)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":12299,"name":"Paweł Koniszewski","email":"pawel.koniszewski@gmail.com","username":"pawel.koniszewski"},"change_message_id":"145853e1fe7abb07f9f88e116a290f3f1a2ce8dc","unresolved":false,"context_lines":[{"line_number":4,"context_line":"Commit:     Eli Qiao \u003cliyong.qiao@intel.com\u003e"},{"line_number":5,"context_line":"CommitDate: 2015-11-23 14:23:09 +0800"},{"line_number":6,"context_line":""},{"line_number":7,"context_line":"support auto coverge for live migration"},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"This spec describes the blueprint live-migrate-support-auto-converge"},{"line_number":10,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"ba8a016a_6ee05397","line":7,"updated":"2015-11-23 08:33:12.000000000","message":"It is already supported by nova. You probably want to change this title.","commit_id":"ad52cac26749f2f7bdf440007f3fefc5daafd8e3"}],"specs/mitaka/approved/live-migration-support-auto-converge.rst":[{"author":{"_account_id":12175,"name":"Eli Qiao","email":"qiaoliyong@gmail.com","username":"Eli"},"change_message_id":"19f3dab0576ece1f571f249894155b1a70cfe387","unresolved":false,"context_lines":[{"line_number":15,"context_line":"increase success chance of migration with busy enterprise workloads on their"},{"line_number":16,"context_line":"instance."},{"line_number":17,"context_line":""},{"line_number":18,"context_line":"This spec proposes a new opition auto-converge for live migration."},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"Problem description"},{"line_number":21,"context_line":"\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":2,"id":"ba8a016a_6a804247","line":18,"updated":"2015-11-23 06:22:24.000000000","message":"this need to be specified libvirt version.","commit_id":"62b141dc09808877883e6420f54632cb7353f345"},{"author":{"_account_id":11189,"name":"yuntongjin","email":"yuntong.jin@intel.com","username":"yuntong"},"change_message_id":"b65748b70c57847120e3140d832cb963795a871f","unresolved":false,"context_lines":[{"line_number":48,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":49,"context_line":""},{"line_number":50,"context_line":"A new option auto-converge will be added in migrate_server API. And it will be"},{"line_number":51,"context_line":"passed to libvirt/driver _live_migration_operation, then libvirt will enable"},{"line_number":52,"context_line":"the auto-converge capability of hypervisor."},{"line_number":53,"context_line":""},{"line_number":54,"context_line":"For low version libvirt which does not support auto-converge, the operator"}],"source_content_type":"text/x-rst","patch_set":2,"id":"ba8a016a_2f3c8c2e","line":51,"updated":"2015-11-23 05:04:36.000000000","message":"specifically, will add VIR_MIGRATE_AUTO_CONVERGE to libvirt migration API.","commit_id":"62b141dc09808877883e6420f54632cb7353f345"},{"author":{"_account_id":11189,"name":"yuntongjin","email":"yuntong.jin@intel.com","username":"yuntong"},"change_message_id":"1bdbea7e417bd9a4501a6e3af53b78b220078f10","unresolved":false,"context_lines":[{"line_number":54,"context_line":"For low version libvirt which does not support auto-converge, the operator"},{"line_number":55,"context_line":"still want auto-converge to help to complete live migration successfully when"},{"line_number":56,"context_line":"VM with heavy workloads. So an extended auto-converge command will be supported"},{"line_number":57,"context_line":"by libvirt qemu-monitor-command."},{"line_number":58,"context_line":""},{"line_number":59,"context_line":"Alternatives"},{"line_number":60,"context_line":"------------"}],"source_content_type":"text/x-rst","patch_set":2,"id":"ba8a016a_94302b09","line":57,"updated":"2015-11-23 04:16:25.000000000","message":"force convergence was added in qemu (version 1.6) to allow a user to choose if they wish to force convergence of their migration via a\nnew migration capability : \"auto-converge\". This feature allows for qemu to auto-detect lack of convergence and trigger a throttle-down of the VCPUs.","commit_id":"62b141dc09808877883e6420f54632cb7353f345"},{"author":{"_account_id":11189,"name":"yuntongjin","email":"yuntong.jin@intel.com","username":"yuntong"},"change_message_id":"d9eb160f2eedbe08b95b075deacadaab882795a5","unresolved":false,"context_lines":[{"line_number":60,"context_line":"------------"},{"line_number":61,"context_line":""},{"line_number":62,"context_line":"Alternative is not doing this and let operator enable auto-converge manually"},{"line_number":63,"context_line":"through hypervisor."},{"line_number":64,"context_line":""},{"line_number":65,"context_line":"Add a new auto-converge action API to enable hypervisor auto-converge"},{"line_number":66,"context_line":"capability."}],"source_content_type":"text/x-rst","patch_set":2,"id":"ba8a016a_2aff9a35","line":63,"updated":"2015-11-23 06:21:14.000000000","message":"operator can add VIR_MIGRATE_AUTO_CONVERGE to live_migration_flag in  nova.conf, but operator will loose the flexibility.","commit_id":"62b141dc09808877883e6420f54632cb7353f345"},{"author":{"_account_id":12175,"name":"Eli Qiao","email":"qiaoliyong@gmail.com","username":"Eli"},"change_message_id":"19f3dab0576ece1f571f249894155b1a70cfe387","unresolved":false,"context_lines":[{"line_number":83,"context_line":"            \u0027properties\u0027: {"},{"line_number":84,"context_line":"                \u0027block_migration\u0027: parameter_types.boolean,"},{"line_number":85,"context_line":"                \u0027disk_over_commit\u0027: parameter_types.boolean,"},{"line_number":86,"context_line":"                \u0027auto_converge\u0027: parameter_types.boolean,"},{"line_number":87,"context_line":"                \u0027host\u0027: host"},{"line_number":88,"context_line":"            },"},{"line_number":89,"context_line":"            \u0027additionalProperties\u0027: False,"}],"source_content_type":"text/x-rst","patch_set":2,"id":"ba8a016a_8acfa6b2","line":86,"updated":"2015-11-23 06:22:24.000000000","message":"Give a default value of auto_converge: it should be False","commit_id":"62b141dc09808877883e6420f54632cb7353f345"},{"author":{"_account_id":11189,"name":"yuntongjin","email":"yuntong.jin@intel.com","username":"yuntong"},"change_message_id":"70d2a9a15c1789e48709ae2f4b01e29a7b3b4ffb","unresolved":false,"context_lines":[{"line_number":143,"context_line":""},{"line_number":144,"context_line":"* A new option auto-converge is added in migrate_server API."},{"line_number":145,"context_line":"* An extended auto-converge command through libvirt qemu-monitor-command"},{"line_number":146,"context_line":"  is added."},{"line_number":147,"context_line":""},{"line_number":148,"context_line":"Dependencies"},{"line_number":149,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":2,"id":"ba8a016a_2a3ebab8","line":146,"updated":"2015-11-23 05:31:50.000000000","message":"If it\u0027s not supported in libvirt API, it won\u0027t be supported in nova.","commit_id":"62b141dc09808877883e6420f54632cb7353f345"},{"author":{"_account_id":12299,"name":"Paweł Koniszewski","email":"pawel.koniszewski@gmail.com","username":"pawel.koniszewski"},"change_message_id":"145853e1fe7abb07f9f88e116a290f3f1a2ce8dc","unresolved":false,"context_lines":[{"line_number":5,"context_line":" http://creativecommons.org/licenses/by/3.0/legalcode"},{"line_number":6,"context_line":""},{"line_number":7,"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\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":8,"context_line":"support auto coverge for live migration"},{"line_number":9,"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\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":10,"context_line":""},{"line_number":11,"context_line":"Blueprint:"}],"source_content_type":"text/x-rst","patch_set":3,"id":"ba8a016a_519f2e5f","line":8,"updated":"2015-11-23 08:33:12.000000000","message":"ditto, it is already supported, you probably want to change title.","commit_id":"ad52cac26749f2f7bdf440007f3fefc5daafd8e3"},{"author":{"_account_id":12299,"name":"Paweł Koniszewski","email":"pawel.koniszewski@gmail.com","username":"pawel.koniszewski"},"change_message_id":"145853e1fe7abb07f9f88e116a290f3f1a2ce8dc","unresolved":false,"context_lines":[{"line_number":31,"context_line":"https://review.openstack.org/#/c/245543"},{"line_number":32,"context_line":""},{"line_number":33,"context_line":"But in another cases, the user want to keep the their workloads on their vm"},{"line_number":34,"context_line":"uninterrupted, they need a new mechenical auto-converge to satisfy their"},{"line_number":35,"context_line":"requirement."},{"line_number":36,"context_line":""},{"line_number":37,"context_line":""}],"source_content_type":"text/x-rst","patch_set":3,"id":"ba8a016a_31bdcad7","line":34,"updated":"2015-11-23 08:33:12.000000000","message":"It\u0027s worth mentioning that performance of a VM will be affected.","commit_id":"ad52cac26749f2f7bdf440007f3fefc5daafd8e3"},{"author":{"_account_id":12299,"name":"Paweł Koniszewski","email":"pawel.koniszewski@gmail.com","username":"pawel.koniszewski"},"change_message_id":"145853e1fe7abb07f9f88e116a290f3f1a2ce8dc","unresolved":false,"context_lines":[{"line_number":45,"context_line":"with it\u0027s help."},{"line_number":46,"context_line":""},{"line_number":47,"context_line":"For some other cases, such as the workloads in the VM we want to migrate is"},{"line_number":48,"context_line":"important, and we don\u0027t want to decrease vcpu, we can disable auto-converge."},{"line_number":49,"context_line":""},{"line_number":50,"context_line":"Proposed change"},{"line_number":51,"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":3,"id":"ba8a016a_1174469a","line":48,"updated":"2015-11-23 08:33:12.000000000","message":"Auto-converge will throttle down memory writes only when there is no progress in live migration so probably we don\u0027t want to turn it off. Otherwise you might end up with never-ending live migration which is super bad for infrastructure.","commit_id":"ad52cac26749f2f7bdf440007f3fefc5daafd8e3"},{"author":{"_account_id":1779,"name":"Daniel Berrange","email":"berrange@redhat.com","username":"berrange"},"change_message_id":"c6c02a76a453366eef7f705c9effde66e116f5e9","unresolved":false,"context_lines":[{"line_number":52,"context_line":""},{"line_number":53,"context_line":"A new option auto-converge will be added in migrate_server API. And it will be"},{"line_number":54,"context_line":"passed to libvirt/driver _live_migration_operation, then libvirt will enable"},{"line_number":55,"context_line":"the auto-converge capability of hypervisor."},{"line_number":56,"context_line":""},{"line_number":57,"context_line":"For lower version libvirt(\u003c1.2.3) which does not support auto-converge, if the"},{"line_number":58,"context_line":"operator still want auto-converge to help to complete live migration"}],"source_content_type":"text/x-rst","patch_set":3,"id":"ba8a016a_e5a45909","line":55,"updated":"2015-11-23 10:48:34.000000000","message":"The auto-converge concept is again a QEMU specific implementation detail that has no business being exposed in the public Nova API. Nova itself should decide if \u0026 when it is appropriate to use it.","commit_id":"ad52cac26749f2f7bdf440007f3fefc5daafd8e3"},{"author":{"_account_id":14131,"name":"shaohef","email":"shaohe.feng@intel.com","username":"shaohefeng"},"change_message_id":"82cf1c206927e35adeb5be519e1e81cab6b7d704","unresolved":false,"context_lines":[{"line_number":52,"context_line":""},{"line_number":53,"context_line":"A new option auto-converge will be added in migrate_server API. And it will be"},{"line_number":54,"context_line":"passed to libvirt/driver _live_migration_operation, then libvirt will enable"},{"line_number":55,"context_line":"the auto-converge capability of hypervisor."},{"line_number":56,"context_line":""},{"line_number":57,"context_line":"For lower version libvirt(\u003c1.2.3) which does not support auto-converge, if the"},{"line_number":58,"context_line":"operator still want auto-converge to help to complete live migration"}],"source_content_type":"text/x-rst","patch_set":3,"id":"ba8a016a_67ccb1bb","line":55,"in_reply_to":"ba8a016a_e5a45909","updated":"2015-11-24 10:37:42.000000000","message":"ACK.\nwill udapte it. \nAdd VIR_MIGRATE_AUTO_CONVERGE to live_migration_flag in nova.conf,\nAnd set the  auto-converge as default value.","commit_id":"ad52cac26749f2f7bdf440007f3fefc5daafd8e3"},{"author":{"_account_id":12299,"name":"Paweł Koniszewski","email":"pawel.koniszewski@gmail.com","username":"pawel.koniszewski"},"change_message_id":"145853e1fe7abb07f9f88e116a290f3f1a2ce8dc","unresolved":false,"context_lines":[{"line_number":57,"context_line":"For lower version libvirt(\u003c1.2.3) which does not support auto-converge, if the"},{"line_number":58,"context_line":"operator still want auto-converge to help to complete live migration"},{"line_number":59,"context_line":"successfully when VM with heavy workloads. So an extended auto-converge"},{"line_number":60,"context_line":"command will be supported by libvirt qemu-monitor-command(require qemu\u003e\u003d1.6.0)."},{"line_number":61,"context_line":""},{"line_number":62,"context_line":"Alternatives"},{"line_number":63,"context_line":"------------"}],"source_content_type":"text/x-rst","patch_set":3,"id":"ba8a016a_91dd9672","line":60,"updated":"2015-11-23 08:33:12.000000000","message":"Can you explain this a bit? I\u0027m not sure what you want to do there.","commit_id":"ad52cac26749f2f7bdf440007f3fefc5daafd8e3"},{"author":{"_account_id":12299,"name":"Paweł Koniszewski","email":"pawel.koniszewski@gmail.com","username":"pawel.koniszewski"},"change_message_id":"145853e1fe7abb07f9f88e116a290f3f1a2ce8dc","unresolved":false,"context_lines":[{"line_number":66,"context_line":"through hypervisor."},{"line_number":67,"context_line":""},{"line_number":68,"context_line":"Or, add VIR_MIGRATE_AUTO_CONVERGE to live_migration_flag in nova.conf, this"},{"line_number":69,"context_line":"is not flexible."},{"line_number":70,"context_line":""},{"line_number":71,"context_line":"Data model impact"},{"line_number":72,"context_line":"-----------------"}],"source_content_type":"text/x-rst","patch_set":3,"id":"ba8a016a_91c4b692","line":69,"updated":"2015-11-23 08:33:12.000000000","message":"IMO this solution is better than adding new parameter to API.","commit_id":"ad52cac26749f2f7bdf440007f3fefc5daafd8e3"},{"author":{"_account_id":14131,"name":"shaohef","email":"shaohe.feng@intel.com","username":"shaohefeng"},"change_message_id":"82cf1c206927e35adeb5be519e1e81cab6b7d704","unresolved":false,"context_lines":[{"line_number":66,"context_line":"through hypervisor."},{"line_number":67,"context_line":""},{"line_number":68,"context_line":"Or, add VIR_MIGRATE_AUTO_CONVERGE to live_migration_flag in nova.conf, this"},{"line_number":69,"context_line":"is not flexible."},{"line_number":70,"context_line":""},{"line_number":71,"context_line":"Data model impact"},{"line_number":72,"context_line":"-----------------"}],"source_content_type":"text/x-rst","patch_set":3,"id":"ba8a016a_a7eff91c","line":69,"in_reply_to":"ba8a016a_91c4b692","updated":"2015-11-24 10:37:42.000000000","message":"ACK.","commit_id":"ad52cac26749f2f7bdf440007f3fefc5daafd8e3"},{"author":{"_account_id":12299,"name":"Paweł Koniszewski","email":"pawel.koniszewski@gmail.com","username":"pawel.koniszewski"},"change_message_id":"145853e1fe7abb07f9f88e116a290f3f1a2ce8dc","unresolved":false,"context_lines":[{"line_number":96,"context_line":"    \u0027additionalProperties\u0027: False,"},{"line_number":97,"context_line":"  }"},{"line_number":98,"context_line":""},{"line_number":99,"context_line":"The default value of auto_converge will be False, this change need new"},{"line_number":100,"context_line":"microversion bump."},{"line_number":101,"context_line":""},{"line_number":102,"context_line":"Security impact"}],"source_content_type":"text/x-rst","patch_set":3,"id":"ba8a016a_11d3e639","line":99,"updated":"2015-11-23 08:33:12.000000000","message":"Why not True? It will only affect VMs that can\u0027t be transferred over network due to dirty pages ratio.","commit_id":"ad52cac26749f2f7bdf440007f3fefc5daafd8e3"},{"author":{"_account_id":4992,"name":"malini-k-bhandaru","email":"malini.k.bhandaru@intel.com","username":"malini-k-bhandaru"},"change_message_id":"e058b838b958b35b332a3d2b6f788827911f30e8","unresolved":false,"context_lines":[{"line_number":15,"context_line":"increase success chance of migration with busy enterprise workloads on their"},{"line_number":16,"context_line":"instance."},{"line_number":17,"context_line":""},{"line_number":18,"context_line":"This spec proposes a new opition auto-converge for live migration."},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"Problem description"},{"line_number":21,"context_line":"\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":4,"id":"7a740942_3fec4cf2","line":18,"updated":"2015-12-08 02:42:01.000000000","message":"s/opition/option","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":14131,"name":"shaohef","email":"shaohe.feng@intel.com","username":"shaohefeng"},"change_message_id":"e93f56f017b0a743bfe0eb2f158fa503f7cd33c5","unresolved":false,"context_lines":[{"line_number":15,"context_line":"increase success chance of migration with busy enterprise workloads on their"},{"line_number":16,"context_line":"instance."},{"line_number":17,"context_line":""},{"line_number":18,"context_line":"This spec proposes a new opition auto-converge for live migration."},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"Problem description"},{"line_number":21,"context_line":"\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":4,"id":"dab17558_de4976d2","line":18,"in_reply_to":"7a740942_3fec4cf2","updated":"2016-05-06 11:26:16.000000000","message":"Done","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":4992,"name":"malini-k-bhandaru","email":"malini.k.bhandaru@intel.com","username":"malini-k-bhandaru"},"change_message_id":"e058b838b958b35b332a3d2b6f788827911f30e8","unresolved":false,"context_lines":[{"line_number":31,"context_line":"https://review.openstack.org/#/c/245543"},{"line_number":32,"context_line":""},{"line_number":33,"context_line":"But in another cases, the user want to keep the their workloads on their vm"},{"line_number":34,"context_line":"uninterrupted, they need a new mechenical auto-converge to satisfy their"},{"line_number":35,"context_line":"requirement."},{"line_number":36,"context_line":""},{"line_number":37,"context_line":"Note, auto converge will decrease performance of VMs with memory-intensive"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7a740942_1f9dd065","line":34,"updated":"2015-12-08 02:42:01.000000000","message":"s/they need a new mechenical/they need the mechanism of","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":14131,"name":"shaohef","email":"shaohe.feng@intel.com","username":"shaohefeng"},"change_message_id":"e93f56f017b0a743bfe0eb2f158fa503f7cd33c5","unresolved":false,"context_lines":[{"line_number":31,"context_line":"https://review.openstack.org/#/c/245543"},{"line_number":32,"context_line":""},{"line_number":33,"context_line":"But in another cases, the user want to keep the their workloads on their vm"},{"line_number":34,"context_line":"uninterrupted, they need a new mechenical auto-converge to satisfy their"},{"line_number":35,"context_line":"requirement."},{"line_number":36,"context_line":""},{"line_number":37,"context_line":"Note, auto converge will decrease performance of VMs with memory-intensive"}],"source_content_type":"text/x-rst","patch_set":4,"id":"dab17558_be4632bd","line":34,"in_reply_to":"7a740942_1f9dd065","updated":"2016-05-06 11:26:16.000000000","message":"Done","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":1779,"name":"Daniel Berrange","email":"berrange@redhat.com","username":"berrange"},"change_message_id":"201f2214b9af4b063314bc6e3fcd83ac405323ce","unresolved":false,"context_lines":[{"line_number":35,"context_line":"requirement."},{"line_number":36,"context_line":""},{"line_number":37,"context_line":"Note, auto converge will decrease performance of VMs with memory-intensive"},{"line_number":38,"context_line":"workloads."},{"line_number":39,"context_line":""},{"line_number":40,"context_line":"Use Cases"},{"line_number":41,"context_line":"----------"}],"source_content_type":"text/x-rst","patch_set":4,"id":"dab17558_d3edf626","line":38,"range":{"start_line":38,"start_character":9,"end_line":38,"end_character":10},"updated":"2016-05-06 09:10:30.000000000","message":"In testing with worst case memory intensive workloads it is shown that auto-converge will have to throttle guest performance by as much as 95% before it will complete. It does this throttling to all guest CPUs indiscriminately, even if only one of the guest CPUs is responsible for the memory workload. Because of the way it does this throttling gradually, the guest may be running with dramatically reduced performance for quite a long period of time.\n\nIOW, auto-converge is not a low-impact solution - we should be clear that performance impact can be very significant.","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":1779,"name":"Daniel Berrange","email":"berrange@redhat.com","username":"berrange"},"change_message_id":"cbebd32beabacec248a3667c0bcea902b5175bcd","unresolved":false,"context_lines":[{"line_number":35,"context_line":"requirement."},{"line_number":36,"context_line":""},{"line_number":37,"context_line":"Note, auto converge will decrease performance of VMs with memory-intensive"},{"line_number":38,"context_line":"workloads."},{"line_number":39,"context_line":""},{"line_number":40,"context_line":"Use Cases"},{"line_number":41,"context_line":"----------"}],"source_content_type":"text/x-rst","patch_set":4,"id":"dab17558_5e6d8603","line":38,"range":{"start_line":38,"start_character":9,"end_line":38,"end_character":10},"in_reply_to":"dab17558_5ee8e67c","updated":"2016-05-06 11:29:28.000000000","message":"For default out of the box config, we should leave auto-converge as None in the config file. Nova should dynamically decide to use post-copy if it is available, and fallback to auto-converge if post-copy is not available.\n\nThe user will only need to set auto_converge\u003don if they wish to force use of auto-converge even when post-copy is available.","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":14131,"name":"shaohef","email":"shaohe.feng@intel.com","username":"shaohefeng"},"change_message_id":"e93f56f017b0a743bfe0eb2f158fa503f7cd33c5","unresolved":false,"context_lines":[{"line_number":35,"context_line":"requirement."},{"line_number":36,"context_line":""},{"line_number":37,"context_line":"Note, auto converge will decrease performance of VMs with memory-intensive"},{"line_number":38,"context_line":"workloads."},{"line_number":39,"context_line":""},{"line_number":40,"context_line":"Use Cases"},{"line_number":41,"context_line":"----------"}],"source_content_type":"text/x-rst","patch_set":4,"id":"dab17558_5ee8e67c","line":38,"range":{"start_line":38,"start_character":9,"end_line":38,"end_character":10},"in_reply_to":"dab17558_d3edf626","updated":"2016-05-06 11:26:16.000000000","message":"hi, Daniel Berrange\n \n from your comments, should we not set the default of auto-converge as \"on\" in nova.conf?","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":4992,"name":"malini-k-bhandaru","email":"malini.k.bhandaru@intel.com","username":"malini-k-bhandaru"},"change_message_id":"e058b838b958b35b332a3d2b6f788827911f30e8","unresolved":false,"context_lines":[{"line_number":40,"context_line":"Use Cases"},{"line_number":41,"context_line":"----------"},{"line_number":42,"context_line":""},{"line_number":43,"context_line":"As an operator of an OpenStack cloud, I would like do the live migration for"},{"line_number":44,"context_line":"some purpose, such as hypervisor load-balance or save energy. To ensure the SLA"},{"line_number":45,"context_line":"uninterrupted business, I observe there\u0027s still heavy workloads on the VM, I"},{"line_number":46,"context_line":"just simply enable auto-converge under libvirt section of nova.conf, and the"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7a740942_9fad4086","line":43,"updated":"2015-12-08 02:42:01.000000000","message":"s/do the/initiate","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":14131,"name":"shaohef","email":"shaohe.feng@intel.com","username":"shaohefeng"},"change_message_id":"e93f56f017b0a743bfe0eb2f158fa503f7cd33c5","unresolved":false,"context_lines":[{"line_number":40,"context_line":"Use Cases"},{"line_number":41,"context_line":"----------"},{"line_number":42,"context_line":""},{"line_number":43,"context_line":"As an operator of an OpenStack cloud, I would like do the live migration for"},{"line_number":44,"context_line":"some purpose, such as hypervisor load-balance or save energy. To ensure the SLA"},{"line_number":45,"context_line":"uninterrupted business, I observe there\u0027s still heavy workloads on the VM, I"},{"line_number":46,"context_line":"just simply enable auto-converge under libvirt section of nova.conf, and the"}],"source_content_type":"text/x-rst","patch_set":4,"id":"dab17558_fe16faa5","line":43,"in_reply_to":"7a740942_9fad4086","updated":"2016-05-06 11:26:16.000000000","message":"Done","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":4992,"name":"malini-k-bhandaru","email":"malini.k.bhandaru@intel.com","username":"malini-k-bhandaru"},"change_message_id":"e058b838b958b35b332a3d2b6f788827911f30e8","unresolved":false,"context_lines":[{"line_number":41,"context_line":"----------"},{"line_number":42,"context_line":""},{"line_number":43,"context_line":"As an operator of an OpenStack cloud, I would like do the live migration for"},{"line_number":44,"context_line":"some purpose, such as hypervisor load-balance or save energy. To ensure the SLA"},{"line_number":45,"context_line":"uninterrupted business, I observe there\u0027s still heavy workloads on the VM, I"},{"line_number":46,"context_line":"just simply enable auto-converge under libvirt section of nova.conf, and the"},{"line_number":47,"context_line":"live migration complete successfully with it\u0027s help."}],"source_content_type":"text/x-rst","patch_set":4,"id":"7a740942_9fe460e0","line":44,"updated":"2015-12-08 02:42:01.000000000","message":"or upgrade the host by way of software patches or other.","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":14131,"name":"shaohef","email":"shaohe.feng@intel.com","username":"shaohefeng"},"change_message_id":"e93f56f017b0a743bfe0eb2f158fa503f7cd33c5","unresolved":false,"context_lines":[{"line_number":41,"context_line":"----------"},{"line_number":42,"context_line":""},{"line_number":43,"context_line":"As an operator of an OpenStack cloud, I would like do the live migration for"},{"line_number":44,"context_line":"some purpose, such as hypervisor load-balance or save energy. To ensure the SLA"},{"line_number":45,"context_line":"uninterrupted business, I observe there\u0027s still heavy workloads on the VM, I"},{"line_number":46,"context_line":"just simply enable auto-converge under libvirt section of nova.conf, and the"},{"line_number":47,"context_line":"live migration complete successfully with it\u0027s help."}],"source_content_type":"text/x-rst","patch_set":4,"id":"dab17558_5e0c6694","line":44,"in_reply_to":"7a740942_9fe460e0","updated":"2016-05-06 11:26:16.000000000","message":"Done","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":4992,"name":"malini-k-bhandaru","email":"malini.k.bhandaru@intel.com","username":"malini-k-bhandaru"},"change_message_id":"e058b838b958b35b332a3d2b6f788827911f30e8","unresolved":false,"context_lines":[{"line_number":50,"context_line":"Proposed change"},{"line_number":51,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":52,"context_line":""},{"line_number":53,"context_line":"Auto-converge will throttle down memory writes only when there is no progress"},{"line_number":54,"context_line":"in live migration so probably we don\u0027t want to turn it off. Otherwise we might"},{"line_number":55,"context_line":"end up with never-ending live migration which is super bad for infrastructure."},{"line_number":56,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"7a740942_5f4118a8","line":53,"updated":"2015-12-08 02:42:01.000000000","message":"s/Auto-converge/Auto-converge is beneficial because it","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":14131,"name":"shaohef","email":"shaohe.feng@intel.com","username":"shaohefeng"},"change_message_id":"e93f56f017b0a743bfe0eb2f158fa503f7cd33c5","unresolved":false,"context_lines":[{"line_number":50,"context_line":"Proposed change"},{"line_number":51,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":52,"context_line":""},{"line_number":53,"context_line":"Auto-converge will throttle down memory writes only when there is no progress"},{"line_number":54,"context_line":"in live migration so probably we don\u0027t want to turn it off. Otherwise we might"},{"line_number":55,"context_line":"end up with never-ending live migration which is super bad for infrastructure."},{"line_number":56,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"dab17558_3e11a2ad","line":53,"in_reply_to":"7a740942_5f4118a8","updated":"2016-05-06 11:26:16.000000000","message":"Done","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":4992,"name":"malini-k-bhandaru","email":"malini.k.bhandaru@intel.com","username":"malini-k-bhandaru"},"change_message_id":"e058b838b958b35b332a3d2b6f788827911f30e8","unresolved":false,"context_lines":[{"line_number":55,"context_line":"end up with never-ending live migration which is super bad for infrastructure."},{"line_number":56,"context_line":""},{"line_number":57,"context_line":"Also in public cloud you have no idea what\u0027s running on virtual machines so you"},{"line_number":58,"context_line":"never know when you should use auto converge, and when you should not. In case"},{"line_number":59,"context_line":"you didn\u0027t turn it ON and there is no progress in VM transfer you will need to"},{"line_number":60,"context_line":"pause VM and affect VM performance a lot or cancel live migration and start"},{"line_number":61,"context_line":"another one with auto converge ON."}],"source_content_type":"text/x-rst","patch_set":4,"id":"7a740942_3f1beccb","line":58,"updated":"2015-12-08 02:42:01.000000000","message":"s/, and when you should not./ or not.","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":14131,"name":"shaohef","email":"shaohe.feng@intel.com","username":"shaohefeng"},"change_message_id":"e93f56f017b0a743bfe0eb2f158fa503f7cd33c5","unresolved":false,"context_lines":[{"line_number":55,"context_line":"end up with never-ending live migration which is super bad for infrastructure."},{"line_number":56,"context_line":""},{"line_number":57,"context_line":"Also in public cloud you have no idea what\u0027s running on virtual machines so you"},{"line_number":58,"context_line":"never know when you should use auto converge, and when you should not. In case"},{"line_number":59,"context_line":"you didn\u0027t turn it ON and there is no progress in VM transfer you will need to"},{"line_number":60,"context_line":"pause VM and affect VM performance a lot or cancel live migration and start"},{"line_number":61,"context_line":"another one with auto converge ON."}],"source_content_type":"text/x-rst","patch_set":4,"id":"dab17558_9e26ce14","line":58,"in_reply_to":"7a740942_3f1beccb","updated":"2016-05-06 11:26:16.000000000","message":"Done","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":4992,"name":"malini-k-bhandaru","email":"malini.k.bhandaru@intel.com","username":"malini-k-bhandaru"},"change_message_id":"e058b838b958b35b332a3d2b6f788827911f30e8","unresolved":false,"context_lines":[{"line_number":60,"context_line":"pause VM and affect VM performance a lot or cancel live migration and start"},{"line_number":61,"context_line":"another one with auto converge ON."},{"line_number":62,"context_line":""},{"line_number":63,"context_line":"So we think that we don\u0027t need to enable auto_converge flexibility."},{"line_number":64,"context_line":"It is OK that a new option auto_converge\u003don|off will be added under libvirt"},{"line_number":65,"context_line":"section in nova.conf."},{"line_number":66,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"7a740942_dfef4885","line":63,"updated":"2015-12-08 02:42:01.000000000","message":"Given the above, and its general useful, we believe it is useful to leave it enabled by default, as opposed to making it a configurable option and as input parameter.","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":14131,"name":"shaohef","email":"shaohe.feng@intel.com","username":"shaohefeng"},"change_message_id":"e93f56f017b0a743bfe0eb2f158fa503f7cd33c5","unresolved":false,"context_lines":[{"line_number":60,"context_line":"pause VM and affect VM performance a lot or cancel live migration and start"},{"line_number":61,"context_line":"another one with auto converge ON."},{"line_number":62,"context_line":""},{"line_number":63,"context_line":"So we think that we don\u0027t need to enable auto_converge flexibility."},{"line_number":64,"context_line":"It is OK that a new option auto_converge\u003don|off will be added under libvirt"},{"line_number":65,"context_line":"section in nova.conf."},{"line_number":66,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"dab17558_7e232a06","line":63,"in_reply_to":"7a740942_dfef4885","updated":"2016-05-06 11:26:16.000000000","message":"Done","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":4992,"name":"malini-k-bhandaru","email":"malini.k.bhandaru@intel.com","username":"malini-k-bhandaru"},"change_message_id":"e058b838b958b35b332a3d2b6f788827911f30e8","unresolved":false,"context_lines":[{"line_number":67,"context_line":"It is better to turn it ON by default in nova.conf, and without this config in"},{"line_number":68,"context_line":"nova.conf, there is a default option in nova to turn auto converge ON."},{"line_number":69,"context_line":""},{"line_number":70,"context_line":"We also will need to check the libvirt virson and qemu virsion. If qemu support"},{"line_number":71,"context_line":"auto-converge, but libvirt version is lower than 1.2.3, that means libvirt does"},{"line_number":72,"context_line":"not support auto-converge, then a qemu-monitor-command will be passed to qemu"},{"line_number":73,"context_line":"by libvirt to enable auto-converge."}],"source_content_type":"text/x-rst","patch_set":4,"id":"7a740942_bfa11c48","line":70,"updated":"2015-12-08 02:42:01.000000000","message":"s/virson/version","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":14131,"name":"shaohef","email":"shaohe.feng@intel.com","username":"shaohefeng"},"change_message_id":"e93f56f017b0a743bfe0eb2f158fa503f7cd33c5","unresolved":false,"context_lines":[{"line_number":67,"context_line":"It is better to turn it ON by default in nova.conf, and without this config in"},{"line_number":68,"context_line":"nova.conf, there is a default option in nova to turn auto converge ON."},{"line_number":69,"context_line":""},{"line_number":70,"context_line":"We also will need to check the libvirt virson and qemu virsion. If qemu support"},{"line_number":71,"context_line":"auto-converge, but libvirt version is lower than 1.2.3, that means libvirt does"},{"line_number":72,"context_line":"not support auto-converge, then a qemu-monitor-command will be passed to qemu"},{"line_number":73,"context_line":"by libvirt to enable auto-converge."}],"source_content_type":"text/x-rst","patch_set":4,"id":"dab17558_1e793efd","line":70,"in_reply_to":"7a740942_bfa11c48","updated":"2016-05-06 11:26:16.000000000","message":"Done","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":1779,"name":"Daniel Berrange","email":"berrange@redhat.com","username":"berrange"},"change_message_id":"201f2214b9af4b063314bc6e3fcd83ac405323ce","unresolved":false,"context_lines":[{"line_number":70,"context_line":"We also will need to check the libvirt virson and qemu virsion. If qemu support"},{"line_number":71,"context_line":"auto-converge, but libvirt version is lower than 1.2.3, that means libvirt does"},{"line_number":72,"context_line":"not support auto-converge, then a qemu-monitor-command will be passed to qemu"},{"line_number":73,"context_line":"by libvirt to enable auto-converge."},{"line_number":74,"context_line":""},{"line_number":75,"context_line":"The command is:"},{"line_number":76,"context_line":"\u0027{ \"execute\": \"migrate-set-capabilities\" , \"arguments\": { \"capabilities\": [ {"}],"source_content_type":"text/x-rst","patch_set":4,"id":"dab17558_d32a167e","line":73,"range":{"start_line":73,"start_character":34,"end_line":73,"end_character":35},"updated":"2016-05-06 09:10:30.000000000","message":"No, if the version of libvirt is insufficient, then we should simply not use auto-converge. Using monitor passthrough is not permitted in nova.","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":14131,"name":"shaohef","email":"shaohe.feng@intel.com","username":"shaohefeng"},"change_message_id":"e93f56f017b0a743bfe0eb2f158fa503f7cd33c5","unresolved":false,"context_lines":[{"line_number":70,"context_line":"We also will need to check the libvirt virson and qemu virsion. If qemu support"},{"line_number":71,"context_line":"auto-converge, but libvirt version is lower than 1.2.3, that means libvirt does"},{"line_number":72,"context_line":"not support auto-converge, then a qemu-monitor-command will be passed to qemu"},{"line_number":73,"context_line":"by libvirt to enable auto-converge."},{"line_number":74,"context_line":""},{"line_number":75,"context_line":"The command is:"},{"line_number":76,"context_line":"\u0027{ \"execute\": \"migrate-set-capabilities\" , \"arguments\": { \"capabilities\": [ {"}],"source_content_type":"text/x-rst","patch_set":4,"id":"dab17558_fe7d5ae8","line":73,"range":{"start_line":73,"start_character":34,"end_line":73,"end_character":35},"in_reply_to":"dab17558_d32a167e","updated":"2016-05-06 11:26:16.000000000","message":"Done","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":1779,"name":"Daniel Berrange","email":"berrange@redhat.com","username":"berrange"},"change_message_id":"201f2214b9af4b063314bc6e3fcd83ac405323ce","unresolved":false,"context_lines":[{"line_number":81,"context_line":"------------"},{"line_number":82,"context_line":""},{"line_number":83,"context_line":"Alternative is not doing this and let operator enable auto-converge manually"},{"line_number":84,"context_line":"through hypervisor."},{"line_number":85,"context_line":""},{"line_number":86,"context_line":"Or, a new option auto-converge is added in migrate_server API."},{"line_number":87,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"dab17558_4812f163","line":84,"range":{"start_line":84,"start_character":18,"end_line":84,"end_character":19},"updated":"2016-05-06 09:10:30.000000000","message":"The major alternative is post-copy migration, which can ensure completion with less of an impact on guest performance overall, albeit at the cost of a small risk in event of network failure","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":14131,"name":"shaohef","email":"shaohe.feng@intel.com","username":"shaohefeng"},"change_message_id":"e93f56f017b0a743bfe0eb2f158fa503f7cd33c5","unresolved":false,"context_lines":[{"line_number":81,"context_line":"------------"},{"line_number":82,"context_line":""},{"line_number":83,"context_line":"Alternative is not doing this and let operator enable auto-converge manually"},{"line_number":84,"context_line":"through hypervisor."},{"line_number":85,"context_line":""},{"line_number":86,"context_line":"Or, a new option auto-converge is added in migrate_server API."},{"line_number":87,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"dab17558_5e83c6cc","line":84,"range":{"start_line":84,"start_character":18,"end_line":84,"end_character":19},"in_reply_to":"dab17558_4812f163","updated":"2016-05-06 11:26:16.000000000","message":"Done","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":1779,"name":"Daniel Berrange","email":"berrange@redhat.com","username":"berrange"},"change_message_id":"201f2214b9af4b063314bc6e3fcd83ac405323ce","unresolved":false,"context_lines":[{"line_number":143,"context_line":""},{"line_number":144,"context_line":"* A new option auto_converge\u003don|off is added."},{"line_number":145,"context_line":"* An extended auto-converge command through libvirt qemu-monitor-command"},{"line_number":146,"context_line":"  is added."},{"line_number":147,"context_line":""},{"line_number":148,"context_line":"Dependencies"},{"line_number":149,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":4,"id":"dab17558_5310e64e","line":146,"range":{"start_line":146,"start_character":10,"end_line":146,"end_character":11},"updated":"2016-05-06 09:10:30.000000000","message":"No, Nova must *never* use qemu-monitor-command. That is only for developer experimentation, never for use in production deployment.","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"},{"author":{"_account_id":14131,"name":"shaohef","email":"shaohe.feng@intel.com","username":"shaohefeng"},"change_message_id":"e93f56f017b0a743bfe0eb2f158fa503f7cd33c5","unresolved":false,"context_lines":[{"line_number":143,"context_line":""},{"line_number":144,"context_line":"* A new option auto_converge\u003don|off is added."},{"line_number":145,"context_line":"* An extended auto-converge command through libvirt qemu-monitor-command"},{"line_number":146,"context_line":"  is added."},{"line_number":147,"context_line":""},{"line_number":148,"context_line":"Dependencies"},{"line_number":149,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":4,"id":"dab17558_7e6a0a1c","line":146,"range":{"start_line":146,"start_character":10,"end_line":146,"end_character":11},"in_reply_to":"dab17558_5310e64e","updated":"2016-05-06 11:26:16.000000000","message":"Done","commit_id":"d77c9a4e5bc5960d6f84c2b20639c7e1c055ead6"}],"specs/newton/approved/live-migration-support-auto-converge.rst":[{"author":{"_account_id":12299,"name":"Paweł Koniszewski","email":"pawel.koniszewski@gmail.com","username":"pawel.koniszewski"},"change_message_id":"a838cdb17e800592433d9bb610aeb096328ce319","unresolved":false,"context_lines":[{"line_number":76,"context_line":"parameter. It is OK that a new option auto_converge\u003don|off will be added under"},{"line_number":77,"context_line":"libvirt section in nova.conf."},{"line_number":78,"context_line":""},{"line_number":79,"context_line":"It is better to turn it ON by default in nova.conf, and without this config in"},{"line_number":80,"context_line":"nova.conf, there is a default option in nova to turn auto converge ON."},{"line_number":81,"context_line":""},{"line_number":82,"context_line":"We also will need to check the libvirt version, and do not need to check qemu"}],"source_content_type":"text/x-rst","patch_set":5,"id":"dab17558_465102d8","line":79,"updated":"2016-05-09 08:27:27.000000000","message":"I agree with Daniel. Let\u0027s keep auto converge off by default (as it is right now). Let nova decide automatically what should be used to speed up live migration process. auto_converge\u003don should only be there for forcing auto converge during live migrations","commit_id":"346c70b9bb268ece480f1ead80cd40fac40076e3"}]}