)]}'
{"reference/tags/assert_supports-rolling-upgrade.rst":[{"author":{"_account_id":2472,"name":"Doug Hellmann","email":"dhellmann@redhat.com","username":"doug-hellmann"},"change_message_id":"9cf252aea320adda0b1bf340206e13268d77dfeb","unresolved":false,"context_lines":[{"line_number":47,"context_line":"  elimination of downtime during upgrades, but rather reducing the"},{"line_number":48,"context_line":"  scope from \"all services\" to \"some services at a time.\" In other"},{"line_number":49,"context_line":"  words, \"restarting all API services together\" is a reasonable restriction. A"},{"line_number":50,"context_line":"  more restrictive tag, ``asserts:supports-zero-downtime-upgrade`` removes this"},{"line_number":51,"context_line":"  allowance."},{"line_number":52,"context_line":"* Full stack integration testing with services arranged in a"},{"line_number":53,"context_line":"  mid-upgrade manner is performed on every proposed commit to validate"}],"source_content_type":"text/x-rst","patch_set":3,"id":"5a8bc5a2_c40bc7d4","line":50,"updated":"2016-09-26 14:26:52.000000000","message":"If you need another revision of this patch for another reason, please make this a reference to the target tag-assert:supports-rolling-upgrade","commit_id":"925feee1167158ab414b64aec97c300f555c20a6"},{"author":{"_account_id":4,"name":"Dolph Mathews","email":"dolph.mathews@gmail.com","username":"dolph"},"change_message_id":"fa07ba5cf4d5e154f99df9f878ef152d4407f49f","unresolved":false,"context_lines":[{"line_number":47,"context_line":"  elimination of downtime during upgrades, but rather reducing the"},{"line_number":48,"context_line":"  scope from \"all services\" to \"some services at a time.\" In other"},{"line_number":49,"context_line":"  words, \"restarting all API services together\" is a reasonable restriction. A"},{"line_number":50,"context_line":"  more restrictive tag, ``asserts:supports-zero-downtime-upgrade`` removes this"},{"line_number":51,"context_line":"  allowance."},{"line_number":52,"context_line":"* Full stack integration testing with services arranged in a"},{"line_number":53,"context_line":"  mid-upgrade manner is performed on every proposed commit to validate"}],"source_content_type":"text/x-rst","patch_set":3,"id":"fa6399be_027939c8","line":50,"in_reply_to":"5a8bc5a2_c40bc7d4","updated":"2016-10-20 17:28:00.000000000","message":"Done","commit_id":"925feee1167158ab414b64aec97c300f555c20a6"}],"reference/tags/assert_supports-zero-downtime-upgrade.rst":[{"author":{"_account_id":2472,"name":"Doug Hellmann","email":"dhellmann@redhat.com","username":"doug-hellmann"},"change_message_id":"9cf252aea320adda0b1bf340206e13268d77dfeb","unresolved":false,"context_lines":[{"line_number":16,"context_line":"The ``assert:supports-zero-downtime-upgrade`` tag asserts that in addition to a"},{"line_number":17,"context_line":"project supporting `minimal rolling upgrade capabilities"},{"line_number":18,"context_line":"\u003chttps://governance.openstack.org/reference/tags/assert_supports-rolling-upgrade.html\u003e`_,"},{"line_number":19,"context_line":"it do so *without incurring any disruption to API availability of the control"},{"line_number":20,"context_line":"plane*."},{"line_number":21,"context_line":""},{"line_number":22,"context_line":"Application to current projects"}],"source_content_type":"text/x-rst","patch_set":3,"id":"5a8bc5a2_c432e7a6","line":19,"updated":"2016-09-26 14:26:52.000000000","message":"\"it does so\"","commit_id":"925feee1167158ab414b64aec97c300f555c20a6"},{"author":{"_account_id":4,"name":"Dolph Mathews","email":"dolph.mathews@gmail.com","username":"dolph"},"change_message_id":"fa07ba5cf4d5e154f99df9f878ef152d4407f49f","unresolved":false,"context_lines":[{"line_number":16,"context_line":"The ``assert:supports-zero-downtime-upgrade`` tag asserts that in addition to a"},{"line_number":17,"context_line":"project supporting `minimal rolling upgrade capabilities"},{"line_number":18,"context_line":"\u003chttps://governance.openstack.org/reference/tags/assert_supports-rolling-upgrade.html\u003e`_,"},{"line_number":19,"context_line":"it do so *without incurring any disruption to API availability of the control"},{"line_number":20,"context_line":"plane*."},{"line_number":21,"context_line":""},{"line_number":22,"context_line":"Application to current projects"}],"source_content_type":"text/x-rst","patch_set":3,"id":"fa6399be_824b8986","line":19,"in_reply_to":"5a8bc5a2_c432e7a6","updated":"2016-10-20 17:28:00.000000000","message":"Done","commit_id":"925feee1167158ab414b64aec97c300f555c20a6"},{"author":{"_account_id":15343,"name":"Tim Burke","email":"tburke@nvidia.com","username":"tburke"},"change_message_id":"8a01e1ae6575ded3abd896eeebecd8c67ae948d4","unresolved":false,"context_lines":[{"line_number":16,"context_line":"The ``assert:supports-zero-downtime-upgrade`` tag asserts that in addition to a"},{"line_number":17,"context_line":"project supporting `minimal rolling upgrade capabilities"},{"line_number":18,"context_line":"\u003chttps://governance.openstack.org/reference/tags/assert_supports-rolling-upgrade.html\u003e`_,"},{"line_number":19,"context_line":"it do so *without incurring any disruption to API availability of the control"},{"line_number":20,"context_line":"plane*."},{"line_number":21,"context_line":""},{"line_number":22,"context_line":"Application to current projects"},{"line_number":23,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":3,"id":"7a8ec9b2_9d47e61c","line":20,"range":{"start_line":19,"start_character":70,"end_line":20,"end_character":5},"updated":"2016-09-19 20:50:17.000000000","message":"Will this tag also apply to projects that are rather squarely in the data plane?\n\nWould it be better to talk about the \"service\" rather than \"control plane\"?","commit_id":"925feee1167158ab414b64aec97c300f555c20a6"},{"author":{"_account_id":6537,"name":"gordon chung","email":"gord@live.ca","username":"chungg"},"change_message_id":"f858f7de9573f4e686c4d7081ead1f98987ea15b","unresolved":false,"context_lines":[{"line_number":16,"context_line":"The ``assert:supports-zero-downtime-upgrade`` tag asserts that in addition to a"},{"line_number":17,"context_line":"project supporting `minimal rolling upgrade capabilities"},{"line_number":18,"context_line":"\u003chttps://governance.openstack.org/reference/tags/assert_supports-rolling-upgrade.html\u003e`_,"},{"line_number":19,"context_line":"it do so *without incurring any disruption to API availability of the control"},{"line_number":20,"context_line":"plane*."},{"line_number":21,"context_line":""},{"line_number":22,"context_line":"Application to current projects"},{"line_number":23,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":3,"id":"5a8bc5a2_77fb3048","line":20,"range":{"start_line":19,"start_character":9,"end_line":20,"end_character":5},"updated":"2016-09-27 19:54:47.000000000","message":"does anyone know why we have the unavailable API allowance in rolling upgrades? is it because it doesn\u0027t support POST?\n\nthis seems to be the one difference between the two tags but we have a lot more words :)","commit_id":"925feee1167158ab414b64aec97c300f555c20a6"},{"author":{"_account_id":4,"name":"Dolph Mathews","email":"dolph.mathews@gmail.com","username":"dolph"},"change_message_id":"fa07ba5cf4d5e154f99df9f878ef152d4407f49f","unresolved":false,"context_lines":[{"line_number":17,"context_line":"project supporting `minimal rolling upgrade capabilities"},{"line_number":18,"context_line":"\u003chttps://governance.openstack.org/reference/tags/assert_supports-rolling-upgrade.html\u003e`_,"},{"line_number":19,"context_line":"it do so *without incurring any disruption to API availability of the control"},{"line_number":20,"context_line":"plane*."},{"line_number":21,"context_line":""},{"line_number":22,"context_line":"Application to current projects"},{"line_number":23,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":3,"id":"fa6399be_55b4cd33","line":20,"in_reply_to":"5a8bc5a2_3201b63a","updated":"2016-10-20 17:28:00.000000000","message":"I like Tim\u0027s simplification here, and will be including it in the next patch. The intent was certainly not to allow an outage in the data plane or to apply to a subset of services, but rather to clarify \"API availability\" -- but I think it just made things more confusing.","commit_id":"925feee1167158ab414b64aec97c300f555c20a6"},{"author":{"_account_id":782,"name":"John Garbutt","email":"john@johngarbutt.com","username":"johngarbutt"},"change_message_id":"f4f5da9ba9bdfbb1f7314d49cfb2e20c9c723bc7","unresolved":false,"context_lines":[{"line_number":16,"context_line":"The ``assert:supports-zero-downtime-upgrade`` tag asserts that in addition to a"},{"line_number":17,"context_line":"project supporting `minimal rolling upgrade capabilities"},{"line_number":18,"context_line":"\u003chttps://governance.openstack.org/reference/tags/assert_supports-rolling-upgrade.html\u003e`_,"},{"line_number":19,"context_line":"it do so *without incurring any disruption to API availability of the control"},{"line_number":20,"context_line":"plane*."},{"line_number":21,"context_line":""},{"line_number":22,"context_line":"Application to current projects"},{"line_number":23,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":3,"id":"5a8bc5a2_3201b63a","line":20,"range":{"start_line":19,"start_character":9,"end_line":20,"end_character":5},"in_reply_to":"5a8bc5a2_77fb3048","updated":"2016-09-27 20:14:47.000000000","message":"Rolling upgrade was a very Nova view of the world. The idea was to ensure the control plane quickly gets upgraded via restart in a few seconds, but the long process of gracefully shutting down and restarting each of 100s/1000s of workers can be done without API downtime.\n\nRolling upgrade assumes no data plane outage (at least in my head, I should double check), so I see why there is a distinction here. But we could instead say no control plane or data plane outage.","commit_id":"925feee1167158ab414b64aec97c300f555c20a6"},{"author":{"_account_id":308,"name":"Thierry Carrez","email":"thierry@openstack.org","username":"ttx"},"change_message_id":"74b5b622f34463672ca2290d0c9cee5c68b13400","unresolved":false,"context_lines":[{"line_number":16,"context_line":"The ``assert:supports-zero-downtime-upgrade`` tag asserts that in addition to a"},{"line_number":17,"context_line":"project supporting `minimal rolling upgrade capabilities"},{"line_number":18,"context_line":"\u003chttps://governance.openstack.org/reference/tags/assert_supports-rolling-upgrade.html\u003e`_,"},{"line_number":19,"context_line":"it do so *without incurring any disruption to API availability of the control"},{"line_number":20,"context_line":"plane*."},{"line_number":21,"context_line":""},{"line_number":22,"context_line":"Application to current projects"},{"line_number":23,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":3,"id":"5a8bc5a2_a9574124","line":20,"range":{"start_line":19,"start_character":70,"end_line":20,"end_character":5},"in_reply_to":"7a8ec9b2_84be71fd","updated":"2016-09-27 09:41:29.000000000","message":"Agreed, not sure what the \"control plane\" distinction buys us here.","commit_id":"925feee1167158ab414b64aec97c300f555c20a6"},{"author":{"_account_id":20680,"name":"Krish Raghuram","email":"krishnan.raghuram@intel.com","username":"kraghuram"},"change_message_id":"66ed8a89d155440006cf68fc93bcdcb5021eb27a","unresolved":false,"context_lines":[{"line_number":16,"context_line":"The ``assert:supports-zero-downtime-upgrade`` tag asserts that in addition to a"},{"line_number":17,"context_line":"project supporting `minimal rolling upgrade capabilities"},{"line_number":18,"context_line":"\u003chttps://governance.openstack.org/reference/tags/assert_supports-rolling-upgrade.html\u003e`_,"},{"line_number":19,"context_line":"it do so *without incurring any disruption to API availability of the control"},{"line_number":20,"context_line":"plane*."},{"line_number":21,"context_line":""},{"line_number":22,"context_line":"Application to current projects"},{"line_number":23,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":3,"id":"7a8ec9b2_b862bb3b","line":20,"range":{"start_line":19,"start_character":70,"end_line":20,"end_character":5},"in_reply_to":"7a8ec9b2_9d47e61c","updated":"2016-09-20 16:01:33.000000000","message":"Any examples of projects purely in the data plane? Won\u0027t any service be called through an API, in which case the availability of the API becomes the matter being addressed, not the uptime of existing service sessions during the upgrade? The latter may require a \"zero data plane downtime\" tag","commit_id":"925feee1167158ab414b64aec97c300f555c20a6"},{"author":{"_account_id":15343,"name":"Tim Burke","email":"tburke@nvidia.com","username":"tburke"},"change_message_id":"b98e288edb92974d8c35f1e9c1b3c5fe54c80edd","unresolved":false,"context_lines":[{"line_number":16,"context_line":"The ``assert:supports-zero-downtime-upgrade`` tag asserts that in addition to a"},{"line_number":17,"context_line":"project supporting `minimal rolling upgrade capabilities"},{"line_number":18,"context_line":"\u003chttps://governance.openstack.org/reference/tags/assert_supports-rolling-upgrade.html\u003e`_,"},{"line_number":19,"context_line":"it do so *without incurring any disruption to API availability of the control"},{"line_number":20,"context_line":"plane*."},{"line_number":21,"context_line":""},{"line_number":22,"context_line":"Application to current projects"},{"line_number":23,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":3,"id":"7a8ec9b2_84be71fd","line":20,"range":{"start_line":19,"start_character":70,"end_line":20,"end_character":5},"in_reply_to":"7a8ec9b2_b862bb3b","updated":"2016-09-20 18:16:24.000000000","message":"I was thinking of Swift in particular, since that\u0027s what I\u0027m most familiar with. Others that seem more \"data plane\" than \"control plane\" include Glance, Designate, Gnocchi, and Zaqar.\n\nYes, all of these are called through an API, but it\u0027s generally a very data-oriented API. Would it be acceptable for Designate to claim this tag if the control-oriented APIs (pushing records to DNS servers, say) were available during an upgrade while the data-oriented APIs (actually creating records, for example) weren\u0027t?\n\nWould it be better to just say something like this?\n\n\u003e The assert:supports-zero-downtime-upgrade tag asserts that in addition to a project supporting minimal rolling upgrade capabilities, it does so without incurring any disruption to API availability.\n\nWith similar edits in the sections below:\n\n\u003e Operators need to know what projects support an upgrade process that eliminates downtime of the API entirely.\n\n\u003e ... this tag requires services to completely eliminate API downtime during the upgrade.\n\n\u003e While all requests to the API must be eventually satisfied, ...","commit_id":"925feee1167158ab414b64aec97c300f555c20a6"},{"author":{"_account_id":16157,"name":"Ken Johnston","email":"kenny@kencjohnston.com","username":"kencjohnston"},"change_message_id":"65473dab6c2a1ebb02d041f14aead757f5d23dae","unresolved":false,"context_lines":[{"line_number":47,"context_line":"  tag which allows for minimal downtime, this tag requires services to"},{"line_number":48,"context_line":"  completely eliminate API downtime of the control plane during the upgrade. In"},{"line_number":49,"context_line":"  other words, requiring operators to \"restart all API services together\" is"},{"line_number":50,"context_line":"  not reasonable under this tag. Additionally, it is not reasonable for"},{"line_number":51,"context_line":"  services to respond to requests with HTTP 5xx codes when they otherwise would"},{"line_number":52,"context_line":"  have returned successful responses, even intermittently, during the upgrade."},{"line_number":53,"context_line":""},{"line_number":54,"context_line":"* While all requests to the control plane must be eventually satisfied,"},{"line_number":55,"context_line":"  performance degradation during the upgrade is acceptable. This may include"}],"source_content_type":"text/x-rst","patch_set":3,"id":"7a8ec9b2_1d867665","line":52,"range":{"start_line":50,"start_character":33,"end_line":52,"end_character":77},"updated":"2016-09-19 20:39:18.000000000","message":"I\u0027m not sure what action \"during the upgrade\" was referring to. Suggestion:\n\"Additionally, it is not reasonable for services to respond to requests with HTTP 5xx codes during an upgrade when they otherwise would have returned successful responses, even intermittently.\"","commit_id":"925feee1167158ab414b64aec97c300f555c20a6"},{"author":{"_account_id":4,"name":"Dolph Mathews","email":"dolph.mathews@gmail.com","username":"dolph"},"change_message_id":"fa07ba5cf4d5e154f99df9f878ef152d4407f49f","unresolved":false,"context_lines":[{"line_number":49,"context_line":"  other words, requiring operators to \"restart all API services together\" is"},{"line_number":50,"context_line":"  not reasonable under this tag. Additionally, it is not reasonable for"},{"line_number":51,"context_line":"  services to respond to requests with HTTP 5xx codes when they otherwise would"},{"line_number":52,"context_line":"  have returned successful responses, even intermittently, during the upgrade."},{"line_number":53,"context_line":""},{"line_number":54,"context_line":"* While all requests to the control plane must be eventually satisfied,"},{"line_number":55,"context_line":"  performance degradation during the upgrade is acceptable. This may include"}],"source_content_type":"text/x-rst","patch_set":3,"id":"fa6399be_d5f53ded","line":52,"in_reply_to":"7a8ec9b2_1d867665","updated":"2016-10-20 17:28:00.000000000","message":"++","commit_id":"925feee1167158ab414b64aec97c300f555c20a6"},{"author":{"_account_id":6537,"name":"gordon chung","email":"gord@live.ca","username":"chungg"},"change_message_id":"f858f7de9573f4e686c4d7081ead1f98987ea15b","unresolved":false,"context_lines":[{"line_number":52,"context_line":"  have returned successful responses, even intermittently, during the upgrade."},{"line_number":53,"context_line":""},{"line_number":54,"context_line":"* While all requests to the control plane must be eventually satisfied,"},{"line_number":55,"context_line":"  performance degradation during the upgrade is acceptable. This may include"},{"line_number":56,"context_line":"  slow HTTP responses and delayed request handling (such as messages queuing up"},{"line_number":57,"context_line":"  on the message bus)."},{"line_number":58,"context_line":""},{"line_number":59,"context_line":"* In addition to the full stack integration testing required by the"},{"line_number":60,"context_line":"  ``assert:supports-rolling-upgrade`` tag, services should be capable of"}],"source_content_type":"text/x-rst","patch_set":3,"id":"5a8bc5a2_b7613814","line":57,"range":{"start_line":55,"start_character":60,"end_line":57,"end_character":22},"updated":"2016-09-27 19:54:47.000000000","message":"is this a hypothetical or is this a byproduct of your solution? just curious as to why this allowance is required.","commit_id":"925feee1167158ab414b64aec97c300f555c20a6"},{"author":{"_account_id":4,"name":"Dolph Mathews","email":"dolph.mathews@gmail.com","username":"dolph"},"change_message_id":"fa07ba5cf4d5e154f99df9f878ef152d4407f49f","unresolved":false,"context_lines":[{"line_number":54,"context_line":"* While all requests to the control plane must be eventually satisfied,"},{"line_number":55,"context_line":"  performance degradation during the upgrade is acceptable. This may include"},{"line_number":56,"context_line":"  slow HTTP responses and delayed request handling (such as messages queuing up"},{"line_number":57,"context_line":"  on the message bus)."},{"line_number":58,"context_line":""},{"line_number":59,"context_line":"* In addition to the full stack integration testing required by the"},{"line_number":60,"context_line":"  ``assert:supports-rolling-upgrade`` tag, services should be capable of"}],"source_content_type":"text/x-rst","patch_set":3,"id":"fa6399be_7538f1a0","line":57,"in_reply_to":"5a8bc5a2_b7613814","updated":"2016-10-20 17:28:00.000000000","message":"Both. It\u0027s possible to do zero-downtime rolling upgrades while incurring a huge performance penalty because the database is getting overloaded with data migration operations for example (i.e. lots of writes). In my world view, asserting that there is no perceivable performance penalty to upgrading is a slightly stricter assertion (and the next step). So a subsequent tag would eliminate this allowance (i.e. \"assert-zero-impact-upgrades\", or something along those lines).","commit_id":"925feee1167158ab414b64aec97c300f555c20a6"},{"author":{"_account_id":16157,"name":"Ken Johnston","email":"kenny@kencjohnston.com","username":"kencjohnston"},"change_message_id":"65473dab6c2a1ebb02d041f14aead757f5d23dae","unresolved":false,"context_lines":[{"line_number":60,"context_line":"  ``assert:supports-rolling-upgrade`` tag, services should be capable of"},{"line_number":61,"context_line":"  receiving and handling requests throughout the rolling upgrade process with a"},{"line_number":62,"context_line":"  normal success rate."},{"line_number":63,"context_line":""},{"line_number":64,"context_line":"Tag application process"},{"line_number":65,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":66,"context_line":""}],"source_content_type":"text/x-rst","patch_set":3,"id":"7a8ec9b2_ddba4e0a","line":63,"updated":"2016-09-19 20:39:18.000000000","message":"Isn\u0027t it important that projects gate on this assertion? Add:\n\n* In order to assert this tag, services should prevent regression by implementing a zero-downtime upgrade gate check.","commit_id":"925feee1167158ab414b64aec97c300f555c20a6"},{"author":{"_account_id":782,"name":"John Garbutt","email":"john@johngarbutt.com","username":"johngarbutt"},"change_message_id":"950df2508bb2e50f0bf72e7de8a05b3cb075031a","unresolved":false,"context_lines":[{"line_number":60,"context_line":"  ``assert:supports-rolling-upgrade`` tag, services should be capable of"},{"line_number":61,"context_line":"  receiving and handling requests throughout the rolling upgrade process with a"},{"line_number":62,"context_line":"  normal success rate."},{"line_number":63,"context_line":""},{"line_number":64,"context_line":"Tag application process"},{"line_number":65,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":66,"context_line":""}],"source_content_type":"text/x-rst","patch_set":3,"id":"5a8bc5a2_a947184a","line":63,"in_reply_to":"5a8bc5a2_296c1107","updated":"2016-09-27 20:47:56.000000000","message":"Yeah, lets be clear how to validate this.\n\nI think its upstream gate running old and new APIs side by side, and testing both are working, or something a bit like that.","commit_id":"925feee1167158ab414b64aec97c300f555c20a6"},{"author":{"_account_id":308,"name":"Thierry Carrez","email":"thierry@openstack.org","username":"ttx"},"change_message_id":"74b5b622f34463672ca2290d0c9cee5c68b13400","unresolved":false,"context_lines":[{"line_number":60,"context_line":"  ``assert:supports-rolling-upgrade`` tag, services should be capable of"},{"line_number":61,"context_line":"  receiving and handling requests throughout the rolling upgrade process with a"},{"line_number":62,"context_line":"  normal success rate."},{"line_number":63,"context_line":""},{"line_number":64,"context_line":"Tag application process"},{"line_number":65,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":66,"context_line":""}],"source_content_type":"text/x-rst","patch_set":3,"id":"5a8bc5a2_296c1107","line":63,"in_reply_to":"5a8bc5a2_93511939","updated":"2016-09-27 09:41:29.000000000","message":"+1","commit_id":"925feee1167158ab414b64aec97c300f555c20a6"},{"author":{"_account_id":4,"name":"Dolph Mathews","email":"dolph.mathews@gmail.com","username":"dolph"},"change_message_id":"fa07ba5cf4d5e154f99df9f878ef152d4407f49f","unresolved":false,"context_lines":[{"line_number":60,"context_line":"  ``assert:supports-rolling-upgrade`` tag, services should be capable of"},{"line_number":61,"context_line":"  receiving and handling requests throughout the rolling upgrade process with a"},{"line_number":62,"context_line":"  normal success rate."},{"line_number":63,"context_line":""},{"line_number":64,"context_line":"Tag application process"},{"line_number":65,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":66,"context_line":""}],"source_content_type":"text/x-rst","patch_set":3,"id":"fa6399be_151fb508","line":63,"in_reply_to":"5a8bc5a2_a947184a","updated":"2016-10-20 17:28:00.000000000","message":"++ I\u0027ve added this to the bullet on L59 in the next patchset, using the word \"concurrently\" instead of side-by-side (which I think could be interpreted more loosely).","commit_id":"925feee1167158ab414b64aec97c300f555c20a6"},{"author":{"_account_id":2750,"name":"Sean Dague","email":"sean@dague.net","username":"sdague"},"change_message_id":"9f5d5ccecdb19d7654c9807e4722f145b1908702","unresolved":false,"context_lines":[{"line_number":60,"context_line":"  ``assert:supports-rolling-upgrade`` tag, services should be capable of"},{"line_number":61,"context_line":"  receiving and handling requests throughout the rolling upgrade process with a"},{"line_number":62,"context_line":"  normal success rate."},{"line_number":63,"context_line":""},{"line_number":64,"context_line":"Tag application process"},{"line_number":65,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":66,"context_line":""}],"source_content_type":"text/x-rst","patch_set":3,"id":"5a8bc5a2_93511939","line":63,"in_reply_to":"7a8ec9b2_ddba4e0a","updated":"2016-09-26 10:20:03.000000000","message":"I think this is a key point. All the other elements here have included some kind of upstream validation. We should make sure there is upstream validation here as well.","commit_id":"925feee1167158ab414b64aec97c300f555c20a6"},{"author":{"_account_id":2472,"name":"Doug Hellmann","email":"dhellmann@redhat.com","username":"doug-hellmann"},"change_message_id":"78a781c1658f1f414e7fc863f5a5103364181aa3","unresolved":false,"context_lines":[{"line_number":52,"context_line":"  HTTP 5xx codes when they otherwise would have returned successful responses,"},{"line_number":53,"context_line":"  even intermittently."},{"line_number":54,"context_line":""},{"line_number":55,"context_line":"* While all requests to the control plane must be eventually satisfied,"},{"line_number":56,"context_line":"  performance degradation during the upgrade is acceptable. This may include"},{"line_number":57,"context_line":"  slow HTTP responses and delayed request handling (such as messages queuing up"},{"line_number":58,"context_line":"  on the message bus)."}],"source_content_type":"text/x-rst","patch_set":5,"id":"ba5da102_1d0285b4","line":55,"range":{"start_line":55,"start_character":61,"end_line":55,"end_character":70},"updated":"2016-11-01 18:25:56.000000000","message":"The word \"processed\" instead of \"satisfied\" eliminates the suggestion that the service must handle the request with what the user would consider \"success\" even if the accurate behavior is to produce an error.\n\nIt seems pretty likely that an upgrade could introduce conditions that make it impossible to \"satisfy\" some of the requests, especially if we allow degraded performance. For example, maybe during an upgrade a bunch of requests are queued up that would exceed the user\u0027s quota. Since none are actually being processed, the situation may not be recognized so that the requests are rejected as part of the API call itself, but the resources cannot be allocated when the upgrade is finished.","commit_id":"1133c16c413765df3aabc08e40fda6f17cf46d1b"},{"author":{"_account_id":4,"name":"Dolph Mathews","email":"dolph.mathews@gmail.com","username":"dolph"},"change_message_id":"b21b4f20dbe418d6cee37df3911fe3b7342d9456","unresolved":false,"context_lines":[{"line_number":52,"context_line":"  HTTP 5xx codes when they otherwise would have returned successful responses,"},{"line_number":53,"context_line":"  even intermittently."},{"line_number":54,"context_line":""},{"line_number":55,"context_line":"* While all requests to the control plane must be eventually satisfied,"},{"line_number":56,"context_line":"  performance degradation during the upgrade is acceptable. This may include"},{"line_number":57,"context_line":"  slow HTTP responses and delayed request handling (such as messages queuing up"},{"line_number":58,"context_line":"  on the message bus)."}],"source_content_type":"text/x-rst","patch_set":5,"id":"9a629dbe_4069b224","line":55,"in_reply_to":"ba5da102_1d0285b4","updated":"2016-11-08 19:52:18.000000000","message":"+1 on \"processed\"; that better reflects my intent.\n\nI also agree completely about the validity of the edge case (and I\u0027m sure there will be many more such examples), but I don\u0027t find it to be too concerning (at least within this the context of this tag).","commit_id":"1133c16c413765df3aabc08e40fda6f17cf46d1b"},{"author":{"_account_id":3153,"name":"Emilien Macchi","email":"emilien@redhat.com","username":"emilienm"},"change_message_id":"e1ab6342c90477eed77c2cf981cb3de4e6c63c12","unresolved":false,"context_lines":[{"line_number":63,"context_line":"  normal success rate. In order to assert this tag, services should prevent"},{"line_number":64,"context_line":"  regression by implementing a zero-downtime gate job wherein both a new"},{"line_number":65,"context_line":"  version of the service and an old version of the service are run"},{"line_number":66,"context_line":"  concurrently."},{"line_number":67,"context_line":""},{"line_number":68,"context_line":"Tag application process"},{"line_number":69,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":5,"id":"da6895a0_2d26c516","line":66,"updated":"2016-10-20 20:34:06.000000000","message":"Excellent.  It makes me thing we\u0027ll need to find a tool (or does it already exist?) that allows us to keep running tests (with tempest probably) during the actual upgrade process done by Grenade.","commit_id":"1133c16c413765df3aabc08e40fda6f17cf46d1b"},{"author":{"_account_id":970,"name":"Dean Troyer","email":"dtroyer@gmail.com","username":"dtroyer"},"change_message_id":"b790f31758710120f52d40110f6564fa405285c9","unresolved":false,"context_lines":[{"line_number":63,"context_line":"  normal success rate. In order to assert this tag, services should prevent"},{"line_number":64,"context_line":"  regression by implementing a zero-downtime gate job wherein both a new"},{"line_number":65,"context_line":"  version of the service and an old version of the service are run"},{"line_number":66,"context_line":"  concurrently."},{"line_number":67,"context_line":""},{"line_number":68,"context_line":"Tag application process"},{"line_number":69,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":5,"id":"9a629dbe_1d58fc31","line":66,"in_reply_to":"9a629dbe_0b0d1b37","updated":"2016-11-08 20:37:03.000000000","message":"FWIW, if we are not already, we\u0027re getting really close to the point that Grenade is really not the right testing driver for multi-node/multi-\u003cwhatever\u003e upgrade testing...","commit_id":"1133c16c413765df3aabc08e40fda6f17cf46d1b"},{"author":{"_account_id":4,"name":"Dolph Mathews","email":"dolph.mathews@gmail.com","username":"dolph"},"change_message_id":"b21b4f20dbe418d6cee37df3911fe3b7342d9456","unresolved":false,"context_lines":[{"line_number":63,"context_line":"  normal success rate. In order to assert this tag, services should prevent"},{"line_number":64,"context_line":"  regression by implementing a zero-downtime gate job wherein both a new"},{"line_number":65,"context_line":"  version of the service and an old version of the service are run"},{"line_number":66,"context_line":"  concurrently."},{"line_number":67,"context_line":""},{"line_number":68,"context_line":"Tag application process"},{"line_number":69,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":5,"id":"9a629dbe_0b0d1b37","line":66,"in_reply_to":"da6895a0_2d26c516","updated":"2016-11-08 19:52:18.000000000","message":"+1! There are several folks in the community evaluating different approaches to solve this problem. I think grenade + some form of continuous tempest is the most obvious solution given where we are today, but I don\u0027t want to preclude an opportunity for something innovative here.","commit_id":"1133c16c413765df3aabc08e40fda6f17cf46d1b"}]}
