_number":66,"context_line":"specific \u0027broadcast_prefix\u0027 string."},{"line_number":67,"context_line":""},{"line_number":68,"context_line":"For each send request, the appropriate address will be deduced from"},{"line_number":69,"context_line":"the specified target. If the server value on the target is specified,"}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_facf72b1","line":66,"in_reply_to":"1ae5cdf2_4bb41ff2","updated":"2014-06-17 15:13:42.000000000","message":"Yes. There is no concept of an \u0027exchange\u0027 in AMQP 1.0. If an exchange is specified in the oslo.messaging configuration it is added into the address used such that agiven topic or server name etc need only be unique within the context of the exchange.","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"cd202cb4f5306eb6df165dc90b16f6cd2e914758","unresolved":false,"context_lines":[{"line_number":63,"context_line":"topic name with \u0027/broadcast/\u0027 prepended (to distinguish it from the"},{"line_number":64,"context_line":"address used for invocations on only one server in the group). It may"},{"line_number":65,"context_line":"also be prefixed by the exchange if specified and/or by a deployment"},{"line_number":66,"context_line":"specific \u0027broadcast_prefix\u0027 string."},{"line_number":67,"context_line":""},{"line_number":68,"context_line":"For each send request, the appropriate address will be deduced from"},{"line_number":69,"context_line":"the specified target. If the server value on the target is specified,"}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_de0b348c","line":66,"in_reply_to":"1ae5cdf2_8318791f","updated":"2014-06-03 20:26:59.000000000","message":"The 1.0 version of the specification differs significantly from either of the older versions used in oslo.messaging already.  The mapping used for the exsiting drivers is in some ways a little odd (e.g. creating an exchange per response, or even creating an exchange for all responses, to which only one queue is ever bound). In 1.0 (as mentioned above) there is no longer any equivalent to queue-declare or exchange-declare. The exchange types defined pre 1.0 are also no longer in the spec.\n\nSo for this reason the mapping is quite different in 1.0, but I believe is quite natural in 1.0 terms.","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":2472,"name":"Doug Hellmann","email":"dhellmann@redhat.com","username":"doug-hellmann"},"change_message_id":"aff3847f164b6ac474d6d699b783730abf8fee41","unresolved":false,"context_lines":[{"line_number":63,"context_line":"topic name with \u0027/broadcast/\u0027 prepended (to distinguish it from the"},{"line_number":64,"context_line":"address used for invocations on only one server in the group). It may"},{"line_number":65,"context_line":"also be prefixed by the exchange if specified and/or by a deployment"},{"line_number":66,"context_line":"specific \u0027broadcast_prefix\u0027 string."},{"line_number":67,"context_line":""},{"line_number":68,"context_line":"For each send request, the appropriate address will be deduced from"},{"line_number":69,"context_line":"the specified target. If the server value on the target is specified,"}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_4bb41ff2","line":66,"in_reply_to":"1ae5cdf2_de0b348c","updated":"2014-06-09 15:23:35.000000000","message":"So AMQP 1.0 doesn\u0027t require an exchange, but if one is given in the configuration it will be used in the addresses generated by the driver?","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":2472,"name":"Doug Hellmann","email":"dhellmann@redhat.com","username":"doug-hellmann"},"change_message_id":"02424ae5d6c2fdbef8c810def31c82f51614930c","unresolved":false,"context_lines":[{"line_number":103,"context_line":"subscribe to broadcasts for all the servers in the group (i.e. one for"},{"line_number":104,"context_line":"each form of address described in A, B and C above).  The first two"},{"line_number":105,"context_line":"receiver links will have the distribution mode set to move, the third"},{"line_number":106,"context_line":"will have the distribution mode set to copy"},{"line_number":107,"context_line":""},{"line_number":108,"context_line":"Alternatives"},{"line_number":109,"context_line":"------------"}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_a3ec95f4","line":106,"updated":"2014-06-03 19:04:09.000000000","message":"What is the \"distribution mode\"?","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"cd202cb4f5306eb6df165dc90b16f6cd2e914758","unresolved":false,"context_lines":[{"line_number":103,"context_line":"subscribe to broadcasts for all the servers in the group (i.e. one for"},{"line_number":104,"context_line":"each form of address described in A, B and C above).  The first two"},{"line_number":105,"context_line":"receiver links will have the distribution mode set to move, the third"},{"line_number":106,"context_line":"will have the distribution mode set to copy"},{"line_number":107,"context_line":""},{"line_number":108,"context_line":"Alternatives"},{"line_number":109,"context_line":"------------"}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_9eed2cba","line":106,"in_reply_to":"1ae5cdf2_a3ec95f4","updated":"2014-06-03 20:26:59.000000000","message":"Again, terminlogoy from the spec. That mode determines whether a message is acquired by one of a set of competing consumers or whether it is delivered to all subscribers (i.e. whether fanout is used).","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":2472,"name":"Doug Hellmann","email":"dhellmann@redhat.com","username":"doug-hellmann"},"change_message_id":"02424ae5d6c2fdbef8c810def31c82f51614930c","unresolved":false,"context_lines":[{"line_number":104,"context_line":"each form of address described in A, B and C above).  The first two"},{"line_number":105,"context_line":"receiver links will have the distribution mode set to move, the third"},{"line_number":106,"context_line":"will have the distribution mode set to copy"},{"line_number":107,"context_line":""},{"line_number":108,"context_line":"Alternatives"},{"line_number":109,"context_line":"------------"},{"line_number":110,"context_line":""}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_83c6b966","line":107,"updated":"2014-06-03 19:04:09.000000000","message":"Could you add something to address how message acknowledgement works and how retransmission is handled? In the existing AMQP driver we acknowledge the message before we handle it, except for notifications where no response is going to be generated and there is not likely to be an OpenStack service that can retry the transmission.","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"cd202cb4f5306eb6df165dc90b16f6cd2e914758","unresolved":false,"context_lines":[{"line_number":104,"context_line":"each form of address described in A, B and C above).  The first two"},{"line_number":105,"context_line":"receiver links will have the distribution mode set to move, the third"},{"line_number":106,"context_line":"will have the distribution mode set to copy"},{"line_number":107,"context_line":""},{"line_number":108,"context_line":"Alternatives"},{"line_number":109,"context_line":"------------"},{"line_number":110,"context_line":""}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_5e3e4429","line":107,"in_reply_to":"1ae5cdf2_83c6b966","updated":"2014-06-03 20:26:59.000000000","message":"Will do. At present the code is essentially the same as the qpid driver (based on AMQP 0-10) in that messages are pre-settled, i.e. no explicit acknowledgement is used. That is probably the best choice for the RPC case, but notify would probably want to turn acknowledgements on.\n\nI\u0027ll update the spec to indicate that and note that the acknowledging of notifications has not yet been completed in code.","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":2472,"name":"Doug Hellmann","email":"dhellmann@redhat.com","username":"doug-hellmann"},"change_message_id":"02424ae5d6c2fdbef8c810def31c82f51614930c","unresolved":false,"context_lines":[{"line_number":133,"context_line":"consolidation as it could also accommodate the non-intermediated style"},{"line_number":134,"context_line":"of communication embraced by ZeroMQ. Perhaps even more powerful would"},{"line_number":135,"context_line":"be a hybrid approach, where again the use of a common, standard"},{"line_number":136,"context_line":"protocol would be advantageous."},{"line_number":137,"context_line":""},{"line_number":138,"context_line":"Impact on Existing APIs"},{"line_number":139,"context_line":"-----------------------"}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_63944d43","line":136,"updated":"2014-06-03 19:04:09.000000000","message":"Do you anticipate this driver being the default at some point in the future? I assume that would require brokers shipped in the various distros supporting the newer protocol. Do we know anything about the schedule for that happening (i.e., when *could* this be the default driver, if we wanted it to be)?","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"cd202cb4f5306eb6df165dc90b16f6cd2e914758","unresolved":false,"context_lines":[{"line_number":133,"context_line":"consolidation as it could also accommodate the non-intermediated style"},{"line_number":134,"context_line":"of communication embraced by ZeroMQ. Perhaps even more powerful would"},{"line_number":135,"context_line":"be a hybrid approach, where again the use of a common, standard"},{"line_number":136,"context_line":"protocol would be advantageous."},{"line_number":137,"context_line":""},{"line_number":138,"context_line":"Impact on Existing APIs"},{"line_number":139,"context_line":"-----------------------"}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_deced42d","line":136,"in_reply_to":"1ae5cdf2_63944d43","updated":"2014-06-03 20:26:59.000000000","message":"RabbitMQs AMQP 1.0 support does not at present cover the standard mechanism defined in teh spec for temporary response queues. That could be worked around in the short term if there was sufficient interest in doing so. However I would also anticipate that it is something that would be addressed fairly easily in the broker itself if requested by users.\n\nIn terms of packaging, Qpid is updated in Fedora relatively quickly (I believe the current packages there are sufficient). Qpid packages for debian based distros are I believe being undertaken at present also.\n\nMy view would be that a \u0027beta\u0027 phase where real users get to try out this driver and provide any feedback would be the first essential step.","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":2472,"name":"Doug Hellmann","email":"dhellmann@redhat.com","username":"doug-hellmann"},"change_message_id":"aff3847f164b6ac474d6d699b783730abf8fee41","unresolved":false,"context_lines":[{"line_number":133,"context_line":"consolidation as it could also accommodate the non-intermediated style"},{"line_number":134,"context_line":"of communication embraced by ZeroMQ. Perhaps even more powerful would"},{"line_number":135,"context_line":"be a hybrid approach, where again the use of a common, standard"},{"line_number":136,"context_line":"protocol would be advantageous."},{"line_number":137,"context_line":""},{"line_number":138,"context_line":"Impact on Existing APIs"},{"line_number":139,"context_line":"-----------------------"}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_2b17d3a4","line":136,"in_reply_to":"1ae5cdf2_deced42d","updated":"2014-06-09 15:23:35.000000000","message":"Sure, I\u0027m trying to understand the full anticipated roadmap.","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":2472,"name":"Doug Hellmann","email":"dhellmann@redhat.com","username":"doug-hellmann"},"change_message_id":"02424ae5d6c2fdbef8c810def31c82f51614930c","unresolved":false,"context_lines":[{"line_number":166,"context_line":""},{"line_number":167,"context_line":"The driver would be selected via the existing transport_url"},{"line_number":168,"context_line":"option. The \u0027amqp\u0027 scheme in the url is used to select the AMQP"},{"line_number":169,"context_line":"protocol."},{"line_number":170,"context_line":""},{"line_number":171,"context_line":"There would be further configuration options to tailor the behaviour"},{"line_number":172,"context_line":"of the driver to different deployments if desired, such as the various"}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_2386c513","line":169,"updated":"2014-06-03 19:04:09.000000000","message":"Should we include the version in the scheme? \"amqp1\"?","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"cd202cb4f5306eb6df165dc90b16f6cd2e914758","unresolved":false,"context_lines":[{"line_number":166,"context_line":""},{"line_number":167,"context_line":"The driver would be selected via the existing transport_url"},{"line_number":168,"context_line":"option. The \u0027amqp\u0027 scheme in the url is used to select the AMQP"},{"line_number":169,"context_line":"protocol."},{"line_number":170,"context_line":""},{"line_number":171,"context_line":"There would be further configuration options to tailor the behaviour"},{"line_number":172,"context_line":"of the driver to different deployments if desired, such as the various"}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_7efa2044","line":169,"in_reply_to":"1ae5cdf2_2386c513","updated":"2014-06-03 20:26:59.000000000","message":"I would personally rather not as it looks quite ugly. AMQP 1.0 is the ratified version of the protocol, as published by OASIS and now an ISO standard. As such I myself think the simple name is justified.","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":2472,"name":"Doug Hellmann","email":"dhellmann@redhat.com","username":"doug-hellmann"},"change_message_id":"aff3847f164b6ac474d6d699b783730abf8fee41","unresolved":false,"context_lines":[{"line_number":166,"context_line":""},{"line_number":167,"context_line":"The driver would be selected via the existing transport_url"},{"line_number":168,"context_line":"option. The \u0027amqp\u0027 scheme in the url is used to select the AMQP"},{"line_number":169,"context_line":"protocol."},{"line_number":170,"context_line":""},{"line_number":171,"context_line":"There would be further configuration options to tailor the behaviour"},{"line_number":172,"context_line":"of the driver to different deployments if desired, such as the various"}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_ab50435f","line":169,"in_reply_to":"1ae5cdf2_7efa2044","updated":"2014-06-09 15:23:35.000000000","message":"What about when AMQP 1.1 comes out? Or 2.0?","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"28ff0e03cde130340b177841eaa990201617ee05","unresolved":false,"context_lines":[{"line_number":166,"context_line":""},{"line_number":167,"context_line":"The driver would be selected via the existing transport_url"},{"line_number":168,"context_line":"option. The \u0027amqp\u0027 scheme in the url is used to select the AMQP"},{"line_number":169,"context_line":"protocol."},{"line_number":170,"context_line":""},{"line_number":171,"context_line":"There would be further configuration options to tailor the behaviour"},{"line_number":172,"context_line":"of the driver to different deployments if desired, such as the various"}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_7ae60218","line":169,"in_reply_to":"1ae5cdf2_ab50435f","updated":"2014-06-17 15:13:42.000000000","message":"If there were a new version of the core specification, which personally I think would be many years away, it would need to be backward compatible at the wire level with 1.0 (that was a commitment the AMQP working group made). SO in that case I would hope that the same driver would be used for olso.messaging, but it would include the ability to negotiate up if needed.","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":2472,"name":"Doug Hellmann","email":"dhellmann@redhat.com","username":"doug-hellmann"},"change_message_id":"02424ae5d6c2fdbef8c810def31c82f51614930c","unresolved":false,"context_lines":[{"line_number":173,"context_line":"prefix options described above. The defaults have been selected to"},{"line_number":174,"context_line":"make broker configuration simple. In general configuration of the"},{"line_number":175,"context_line":"messaging infrastructure is preferred to configuration of the driver"},{"line_number":176,"context_line":"as it allows for more transparent, centralised changes."},{"line_number":177,"context_line":""},{"line_number":178,"context_line":"Developer Impact"},{"line_number":179,"context_line":"----------------"}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_e3969d3e","line":176,"updated":"2014-06-03 19:04:09.000000000","message":"Could you list which brokers other than qpid support AMQP 1.x at a level that would make them compatible with this driver?","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"28ff0e03cde130340b177841eaa990201617ee05","unresolved":false,"context_lines":[{"line_number":173,"context_line":"prefix options described above. The defaults have been selected to"},{"line_number":174,"context_line":"make broker configuration simple. In general configuration of the"},{"line_number":175,"context_line":"messaging infrastructure is preferred to configuration of the driver"},{"line_number":176,"context_line":"as it allows for more transparent, centralised changes."},{"line_number":177,"context_line":""},{"line_number":178,"context_line":"Developer Impact"},{"line_number":179,"context_line":"----------------"}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_1a097e95","line":176,"in_reply_to":"1ae5cdf2_8bdaa7e7","updated":"2014-06-17 15:13:42.000000000","message":"Done","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":2472,"name":"Doug Hellmann","email":"dhellmann@redhat.com","username":"doug-hellmann"},"change_message_id":"aff3847f164b6ac474d6d699b783730abf8fee41","unresolved":false,"context_lines":[{"line_number":173,"context_line":"prefix options described above. The defaults have been selected to"},{"line_number":174,"context_line":"make broker configuration simple. In general configuration of the"},{"line_number":175,"context_line":"messaging infrastructure is preferred to configuration of the driver"},{"line_number":176,"context_line":"as it allows for more transparent, centralised changes."},{"line_number":177,"context_line":""},{"line_number":178,"context_line":"Developer Impact"},{"line_number":179,"context_line":"----------------"}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_8bdaa7e7","line":176,"in_reply_to":"1ae5cdf2_de3894f1","updated":"2014-06-09 15:23:35.000000000","message":"Could you add that information to the spec, so in the future when we look back at the document we can understand what the expectations were for where the new driver could be deployed? Couch it in whatever terms make you feel comfortable, since the point isn\u0027t to commit to working configurations but to point out that we anticipate support to vary.","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"cd202cb4f5306eb6df165dc90b16f6cd2e914758","unresolved":false,"context_lines":[{"line_number":173,"context_line":"prefix options described above. The defaults have been selected to"},{"line_number":174,"context_line":"make broker configuration simple. In general configuration of the"},{"line_number":175,"context_line":"messaging infrastructure is preferred to configuration of the driver"},{"line_number":176,"context_line":"as it allows for more transparent, centralised changes."},{"line_number":177,"context_line":""},{"line_number":178,"context_line":"Developer Impact"},{"line_number":179,"context_line":"----------------"}],"source_content_type":"text/x-rst","patch_set":2,"id":"1ae5cdf2_de3894f1","line":176,"in_reply_to":"1ae5cdf2_e3969d3e","updated":"2014-06-03 20:26:59.000000000","message":"I haven\u0027t tested against anything other the qpidd and Qpid dispatch router. Based on my experience with other applications, I believe it would work against the current ActiveMQ release (as well as ApolloMQ). For RabbitMQ, the lack of support for the \u0027dynamic\u0027 flag on link source is the main issue at present. \n\nIn theory it should also work against Microsofts ServiceBus and IITs SwiftMQ as they support the aspects of the protocol used. Those might require much more manual configuration at present, I haven\u0027t used them so am unable to say for sure.","commit_id":"cd01468e977b422b38f96652e07acbd2ee140330"},{"author":{"_account_id":1247,"name":"Mark McLoughlin","email":"markmc@redhat.com","username":"markmc"},"change_message_id":"ad968330d11ff86f417a53496734cda1ed048f86","unresolved":false,"context_lines":[{"line_number":22,"context_line":"wishing to move to any other system that spoke AMQP 1.0[1]. That would no"},{"line_number":23,"context_line":"longer require a new driver, merely different configuration and"},{"line_number":24,"context_line":"deployment. That would then allow openstack to more easily take"},{"line_number":25,"context_line":"advantage of any emerging innovations in this space."},{"line_number":26,"context_line":""},{"line_number":27,"context_line":"Proposed change"},{"line_number":28,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":4,"id":"dab895d6_333e95b9","line":25,"updated":"2014-06-30 16:01:15.000000000","message":"This is good, but part of the problem this driver is addressing is to support running OpenStack on messaging systems other than RabbitMQ, Qpid, or ZeroMQ\n\nAIUI the Qpid Dispatch router is the first new thing we want to support (http://qpid.apache.org/components/dispatch-router/) but perhaps ActiveMQ too?\n\nWhichever it is, I think it\u0027s worth calling out - it goes to the more immediate benefit OpenStack users will see","commit_id":"3fdd69b57a1264fd53f52a145cdd70a4bd745034"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"f5e55aea63f9944dcd65445054a8c7414bd5cbba","unresolved":false,"context_lines":[{"line_number":22,"context_line":"wishing to move to any other system that spoke AMQP 1.0[1]. That would no"},{"line_number":23,"context_line":"longer require a new driver, merely different configuration and"},{"line_number":24,"context_line":"deployment. That would then allow openstack to more easily take"},{"line_number":25,"context_line":"advantage of any emerging innovations in this space."},{"line_number":26,"context_line":""},{"line_number":27,"context_line":"Proposed change"},{"line_number":28,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":4,"id":"baada198_711ab36f","line":25,"in_reply_to":"dab895d6_333e95b9","updated":"2014-07-01 13:11:39.000000000","message":"Ok, I\u0027ve mentioned Qpid Dispatch Router along with a brief summary of why it might be an appealing choice.","commit_id":"3fdd69b57a1264fd53f52a145cdd70a4bd745034"},{"author":{"_account_id":1247,"name":"Mark McLoughlin","email":"markmc@redhat.com","username":"markmc"},"change_message_id":"ad968330d11ff86f417a53496734cda1ed048f86","unresolved":false,"context_lines":[{"line_number":35,"context_line":"brokers). AMQP 1.0 would allow communication (in part or as a whole)"},{"line_number":36,"context_line":"to be conducted directly between peers, without the use of"},{"line_number":37,"context_line":"intermediaries. Development of this aspect however is considered a future"},{"line_number":38,"context_line":"extension."},{"line_number":39,"context_line":""},{"line_number":40,"context_line":"The aim is to support as many of the existing 1.0 enabled"},{"line_number":41,"context_line":"intermediaries as we can without overly distorting the mapping."}],"source_content_type":"text/x-rst","patch_set":4,"id":"dab895d6_b3870561","line":38,"updated":"2014-06-30 16:01:15.000000000","message":"It might be worth going into how you\u0027d see the driver being extended to support this - e.g. it would be good to be confident we can do it in a backwards compatible way","commit_id":"3fdd69b57a1264fd53f52a145cdd70a4bd745034"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"f5e55aea63f9944dcd65445054a8c7414bd5cbba","unresolved":false,"context_lines":[{"line_number":35,"context_line":"brokers). AMQP 1.0 would allow communication (in part or as a whole)"},{"line_number":36,"context_line":"to be conducted directly between peers, without the use of"},{"line_number":37,"context_line":"intermediaries. Development of this aspect however is considered a future"},{"line_number":38,"context_line":"extension."},{"line_number":39,"context_line":""},{"line_number":40,"context_line":"The aim is to support as many of the existing 1.0 enabled"},{"line_number":41,"context_line":"intermediaries as we can without overly distorting the mapping."}],"source_content_type":"text/x-rst","patch_set":4,"id":"baada198_f15c0328","line":38,"in_reply_to":"dab895d6_b3870561","updated":"2014-07-01 13:11:39.000000000","message":"I\u0027ve described the possible approaches in an appendix. I actually had a patch nearly finished at one point that allowed for direct communication, so I am quite confident it can be done in an entirely backward compatible manner (requiring only new configuration to enable it if desired)","commit_id":"3fdd69b57a1264fd53f52a145cdd70a4bd745034"},{"author":{"_account_id":1247,"name":"Mark McLoughlin","email":"markmc@redhat.com","username":"markmc"},"change_message_id":"ad968330d11ff86f417a53496734cda1ed048f86","unresolved":false,"context_lines":[{"line_number":38,"context_line":"extension."},{"line_number":39,"context_line":""},{"line_number":40,"context_line":"The aim is to support as many of the existing 1.0 enabled"},{"line_number":41,"context_line":"intermediaries as we can without overly distorting the mapping."},{"line_number":42,"context_line":""},{"line_number":43,"context_line":"AMQP 1.0 does not provide any standard configuration mechanism for"},{"line_number":44,"context_line":"configuring queues/topics (known as \u0027nodes\u0027 in the AMQP 1.0 spec) on"}],"source_content_type":"text/x-rst","patch_set":4,"id":"dab895d6_738d6d7b","line":41,"updated":"2014-06-30 16:01:15.000000000","message":"Not immediately clear what you mean by \"mapping\" - could be explained upfront. Is it about address patterns like http://qpid.apache.org/releases/qpid-dispatch-0.2/book/amqp-mapping.html ?","commit_id":"3fdd69b57a1264fd53f52a145cdd70a4bd745034"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"f5e55aea63f9944dcd65445054a8c7414bd5cbba","unresolved":false,"context_lines":[{"line_number":38,"context_line":"extension."},{"line_number":39,"context_line":""},{"line_number":40,"context_line":"The aim is to support as many of the existing 1.0 enabled"},{"line_number":41,"context_line":"intermediaries as we can without overly distorting the mapping."},{"line_number":42,"context_line":""},{"line_number":43,"context_line":"AMQP 1.0 does not provide any standard configuration mechanism for"},{"line_number":44,"context_line":"configuring queues/topics (known as \u0027nodes\u0027 in the AMQP 1.0 spec) on"}],"source_content_type":"text/x-rst","patch_set":4,"id":"baada198_b1d52b45","line":41,"in_reply_to":"dab895d6_738d6d7b","updated":"2014-07-01 13:11:39.000000000","message":"I just mean the \u0027mapping on to a protocol\u0027 as mentioned in the problem description above. I\u0027ve reworded this to try and convey the point more clearly. It is really just saying that we don\u0027t want to bend the protocol to fit a particular implementation, but want to keep to standard, conventional usage patterns.","commit_id":"3fdd69b57a1264fd53f52a145cdd70a4bd745034"},{"author":{"_account_id":1247,"name":"Mark McLoughlin","email":"markmc@redhat.com","username":"markmc"},"change_message_id":"ad968330d11ff86f417a53496734cda1ed048f86","unresolved":false,"context_lines":[{"line_number":46,"context_line":"addresses specified in the source and target for receiving and sending"},{"line_number":47,"context_line":"links respectively, it should be possible to get the required"},{"line_number":48,"context_line":"functionality working with a variety of implementations and offer"},{"line_number":49,"context_line":"considerable flexibility to deployers."},{"line_number":50,"context_line":""},{"line_number":51,"context_line":"In RPC, there are three patterns for invocations, and a different"},{"line_number":52,"context_line":"address format will be used for each."}],"source_content_type":"text/x-rst","patch_set":4,"id":"baada198_a12424c7","line":49,"updated":"2014-06-30 16:01:15.000000000","message":"This could do with some elaboration, I think\n\nSome of the basic AMQP 1.0 concepts could be explained first - e.g. nodes, addresses, links, distribution modes, etc.\n\nBasically, I think the below shouldn\u0027t require a detailed understanding of AMQP 1.0 to understand how the proposal implements the semantics we\u0027re looking for","commit_id":"3fdd69b57a1264fd53f52a145cdd70a4bd745034"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"f5e55aea63f9944dcd65445054a8c7414bd5cbba","unresolved":false,"context_lines":[{"line_number":46,"context_line":"addresses specified in the source and target for receiving and sending"},{"line_number":47,"context_line":"links respectively, it should be possible to get the required"},{"line_number":48,"context_line":"functionality working with a variety of implementations and offer"},{"line_number":49,"context_line":"considerable flexibility to deployers."},{"line_number":50,"context_line":""},{"line_number":51,"context_line":"In RPC, there are three patterns for invocations, and a different"},{"line_number":52,"context_line":"address format will be used for each."}],"source_content_type":"text/x-rst","patch_set":4,"id":"baada198_d141bf0c","line":49,"in_reply_to":"baada198_a12424c7","updated":"2014-07-01 13:11:39.000000000","message":"I\u0027ve reworded this a little along the lines of your suggestion to try and make it clearer.","commit_id":"3fdd69b57a1264fd53f52a145cdd70a4bd745034"},{"author":{"_account_id":1247,"name":"Mark McLoughlin","email":"markmc@redhat.com","username":"markmc"},"change_message_id":"ad968330d11ff86f417a53496734cda1ed048f86","unresolved":false,"context_lines":[{"line_number":64,"context_line":"topic name with \u0027/broadcast/\u0027 prepended (to distinguish it from the"},{"line_number":65,"context_line":"address used for invocations on only one server in the group). It may"},{"line_number":66,"context_line":"also be prefixed by the exchange if specified and/or by a deployment"},{"line_number":67,"context_line":"specific \u0027broadcast_prefix\u0027 string."},{"line_number":68,"context_line":""},{"line_number":69,"context_line":"For each send request, the appropriate address will be deduced from"},{"line_number":70,"context_line":"the specified target. If the server value on the target is specified,"}],"source_content_type":"text/x-rst","patch_set":4,"id":"baada198_0cb6e73d","line":67,"updated":"2014-06-30 16:01:15.000000000","message":"How should an operator choose these strings? Will they likely need to configure their router/broker to treat addresses with these prefixes in certain ways?\n\nA concrete example with the dispatch router would be good\n\nThese have default values of \u0027exclusive\u0027, \u0027unicast\u0027 and \u0027broadcast\u0027 respectively, right?\n\nThis describes the format, right?\n\n  %(prefix)s.%(exchange)s.%(topic)s[|.%(server)s|.all]","commit_id":"3fdd69b57a1264fd53f52a145cdd70a4bd745034"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"f5e55aea63f9944dcd65445054a8c7414bd5cbba","unresolved":false,"context_lines":[{"line_number":64,"context_line":"topic name with \u0027/broadcast/\u0027 prepended (to distinguish it from the"},{"line_number":65,"context_line":"address used for invocations on only one server in the group). It may"},{"line_number":66,"context_line":"also be prefixed by the exchange if specified and/or by a deployment"},{"line_number":67,"context_line":"specific \u0027broadcast_prefix\u0027 string."},{"line_number":68,"context_line":""},{"line_number":69,"context_line":"For each send request, the appropriate address will be deduced from"},{"line_number":70,"context_line":"the specified target. If the server value on the target is specified,"}],"source_content_type":"text/x-rst","patch_set":4,"id":"baada198_b1636bb0","line":67,"in_reply_to":"baada198_0cb6e73d","updated":"2014-07-01 13:11:39.000000000","message":"I\u0027ve added quite a bit more explanation of this point as I agree it was porrly described. Hopefully it makes more sense now. I\u0027ve also given two different examples of deployment.","commit_id":"3fdd69b57a1264fd53f52a145cdd70a4bd745034"},{"author":{"_account_id":1247,"name":"Mark McLoughlin","email":"markmc@redhat.com","username":"markmc"},"change_message_id":"ad968330d11ff86f417a53496734cda1ed048f86","unresolved":false,"context_lines":[{"line_number":81,"context_line":"overhead of recreating them on each request. (The alternative of using a"},{"line_number":82,"context_line":"single sending link and setting the \u0027to\u0027 field is also a possibility,"},{"line_number":83,"context_line":"but it would restrict the number of intermediaries that could be"},{"line_number":84,"context_line":"used). If there is no sender link yet for the specified address one will be"},{"line_number":85,"context_line":"created and added to the cache. The cache will be cleared on"},{"line_number":86,"context_line":"reconnect."},{"line_number":87,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"baada198_8ce2f701","line":84,"updated":"2014-06-30 16:01:15.000000000","message":"That \"number of intermediaries\" restriction could be explained in detail - perhaps in \"Alternatives\" below","commit_id":"3fdd69b57a1264fd53f52a145cdd70a4bd745034"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"f5e55aea63f9944dcd65445054a8c7414bd5cbba","unresolved":false,"context_lines":[{"line_number":81,"context_line":"overhead of recreating them on each request. (The alternative of using a"},{"line_number":82,"context_line":"single sending link and setting the \u0027to\u0027 field is also a possibility,"},{"line_number":83,"context_line":"but it would restrict the number of intermediaries that could be"},{"line_number":84,"context_line":"used). If there is no sender link yet for the specified address one will be"},{"line_number":85,"context_line":"created and added to the cache. The cache will be cleared on"},{"line_number":86,"context_line":"reconnect."},{"line_number":87,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"baada198_11f977c8","line":84,"in_reply_to":"baada198_8ce2f701","updated":"2014-07-01 13:11:39.000000000","message":"By this I just mean it could unnecessarily restrict choice, since not every broker is required to support this. I\u0027ve updated the wording to make that (hopefully) clearer.\n\nThis is an example of trying to maximise choice by avoiding less common usage patterns.","commit_id":"3fdd69b57a1264fd53f52a145cdd70a4bd745034"},{"author":{"_account_id":1247,"name":"Mark McLoughlin","email":"markmc@redhat.com","username":"markmc"},"change_message_id":"ad968330d11ff86f417a53496734cda1ed048f86","unresolved":false,"context_lines":[{"line_number":275,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":276,"context_line":""},{"line_number":277,"context_line":"This driver depends on the qpid-proton python library for AMQP 1.0"},{"line_number":278,"context_line":"support. Pyngus may be added as dependency, see Work Items above."},{"line_number":279,"context_line":""},{"line_number":280,"context_line":"References"},{"line_number":281,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":4,"id":"baada198_acc97b5c","line":278,"updated":"2014-06-30 16:01:15.000000000","message":"We need to talk about the viability of taking a dependency on Pyngus and the Proton Python bindings in Juno\n\nKen explained that pyngus is available from PyPI, but requires the Proton Python bindings to already be installed. This won\u0027t work in virtualenvs which do not include \u0027sitelib\u0027 in PYTHONPATH, and this is the default mode for virtualenvs used in OpenStack testing.\n\nThere\u0027s also the question of the Proton C library not being widely available in all relevant distros yet.\n\nSo, two questions to address here: 1) the availability of the Proton Python bindings on PyPI and 2) the availability of the Proton C library in various distros","commit_id":"3fdd69b57a1264fd53f52a145cdd70a4bd745034"},{"author":{"_account_id":1247,"name":"Mark McLoughlin","email":"markmc@redhat.com","username":"markmc"},"change_message_id":"1683699e2afe48197f923482fd38d31aa12f8744","unresolved":false,"context_lines":[{"line_number":26,"context_line":""},{"line_number":27,"context_line":"As an example, this would allow the use of the Dispatch Router from"},{"line_number":28,"context_line":"the Apache Qpid project to be used. This has been designed as a"},{"line_number":29,"context_line":"distributed router rather than a single-process, stor-and-forward"},{"line_number":30,"context_line":"broker and offers simple scalability and redundancy."},{"line_number":31,"context_line":""},{"line_number":32,"context_line":"Proposed change"}],"source_content_type":"text/x-rst","patch_set":5,"id":"baada198_7b30b0e8","line":29,"updated":"2014-07-02 08:23:53.000000000","message":"store-and-forward","commit_id":"fca2bbddc63685e2eec7c3a60c363e6b11ad2d11"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"06e1e2b94cf3c6c8cccd52ddaa6aaf1ea9aba446","unresolved":false,"context_lines":[{"line_number":26,"context_line":""},{"line_number":27,"context_line":"As an example, this would allow the use of the Dispatch Router from"},{"line_number":28,"context_line":"the Apache Qpid project to be used. This has been designed as a"},{"line_number":29,"context_line":"distributed router rather than a single-process, stor-and-forward"},{"line_number":30,"context_line":"broker and offers simple scalability and redundancy."},{"line_number":31,"context_line":""},{"line_number":32,"context_line":"Proposed change"}],"source_content_type":"text/x-rst","patch_set":5,"id":"baada198_7602850c","line":29,"in_reply_to":"baada198_7b30b0e8","updated":"2014-07-02 09:03:45.000000000","message":"Done","commit_id":"fca2bbddc63685e2eec7c3a60c363e6b11ad2d11"},{"author":{"_account_id":1247,"name":"Mark McLoughlin","email":"markmc@redhat.com","username":"markmc"},"change_message_id":"1683699e2afe48197f923482fd38d31aa12f8744","unresolved":false,"context_lines":[{"line_number":88,"context_line":"address starting with \u0027unicast\u0027 (or \u0027exclusive\u0027) would be routed to"},{"line_number":89,"context_line":"only one subscriber for that address. Any message for an address"},{"line_number":90,"context_line":"starting with \u0027broadcast\u0027 would be sent to all subscribers for that"},{"line_number":91,"context_line":"address."},{"line_number":92,"context_line":""},{"line_number":93,"context_line":"SOme other broker might not be configurable in this way, but might"},{"line_number":94,"context_line":"have built in conventions around address patterns. E.g. all address of"}],"source_content_type":"text/x-rst","patch_set":5,"id":"baada198_fb868012","line":91,"updated":"2014-07-02 08:23:53.000000000","message":"Perhaps the spec isn\u0027t the place to do this, but we really need concrete instructions on how to configure the dispatch router","commit_id":"fca2bbddc63685e2eec7c3a60c363e6b11ad2d11"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"06e1e2b94cf3c6c8cccd52ddaa6aaf1ea9aba446","unresolved":false,"context_lines":[{"line_number":88,"context_line":"address starting with \u0027unicast\u0027 (or \u0027exclusive\u0027) would be routed to"},{"line_number":89,"context_line":"only one subscriber for that address. Any message for an address"},{"line_number":90,"context_line":"starting with \u0027broadcast\u0027 would be sent to all subscribers for that"},{"line_number":91,"context_line":"address."},{"line_number":92,"context_line":""},{"line_number":93,"context_line":"SOme other broker might not be configurable in this way, but might"},{"line_number":94,"context_line":"have built in conventions around address patterns. E.g. all address of"}],"source_content_type":"text/x-rst","patch_set":5,"id":"baada198_310f3730","line":91,"in_reply_to":"baada198_fb868012","updated":"2014-07-02 09:03:45.000000000","message":"I\u0027ve added another couple of appendices with some concrete detail in them. I agree that at some point a more complete guide for setting things up would be required.","commit_id":"fca2bbddc63685e2eec7c3a60c363e6b11ad2d11"},{"author":{"_account_id":1247,"name":"Mark McLoughlin","email":"markmc@redhat.com","username":"markmc"},"change_message_id":"1683699e2afe48197f923482fd38d31aa12f8744","unresolved":false,"context_lines":[{"line_number":90,"context_line":"starting with \u0027broadcast\u0027 would be sent to all subscribers for that"},{"line_number":91,"context_line":"address."},{"line_number":92,"context_line":""},{"line_number":93,"context_line":"SOme other broker might not be configurable in this way, but might"},{"line_number":94,"context_line":"have built in conventions around address patterns. E.g. all address of"},{"line_number":95,"context_line":"the form \u0027/queues/foo\u0027 would be treated as queues and messages sent to"},{"line_number":96,"context_line":"them would be allocated to only one consumer. Address of the form"}],"source_content_type":"text/x-rst","patch_set":5,"id":"baada198_db59bcff","line":93,"updated":"2014-07-02 08:23:53.000000000","message":"Some other brokers","commit_id":"fca2bbddc63685e2eec7c3a60c363e6b11ad2d11"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"06e1e2b94cf3c6c8cccd52ddaa6aaf1ea9aba446","unresolved":false,"context_lines":[{"line_number":90,"context_line":"starting with \u0027broadcast\u0027 would be sent to all subscribers for that"},{"line_number":91,"context_line":"address."},{"line_number":92,"context_line":""},{"line_number":93,"context_line":"SOme other broker might not be configurable in this way, but might"},{"line_number":94,"context_line":"have built in conventions around address patterns. E.g. all address of"},{"line_number":95,"context_line":"the form \u0027/queues/foo\u0027 would be treated as queues and messages sent to"},{"line_number":96,"context_line":"them would be allocated to only one consumer. Address of the form"}],"source_content_type":"text/x-rst","patch_set":5,"id":"baada198_511e6b64","line":93,"in_reply_to":"baada198_db59bcff","updated":"2014-07-02 09:03:45.000000000","message":"Done","commit_id":"fca2bbddc63685e2eec7c3a60c363e6b11ad2d11"},{"author":{"_account_id":1247,"name":"Mark McLoughlin","email":"markmc@redhat.com","username":"markmc"},"change_message_id":"1683699e2afe48197f923482fd38d31aa12f8744","unresolved":false,"context_lines":[{"line_number":112,"context_line":"overhead of recreating them on each request. (The alternative of using"},{"line_number":113,"context_line":"a single sending link and setting the \u0027to\u0027 field is also a"},{"line_number":114,"context_line":"possibility, but not all intermediaries might support that so it would"},{"line_number":115,"context_line":"be an unnecessary limitationm on choice). If there is no sender link"},{"line_number":116,"context_line":"yet for the specified address one will be created and added to the"},{"line_number":117,"context_line":"cache. The cache will be cleared on reconnect."},{"line_number":118,"context_line":""}],"source_content_type":"text/x-rst","patch_set":5,"id":"baada198_7ba9907a","line":115,"updated":"2014-07-02 08:23:53.000000000","message":"limitation","commit_id":"fca2bbddc63685e2eec7c3a60c363e6b11ad2d11"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"06e1e2b94cf3c6c8cccd52ddaa6aaf1ea9aba446","unresolved":false,"context_lines":[{"line_number":112,"context_line":"overhead of recreating them on each request. (The alternative of using"},{"line_number":113,"context_line":"a single sending link and setting the \u0027to\u0027 field is also a"},{"line_number":114,"context_line":"possibility, but not all intermediaries might support that so it would"},{"line_number":115,"context_line":"be an unnecessary limitationm on choice). If there is no sender link"},{"line_number":116,"context_line":"yet for the specified address one will be created and added to the"},{"line_number":117,"context_line":"cache. The cache will be cleared on reconnect."},{"line_number":118,"context_line":""}],"source_content_type":"text/x-rst","patch_set":5,"id":"baada198_7119af78","line":115,"in_reply_to":"baada198_7ba9907a","updated":"2014-07-02 09:03:45.000000000","message":"Done","commit_id":"fca2bbddc63685e2eec7c3a60c363e6b11ad2d11"},{"author":{"_account_id":1247,"name":"Mark McLoughlin","email":"markmc@redhat.com","username":"markmc"},"change_message_id":"1683699e2afe48197f923482fd38d31aa12f8744","unresolved":false,"context_lines":[{"line_number":313,"context_line":""},{"line_number":314,"context_line":"Supporting direct communication, where messages do not go through an"},{"line_number":315,"context_line":"application layer intermediary user process, requires that"},{"line_number":316,"context_line":"communicants accept incoming connections and can determine the cirrect"},{"line_number":317,"context_line":"hostname and port to connect to."},{"line_number":318,"context_line":""},{"line_number":319,"context_line":"Though this can be done by convention, e.g. using the server name as"}],"source_content_type":"text/x-rst","patch_set":5,"id":"baada198_9b75849a","line":316,"updated":"2014-07-02 08:23:53.000000000","message":"correct","commit_id":"fca2bbddc63685e2eec7c3a60c363e6b11ad2d11"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"06e1e2b94cf3c6c8cccd52ddaa6aaf1ea9aba446","unresolved":false,"context_lines":[{"line_number":313,"context_line":""},{"line_number":314,"context_line":"Supporting direct communication, where messages do not go through an"},{"line_number":315,"context_line":"application layer intermediary user process, requires that"},{"line_number":316,"context_line":"communicants accept incoming connections and can determine the cirrect"},{"line_number":317,"context_line":"hostname and port to connect to."},{"line_number":318,"context_line":""},{"line_number":319,"context_line":"Though this can be done by convention, e.g. using the server name as"}],"source_content_type":"text/x-rst","patch_set":5,"id":"baada198_912723b6","line":316,"in_reply_to":"baada198_9b75849a","updated":"2014-07-02 09:03:45.000000000","message":"Done","commit_id":"fca2bbddc63685e2eec7c3a60c363e6b11ad2d11"},{"author":{"_account_id":1247,"name":"Mark McLoughlin","email":"markmc@redhat.com","username":"markmc"},"change_message_id":"1683699e2afe48197f923482fd38d31aa12f8744","unresolved":false,"context_lines":[{"line_number":346,"context_line":"to a given server."},{"line_number":347,"context_line":""},{"line_number":348,"context_line":"These and indeed other schemes could be easily accomplished in a backward"},{"line_number":349,"context_line":"compatible manner."},{"line_number":350,"context_line":""},{"line_number":351,"context_line":"References"},{"line_number":352,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":5,"id":"baada198_3bd9d883","line":349,"updated":"2014-07-02 08:23:53.000000000","message":"Hmm, interesting. One of the perceived advantages of an intermediary-less approach is that you avoid the overhead of deploying and managing that intermediary. Your proposal requires the intermediary to be still deployed and managed, but it is avoided for much of the time as an optimization.\n\nThen again, the \"reduced management overhead\" argument for an intermediary-less approach is hardly valid if you need to deploy e.g. redis or zookeeper to use as a matchmaker registry.\n\nLots of experimentation remains to be done here, I think.\n\nThanks for writing this up.","commit_id":"fca2bbddc63685e2eec7c3a60c363e6b11ad2d11"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"06e1e2b94cf3c6c8cccd52ddaa6aaf1ea9aba446","unresolved":false,"context_lines":[{"line_number":346,"context_line":"to a given server."},{"line_number":347,"context_line":""},{"line_number":348,"context_line":"These and indeed other schemes could be easily accomplished in a backward"},{"line_number":349,"context_line":"compatible manner."},{"line_number":350,"context_line":""},{"line_number":351,"context_line":"References"},{"line_number":352,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":5,"id":"baada198_11979391","line":349,"in_reply_to":"baada198_3bd9d883","updated":"2014-07-02 09:03:45.000000000","message":"Very much agree that experimentation and debate is required. I tend to think that the reduced management overhead argument applies only to very small scale setups (i.e. where you can avoid a running component with some static configuration solution). Ease of deployment and management is I think an area where different messaging solutions can compete and improve, providing another example of the benefit of choice (without needing a new driver).\n\nAnother argument for direct communication is for scale, but there I think experimentation is needed to see what actually delivers improvements and what trade-offs there  may be (if any).","commit_id":"fca2bbddc63685e2eec7c3a60c363e6b11ad2d11"},{"author":{"_account_id":1247,"name":"Mark McLoughlin","email":"markmc@redhat.com","username":"markmc"},"change_message_id":"e4241cf442ad1682c6bf4b66d895c7c2aa849a76","unresolved":false,"context_lines":[{"line_number":379,"context_line":"Using the default prefixes, passing the following options to qpidd"},{"line_number":380,"context_line":"(0.28 or later) would setup the required semantics for openstack:"},{"line_number":381,"context_line":""},{"line_number":382,"context_line":"--queue-patterns exclusive --queue-patterns unicast --topic-patterns broadcast"},{"line_number":383,"context_line":""},{"line_number":384,"context_line":""},{"line_number":385,"context_line":"References"}],"source_content_type":"text/x-rst","patch_set":6,"id":"baada198_f124dfb0","line":382,"updated":"2014-07-02 09:00:20.000000000","message":"Great! Thanks for this - really useful, and not obvious","commit_id":"8e065239edaf753e41c3004fb8ef5aa568c8b480"}],"specs/juno/amqp10-driver-implementation.rst":[{"author":{"_account_id":1247,"name":"Mark McLoughlin","email":"markmc@redhat.com","username":"markmc"},"change_message_id":"c0591be3ef94ed3838c2cbed746b47e418426065","unresolved":false,"context_lines":[{"line_number":371,"context_line":"  fixed-address {"},{"line_number":372,"context_line":"      prefix: /broadcast"},{"line_number":373,"context_line":"      fanout: multiple"},{"line_number":374,"context_line":"  }"},{"line_number":375,"context_line":""},{"line_number":376,"context_line":""},{"line_number":377,"context_line":"Appendix C: Configuring Qpidd"}],"source_content_type":"text/x-rst","patch_set":8,"id":"baada198_12c07229","line":374,"updated":"2014-07-02 10:06:28.000000000","message":"Playing with this now ... I wonder\n\nIs it sane for our default prefixes to assume those prefixes aren\u0027t already being used?\n\nWhy not use one of e.g.\n\n  /openstack/{unicast,exclusive,broadcast}\n  /oslo/{unicast,exclusive,broadcast}\n  /oslo.messaging/{unicast,exclusive,broadcast}\n\nas the default?","commit_id":"fbcd34d53782ffaf60f533955cd3861ed51279e3"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"14be70ddd08bd20afe95bf721a20cac1ded81a78","unresolved":false,"context_lines":[{"line_number":371,"context_line":"  fixed-address {"},{"line_number":372,"context_line":"      prefix: /broadcast"},{"line_number":373,"context_line":"      fanout: multiple"},{"line_number":374,"context_line":"  }"},{"line_number":375,"context_line":""},{"line_number":376,"context_line":""},{"line_number":377,"context_line":"Appendix C: Configuring Qpidd"}],"source_content_type":"text/x-rst","patch_set":8,"id":"baada198_12e1529a","line":374,"in_reply_to":"baada198_12c07229","updated":"2014-07-02 10:16:35.000000000","message":"You mean that these address patterns are already in use for some other system using the same dispatch router networks?\n\nIf so, then providing the semantics are the same (and with these names it would be very odd if they weren\u0027t), then there would be no issue.\n\nIf for any reason you wanted to you could always alter the prefixes used, to include openstack/oslo.messaging specific differentiating substrings.\n\nThe actual defaults are of course easy to change in code, so I\u0027m not opposed to doing so if it is felt to be advisable.","commit_id":"fbcd34d53782ffaf60f533955cd3861ed51279e3"},{"author":{"_account_id":1247,"name":"Mark McLoughlin","email":"markmc@redhat.com","username":"markmc"},"change_message_id":"0bd1eff532146cc08a3d49ee39c81655f9044d24","unresolved":false,"context_lines":[{"line_number":371,"context_line":"  fixed-address {"},{"line_number":372,"context_line":"      prefix: /broadcast"},{"line_number":373,"context_line":"      fanout: multiple"},{"line_number":374,"context_line":"  }"},{"line_number":375,"context_line":""},{"line_number":376,"context_line":""},{"line_number":377,"context_line":"Appendix C: Configuring Qpidd"}],"source_content_type":"text/x-rst","patch_set":8,"id":"baada198_324bb688","line":374,"in_reply_to":"baada198_12e1529a","updated":"2014-07-02 10:19:54.000000000","message":"If these address prefix and semantics were defaults in qpid-dispatch itself, that\u0027d be cool .... but assuming that if anyone else uses these prefixes then they\u0027d configure the same semantics is a bit tenuous for my liking\n\nAlso for config management simplicity, it\u0027s easier to just assume that the prefixes we want to add cannot be in the config file already","commit_id":"fbcd34d53782ffaf60f533955cd3861ed51279e3"},{"author":{"_account_id":8784,"name":"Gordon Sim","email":"gsim@redhat.com","username":"gordonsim"},"change_message_id":"639493013f23cb86577ecd9f54f8894184a48544","unresolved":false,"context_lines":[{"line_number":371,"context_line":"  fixed-address {"},{"line_number":372,"context_line":"      prefix: /broadcast"},{"line_number":373,"context_line":"      fanout: multiple"},{"line_number":374,"context_line":"  }"},{"line_number":375,"context_line":""},{"line_number":376,"context_line":""},{"line_number":377,"context_line":"Appendix C: Configuring Qpidd"}],"source_content_type":"text/x-rst","patch_set":8,"id":"baada198_75a0b843","line":374,"in_reply_to":"baada198_324bb688","updated":"2014-07-02 10:54:17.000000000","message":"The \u0027fanout\u0027  option is really the crucial one, and my point was that having an address starting \u0027broadcast\u0027 configured to have fanout: single, or an address starting \u0027unicast\u0027 with fanout: multiple, would be very odd indeed.\n\nThe names were indeed chosen such that any service that wanted to make these default, built-in conventions could do so.\n\nHowever, I\u0027m very happy to add some qualifier to those defaults. Should I do this in the spec doc to begin with?","commit_id":"fbcd34d53782ffaf60f533955cd3861ed51279e3"},{"author":{"_account_id":1247,"name":"Mark McLoughlin","email":"markmc@redhat.com","username":"markmc"},"change_message_id":"5e21adae7a3e5a85f94d6c8ea84ba820854f247b","unresolved":false,"context_lines":[{"line_number":371,"context_line":"  fixed-address {"},{"line_number":372,"context_line":"      prefix: /broadcast"},{"line_number":373,"context_line":"      fanout: multiple"},{"line_number":374,"context_line":"  }"},{"line_number":375,"context_line":""},{"line_number":376,"context_line":""},{"line_number":377,"context_line":"Appendix C: Configuring Qpidd"}],"source_content_type":"text/x-rst","patch_set":8,"id":"baada198_f8be7ffe","line":374,"in_reply_to":"baada198_75a0b843","updated":"2014-07-02 11:06:12.000000000","message":"I think we\u0027re both 50/50 on this, so let\u0027s leave it as-is for now and see if anyone else has opinions.\n\nAs you say, it\u0027s very easy to change (at least until we need to start worrying about backwards compat), so it\u0027s fine to worry about it later","commit_id":"fbcd34d53782ffaf60f533955cd3861ed51279e3"},{"author":{"_account_id":2472,"name":"Doug Hellmann","email":"dhellmann@redhat.com","username":"doug-hellmann"},"change_message_id":"e8a3e9a5b77eaa99e7f8d5fc8d7257c16e6d2cf6","unresolved":false,"context_lines":[{"line_number":371,"context_line":"  fixed-address {"},{"line_number":372,"context_line":"      prefix: /broadcast"},{"line_number":373,"context_line":"      fanout: multiple"},{"line_number":374,"context_line":"  }"},{"line_number":375,"context_line":""},{"line_number":376,"context_line":""},{"line_number":377,"context_line":"Appendix C: Configuring Qpidd"}],"source_content_type":"text/x-rst","patch_set":8,"id":"baada198_0e572dcb","line":374,"in_reply_to":"baada198_f8be7ffe","updated":"2014-07-03 17:47:15.000000000","message":"I do think it makes sense to have an openstack-specific namespace in the address. I like \"openstack\" for clarity, but using one of the oslo variants would make sense, too.","commit_id":"fbcd34d53782ffaf60f533955cd3861ed51279e3"}]}
