)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"54d2cb2c3b79b1d9b00244d42f0b2390b859854f","unresolved":true,"context_lines":[{"line_number":24,"context_line":"allocation ratios are never respected for the NUMA node 0, so NUMA node 0"},{"line_number":25,"context_line":"is always a valid candidate for Live Migrations, which means that with"},{"line_number":26,"context_line":"every Live Migration the NUMA node 0 will get more and more overcommited,"},{"line_number":27,"context_line":"eventually causing VMs to get into OOM as well."},{"line_number":28,"context_line":""},{"line_number":29,"context_line":"This commit simply passes those limits received from the scheduler onto"},{"line_number":30,"context_line":"nova-compute, so they can be used to make sure we don\u0027t overcommit NUMA nodes."}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"ac12a64c_515c472b","line":27,"updated":"2021-03-08 15:52:05.000000000","message":"that is not really correct.\npassing the limits will not alther the placment.\nthe vms in your case just have hw:numa_nodes\u003d1 \nthe vm is not claiming any resouce in the host numa toplogy blob so if the vm is smaller then the numa 0s free resouces then numa node 0 will be chosen\nthis is the same behavior you would have had without passing the limtis.\n\n\nwithout functional test to prove that this works and exactuly how this changes behvior we should not merge this.\nthe unit tests are not sufficnt to probley validate this.","commit_id":"6795a47bd1cd9c926d5e0f6d08a616d2214ea7f1"}]}