)]}'
{"/PATCHSET_LEVEL":[{"author":{"_account_id":9816,"name":"Takashi Kajinami","email":"kajinamit@oss.nttdata.com","username":"kajinamit"},"change_message_id":"1121130b9af4e4be03adaaee4ac79df624e9221b","unresolved":true,"context_lines":[],"source_content_type":"","patch_set":1,"id":"5af74db7_60cb5db3","updated":"2026-07-06 07:31:02.000000000","message":"Please note that spec freeze for 2026.2 cycle has already passed. This should be proposed again once 2027.1 development cycle is started.\n\nAlso did you check https://specs.openstack.org/openstack/nova-specs/readme.html to create the spec file ? The submitted content does not include the sections you are supposed to include. (voting -1 due to the incomplete content)","commit_id":"62e93d974d9551d968c86b20b74af409cfb18422"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"fba2144ba100d65796180f7ec7cdaf4d6623dce7","unresolved":true,"context_lines":[],"source_content_type":"","patch_set":1,"id":"e65404e5_298bf82f","in_reply_to":"5af74db7_60cb5db3","updated":"2026-07-13 16:00:49.000000000","message":"i think this was created based on my comment on https://review.opendev.org/c/openstack/nova/+/996058\n\nbut yes this woudl be a 2027.1 item and i think we need a much wider deicusion on if this should be in the scope fo nova  to do or not vs cybrog or watcher or some combindiaton of the 3","commit_id":"62e93d974d9551d968c86b20b74af409cfb18422"}],"specs/2026.2/approved/mig-minimal-displacement-repartition.rst":[{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"fba2144ba100d65796180f7ec7cdaf4d6623dce7","unresolved":true,"context_lines":[{"line_number":11,"context_line":"mdevs automatically and doesn\u0027t reliably support heterogeneous MIG profiles "},{"line_number":12,"context_line":"on one card (bug #2008883). Modern MIG drivers support per-instance "},{"line_number":13,"context_line":"create/destroy without resetting the whole card, so this full-drain "},{"line_number":14,"context_line":"cost is a Nova orchestration gap, not a hardware limitation."},{"line_number":15,"context_line":""},{"line_number":16,"context_line":"Proposed Change"},{"line_number":17,"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":1,"id":"6363b0b1_cd12c5b2","line":14,"updated":"2026-07-13 16:00:49.000000000","message":"that is technicly ture but nova does nto supprot that by desgin\n\nwe declared dynmic partioning of mig out of scope fo nova when we were designing the orgin integration\n\nits currnly also outside the scope of cybrog for that matter but it may someday be consider there.","commit_id":"62e93d974d9551d968c86b20b74af409cfb18422"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"fba2144ba100d65796180f7ec7cdaf4d6623dce7","unresolved":true,"context_lines":[{"line_number":13,"context_line":"create/destroy without resetting the whole card, so this full-drain "},{"line_number":14,"context_line":"cost is a Nova orchestration gap, not a hardware limitation."},{"line_number":15,"context_line":""},{"line_number":16,"context_line":"Proposed Change"},{"line_number":17,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":18,"context_line":"Compute the minimal subset of currently running instances that actually "},{"line_number":19,"context_line":"block a requested repartitioning (an integer-program formulation over "},{"line_number":20,"context_line":"the card\u0027s compute/memory slot occupancy). Live-migrate only that "},{"line_number":21,"context_line":"subset via the existing os-migrations machinery under a configurable "},{"line_number":22,"context_line":"concurrent-disruption budget, then resync Placement inventory and clean "},{"line_number":23,"context_line":"up the stale mdevs as part of the operation."},{"line_number":24,"context_line":""},{"line_number":25,"context_line":"The initial scope is operator-triggered via a new \u0027nova-manage\u0027 "},{"line_number":26,"context_line":"command, rather than an automatic scheduler decision."}],"source_content_type":"text/x-rst","patch_set":1,"id":"a760a2a8_6bc4d65d","line":26,"range":{"start_line":16,"start_character":1,"end_line":26,"end_character":53},"updated":"2026-07-13 16:00:49.000000000","message":"i asked for the probleem descripton nad usecases not the proposed change\n\nbecause directinlly what you started in https://review.opendev.org/c/openstack/nova/+/996058\nis not something that woudl fit in nova.\n\nas proposed this feel much more in scope of watcher where the existing  vm_consolidation goal would gain a new stagey similar to https://docs.openstack.org/watcher/latest/strategies/vm_workload_consolidation.html\n\nwhich would concolidate the vgpu instnace to the smallest subset of host that would fit them.\n\n\nin nova a possible directly although not nessiarly the bset one would be a weigher that would prefer to spread or pack vgpu assignment by ordering the allcoation candites to express that policy.\n\nnova scope fo the project document explictly puts new orchstration api and perect scduling as out of scope \n\nhttps://docs.openstack.org/nova/latest/contributor/project-scope.html#no-more-orchestration\nhttps://docs.openstack.org/nova/latest/contributor/project-scope.html#iaas-not-batch-processing\n\n\nour entire gpu enabling desgin was intentionally based on static partitioning not dynamic irrespective of the hardware capabilities because that is what placement and our config driven workflow is desginged to model and supprot.\n\nnova-mange is not for this and exetending it to trigger this woudl violate a number of our interal design tenants.\n\nnova-manage is a admin only tool intended for priamrly db upgreade related action or other fixes that are requried when upgrading nova.\n\nit has limited repair capablity but orchstrating live migration or caluating them is not in its current or future scope.","commit_id":"62e93d974d9551d968c86b20b74af409cfb18422"}]}