)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"6bb1b3417a37d03dda0f8d8c236218f0728528b2","unresolved":true,"context_lines":[{"line_number":10,"context_line":"OS because QEMU does not set a rotation_rate by default. This"},{"line_number":11,"context_line":"causes suboptimal I/O scheduler selection in the guest kernel"},{"line_number":12,"context_line":"(mq-deadline instead of none/noop for SSDs)."},{"line_number":13,"context_line":""},{"line_number":14,"context_line":"This patch adds:"},{"line_number":15,"context_line":"- A new config option [libvirt]disk_rotation_rate (default: 1/SSD)"},{"line_number":16,"context_line":"- Support for hw:disk_rotation_rate flavor extra spec"},{"line_number":17,"context_line":"- Support for hw_disk_rotation_rate image property"},{"line_number":18,"context_line":"- rotation_rate element in generated libvirt XML"},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"Requires: libvirt \u003e\u003d 9.0.0, QEMU \u003e\u003d 7.1"},{"line_number":21,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"91e564e7_7ff65960","line":18,"range":{"start_line":13,"start_character":1,"end_line":18,"end_character":48},"updated":"2026-03-09 17:16:24.000000000","message":"this is combining 2 diffent features and its incomplete\n\n\nfirst its  intoducing  part of a previously \nproposed but never merged feature\nhttps://review.opendev.org/c/openstack/nova/+/920818\nbut it does nto address the missing part fo the feature namely the lack of object changes to support the new image propertes or the testing element of the feature \n\nim unsure if my comment regarding \nhttps://review.opendev.org/c/openstack/nova/+/920818/comment/9f2bfaa4_5f7119ab/\n\nhave been adressed but we need to ensure we do not break upgrades\nor live migration as well.\n\nthis is partly why we strongly prefer not to have per host config option that affect teh domain xml\n\nif we do we need to consdier the bevhior when we move the instance between host and determin if we shoudl only allow the vm to move ot hosts with the same configuration to keep live migration transparten to the guest or if it can be udpated and when.\n\nin general we try to avoid casue where a vm boot with conficuation A, live migate to a host that uses configuation B and only adopts configuraton B latere the next time it repoots.\n\nthis lead to latent failures.\n\nso typically for somthign like this we woudl only want this to change on rebuild or resize and otherwise be fixed for the lifetime fo the instance.\n\nfor cinder volumes it would be valide to change this via detach adn attach but the current spec does not adress per cinder volume configuration of this.","commit_id":"f9f7fa77bd7424af5f42e90013af53e74b7d7b58"},{"author":{"_account_id":38909,"name":"Thibaut Démaret","display_name":"Thibaut Démaret","email":"thidem@worteks.com","username":"worteks-thidem"},"change_message_id":"b2936d5ecc644445e0495dac287a5f2dd6a1aa4b","unresolved":true,"context_lines":[{"line_number":10,"context_line":"OS because QEMU does not set a rotation_rate by default. This"},{"line_number":11,"context_line":"causes suboptimal I/O scheduler selection in the guest kernel"},{"line_number":12,"context_line":"(mq-deadline instead of none/noop for SSDs)."},{"line_number":13,"context_line":""},{"line_number":14,"context_line":"This patch adds:"},{"line_number":15,"context_line":"- A new config option [libvirt]disk_rotation_rate (default: 1/SSD)"},{"line_number":16,"context_line":"- Support for hw:disk_rotation_rate flavor extra spec"},{"line_number":17,"context_line":"- Support for hw_disk_rotation_rate image property"},{"line_number":18,"context_line":"- rotation_rate element in generated libvirt XML"},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"Requires: libvirt \u003e\u003d 9.0.0, QEMU \u003e\u003d 7.1"},{"line_number":21,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"61e435b8_e91dc9b7","line":18,"range":{"start_line":13,"start_character":1,"end_line":18,"end_character":48},"in_reply_to":"0ac93688_15978767","updated":"2026-03-24 08:39:21.000000000","message":"Normally I resolv this point and I change my PR with your comments.\n\nRegards","commit_id":"f9f7fa77bd7424af5f42e90013af53e74b7d7b58"},{"author":{"_account_id":38909,"name":"Thibaut Démaret","display_name":"Thibaut Démaret","email":"thidem@worteks.com","username":"worteks-thidem"},"change_message_id":"222cbf71be1d03339cde310f653b9e16256a389c","unresolved":false,"context_lines":[{"line_number":10,"context_line":"OS because QEMU does not set a rotation_rate by default. This"},{"line_number":11,"context_line":"causes suboptimal I/O scheduler selection in the guest kernel"},{"line_number":12,"context_line":"(mq-deadline instead of none/noop for SSDs)."},{"line_number":13,"context_line":""},{"line_number":14,"context_line":"This patch adds:"},{"line_number":15,"context_line":"- A new config option [libvirt]disk_rotation_rate (default: 1/SSD)"},{"line_number":16,"context_line":"- Support for hw:disk_rotation_rate flavor extra spec"},{"line_number":17,"context_line":"- Support for hw_disk_rotation_rate image property"},{"line_number":18,"context_line":"- rotation_rate element in generated libvirt XML"},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"Requires: libvirt \u003e\u003d 9.0.0, QEMU \u003e\u003d 7.1"},{"line_number":21,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"386ef13f_bfd6d8f2","line":18,"range":{"start_line":13,"start_character":1,"end_line":18,"end_character":48},"in_reply_to":"61e435b8_e91dc9b7","updated":"2026-04-23 14:48:51.000000000","message":"Acknowledged","commit_id":"f9f7fa77bd7424af5f42e90013af53e74b7d7b58"},{"author":{"_account_id":38909,"name":"Thibaut Démaret","display_name":"Thibaut Démaret","email":"thidem@worteks.com","username":"worteks-thidem"},"change_message_id":"83719348e5792efe6687e69230a79e8d4e58899f","unresolved":true,"context_lines":[{"line_number":10,"context_line":"OS because QEMU does not set a rotation_rate by default. This"},{"line_number":11,"context_line":"causes suboptimal I/O scheduler selection in the guest kernel"},{"line_number":12,"context_line":"(mq-deadline instead of none/noop for SSDs)."},{"line_number":13,"context_line":""},{"line_number":14,"context_line":"This patch adds:"},{"line_number":15,"context_line":"- A new config option [libvirt]disk_rotation_rate (default: 1/SSD)"},{"line_number":16,"context_line":"- Support for hw:disk_rotation_rate flavor extra spec"},{"line_number":17,"context_line":"- Support for hw_disk_rotation_rate image property"},{"line_number":18,"context_line":"- rotation_rate element in generated libvirt XML"},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"Requires: libvirt \u003e\u003d 9.0.0, QEMU \u003e\u003d 7.1"},{"line_number":21,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"0ac93688_15978767","line":18,"range":{"start_line":13,"start_character":1,"end_line":18,"end_character":48},"in_reply_to":"91e564e7_7ff65960","updated":"2026-03-11 09:03:17.000000000","message":"Thank you for your detailed and constructive feedback. I understand your concerns regarding the combination of features and the current incompleteness of this PR.\nRegarding your first point, I acknowledge that this PR introduces a part of a previously proposed but unmerged feature (https://review.opendev.org/c/openstack/nova/+/920818). I will ensure that the missing parts, specifically the object changes required for the new image properties and the testing elements, are addressed.\nI will also verify if my previous comments concerning https://review.opendev.org/c/openstack/nova/+/920818/comment/9f2bfaa4_5f7119ab/ have been addressed, and I am committed to ensuring no regressions occur in upgrades or live migrations.\nI appreciate your emphasis on avoiding per-host configuration options that affect the domain XML. I understand the importance of considering the behavior during instance migrations between hosts with differing configurations. I will explore solutions to ensure either migration to hosts with identical configurations or a transparent update mechanism for the guest.\nYour point about latent failures due to configuration changes post-live migration is very pertinent. I agree that such changes should ideally only occur on rebuild or resize and remain fixed for the instance\u0027s lifetime.\nLastly, I note your comment regarding Cinder volumes. While the change is valid for Cinder volumes during detach/attach, I will ensure the specification properly addresses per-Cinder volume configuration.\nI will work on these points and will provide an updated PR as soon as possible. Your input is highly valued.","commit_id":"f9f7fa77bd7424af5f42e90013af53e74b7d7b58"}]}