)]}'
{"nova/compute/manager.py":[{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"6a8b0d2c0b3567c089820e0d54f507c24cb942d1","unresolved":true,"context_lines":[{"line_number":1759,"context_line":"                    \u0027number of disk devices.\u0027)"},{"line_number":1760,"context_line":"            raise exception.InvalidConfiguration(msg)"},{"line_number":1761,"context_line":""},{"line_number":1762,"context_line":"        self.driver.init_host(host\u003dself.host)"},{"line_number":1763,"context_line":""},{"line_number":1764,"context_line":"        if service_ref:"},{"line_number":1765,"context_line":"            # If we are an existing service, check to see if we need"}],"source_content_type":"text/x-python","patch_set":1,"id":"445b3556_183f7f78","side":"PARENT","line":1762,"updated":"2026-05-28 19:38:30.000000000","message":"it would be nice if we could pass in the instances here since at lest the libivrt driver is already getting all the instnace for the currnet host so that it can register the image metadata defautls.\n\ni am also thinking that it would be preferbael to do the comptue host rename check before we call this for a similar reason.\n\nit would be better to not do any db modifcation on instance object before that validatrion.\n\nthis is out os scope fo this bug to some degree but it might be worth reconsidering the ordering here amd moveing\n\nself.driver.init_host(host\u003dself.host) to line 1788 below right before calling init_virt_events. and after where we currently get the instnaces. so we can pass them in and elimate the addtional call entirly.\n\npassing the instnace would be a virt direr api change however.","commit_id":"db1e6ddfe7d9ac631d0b842810242f8821e1b818"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"a1db7d869d16a2af94d83f0f243ea9ed6e8fb1da","unresolved":true,"context_lines":[{"line_number":1789,"context_line":"        self.init_virt_events()"},{"line_number":1790,"context_line":""},{"line_number":1791,"context_line":"        self._validate_pinning_configuration(instances)"},{"line_number":1792,"context_line":"        self._validate_vtpm_configuration(instances)"},{"line_number":1793,"context_line":""},{"line_number":1794,"context_line":"        # NOTE(gibi): If ironic and vcenter virt driver slow start time"},{"line_number":1795,"context_line":"        # becomes problematic here then we should consider adding a config"}],"source_content_type":"text/x-python","patch_set":1,"id":"97216f2b_1e3f3d03","side":"PARENT","line":1792,"updated":"2026-05-28 20:04:34.000000000","message":"the other wasy to fix this are\n1: wrap this ins an if and only call it if its libvirt\n2: add a second compute capablity flag in addtion to the exisitng one.\n3: delte this check engirely by effectivly reverting \nhttps://opendev.org/openstack/nova/commit/bf5eea6fda23cc709e31cedb39e4054fafe923d2\n4: add a validate_instances driver method and move this to that\n\n\nby the way cpu pinning is alos only implemnted by the libvirt driver \nso _validate_pinning_configuration\n\nhttps://github.com/openstack/nova/blob/380f657b5102707a3da676478e72fd96691e966b/nova/compute/manager.py#L1069\n\nshoudl also probably not be in the compute manger.\n\nnow it is true tha twhiel other driver dont supprot cpu pinnign they do supprot vcpu_pin_set to contole the nubmer of cpus that tehy can used but valdiating the config option don theed the instances.\nthis is just for valid the instance affinciy on startup up and i belive only the libvirt driver care about htat.","commit_id":"db1e6ddfe7d9ac631d0b842810242f8821e1b818"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"6ff7cfad0dfd2c6736686a954eb51fb5e241f6ed","unresolved":true,"context_lines":[{"line_number":1789,"context_line":"        self.init_virt_events()"},{"line_number":1790,"context_line":""},{"line_number":1791,"context_line":"        self._validate_pinning_configuration(instances)"},{"line_number":1792,"context_line":"        self._validate_vtpm_configuration(instances)"},{"line_number":1793,"context_line":""},{"line_number":1794,"context_line":"        # NOTE(gibi): If ironic and vcenter virt driver slow start time"},{"line_number":1795,"context_line":"        # becomes problematic here then we should consider adding a config"}],"source_content_type":"text/x-python","patch_set":1,"id":"883f26b9_dabb0709","side":"PARENT","line":1792,"in_reply_to":"97216f2b_1e3f3d03","updated":"2026-05-28 22:39:03.000000000","message":"the more i think about this the more i feel option 4 is the right path here\n\ni shoudl add \n```\ndef process_instances_at_startup(self, instances):\n  return instances\n```\n\nto the base virt driver adn i should move\n\n_validate_pinning_configuration, _validate_vtpm_configuration\nand _register_all_undefined_instance_details to libvirt implmeation of process_instances_at_startup. \ninstead and pass down the instances form the compute manger to the virt driver so that the virt dirver\n\nthat will allwo use to remove a db call form teh virt dirver and do only isntnace lookup on start up.\n\nif i rebase on https://review.opendev.org/c/openstack/nova/+/977037","commit_id":"db1e6ddfe7d9ac631d0b842810242f8821e1b818"}]}