)]}'
{"reference/tags/assert_supports-zero-impact-upgrade.rst":[{"author":{"_account_id":4,"name":"Dolph Mathews","email":"dolph.mathews@gmail.com","username":"dolph"},"change_message_id":"3ba2c037eb34e92a385b51e0f9597ef31e59f52d","unresolved":false,"context_lines":[{"line_number":61,"context_line":"  wherein both a new version of the service and an old version of the service"},{"line_number":62,"context_line":"  are run concurrently under load. *A measurement of API response times must"},{"line_number":63,"context_line":"  show that there are no statistically significant outliers during the upgrade"},{"line_number":64,"context_line":"  process when compared to normal operations.*"},{"line_number":65,"context_line":""},{"line_number":66,"context_line":"Tag application process"},{"line_number":67,"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":1,"id":"fa6399be_584404b6","line":64,"updated":"2016-10-20 18:23:51.000000000","message":"Basically this requires that we run a benchmark before the upgrade process, compute a statistically significant increase in response time based on the results (essentially a timeout threshold), and then run the same benchmark again during the upgrade process.","commit_id":"ce48485756ea9b178aae7509e0becfea2675b144"},{"author":{"_account_id":4,"name":"Dolph Mathews","email":"dolph.mathews@gmail.com","username":"dolph"},"change_message_id":"45db803866b12d3f7db6f48d80e0df38558118ff","unresolved":false,"context_lines":[{"line_number":61,"context_line":"  wherein both a new version of the service and an old version of the service"},{"line_number":62,"context_line":"  are run concurrently under load. *A measurement of API response times must"},{"line_number":63,"context_line":"  show that there are no statistically significant outliers during the upgrade"},{"line_number":64,"context_line":"  process when compared to normal operations.*"},{"line_number":65,"context_line":""},{"line_number":66,"context_line":"Tag application process"},{"line_number":67,"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":1,"id":"ba5da102_cc1c6f7a","line":64,"in_reply_to":"da6895a0_8d4b51b0","updated":"2016-10-26 09:18:57.000000000","message":"I certainly don\u0027t think the tag should make assertions about the testing tool, but I think rally is an obvious contender to help us satisfy this requirement.","commit_id":"ce48485756ea9b178aae7509e0becfea2675b144"},{"author":{"_account_id":3153,"name":"Emilien Macchi","email":"emilien@redhat.com","username":"emilienm"},"change_message_id":"0eb1e58419d7f213841a65330ee615e7f4a927c0","unresolved":false,"context_lines":[{"line_number":61,"context_line":"  wherein both a new version of the service and an old version of the service"},{"line_number":62,"context_line":"  are run concurrently under load. *A measurement of API response times must"},{"line_number":63,"context_line":"  show that there are no statistically significant outliers during the upgrade"},{"line_number":64,"context_line":"  process when compared to normal operations.*"},{"line_number":65,"context_line":""},{"line_number":66,"context_line":"Tag application process"},{"line_number":67,"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":1,"id":"da6895a0_8d4b51b0","line":64,"in_reply_to":"fa6399be_584404b6","updated":"2016-10-20 20:35:32.000000000","message":"I\u0027m curious about which tooling will be used for benchmark. Are we going to investigate Rally?","commit_id":"ce48485756ea9b178aae7509e0becfea2675b144"}]}
