)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":1779,"name":"Daniel Berrange","email":"berrange@redhat.com","username":"berrange"},"change_message_id":"326bd90f2e8c56a9b2700d6d9eb0f7967e80560f","unresolved":false,"context_lines":[{"line_number":12,"context_line":"This happens when:"},{"line_number":13,"context_line":"    a) NUMATopologyFilter is unset"},{"line_number":14,"context_line":"        NUMA filter loads oversubscription limits for ram and cpu"},{"line_number":15,"context_line":"        (ram_allocation_ratio and cpu_allocation_ratio)"},{"line_number":16,"context_line":"OR"},{"line_number":17,"context_line":"    b) --availability-zone zone:host is set (force_hosts)"},{"line_number":18,"context_line":"        force_hosts skips filter logic on scheduler"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"dada55a8_ceaed3e6","line":15,"updated":"2016-07-22 17:15:51.000000000","message":"Using numa while not having NUMATopologyFilter present is simply an deployment error scenario. We shouldn\u0027t be trying todo numa placement of the Vm at all if the NUMATopologyFilter is not enabled.","commit_id":"83681fe8b7f1f3a77d3aa5e718d86d1376d83869"},{"author":{"_account_id":8175,"name":"Rafael Folco","email":"rfolco@redhat.com","username":"rfolco"},"change_message_id":"cab910c2ec01fd3c9451b724a932cac155f48ef1","unresolved":false,"context_lines":[{"line_number":12,"context_line":"This happens when:"},{"line_number":13,"context_line":"    a) NUMATopologyFilter is unset"},{"line_number":14,"context_line":"        NUMA filter loads oversubscription limits for ram and cpu"},{"line_number":15,"context_line":"        (ram_allocation_ratio and cpu_allocation_ratio)"},{"line_number":16,"context_line":"OR"},{"line_number":17,"context_line":"    b) --availability-zone zone:host is set (force_hosts)"},{"line_number":18,"context_line":"        force_hosts skips filter logic on scheduler"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"dada55a8_997b877b","line":15,"in_reply_to":"dada55a8_ceaed3e6","updated":"2016-07-22 17:52:04.000000000","message":"Indeed the exception should be raised much earlier in this case.","commit_id":"83681fe8b7f1f3a77d3aa5e718d86d1376d83869"},{"author":{"_account_id":1779,"name":"Daniel Berrange","email":"berrange@redhat.com","username":"berrange"},"change_message_id":"326bd90f2e8c56a9b2700d6d9eb0f7967e80560f","unresolved":false,"context_lines":[{"line_number":15,"context_line":"        (ram_allocation_ratio and cpu_allocation_ratio)"},{"line_number":16,"context_line":"OR"},{"line_number":17,"context_line":"    b) --availability-zone zone:host is set (force_hosts)"},{"line_number":18,"context_line":"        force_hosts skips filter logic on scheduler"},{"line_number":19,"context_line":"AND"},{"line_number":20,"context_line":"    c) cpu pinning is disabled (hw:cpu_policy unset)"},{"line_number":21,"context_line":"AND"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"dada55a8_0e90eb94","line":18,"updated":"2016-07-22 17:15:51.000000000","message":"This is a known issue whereby we need to re-run schedular filters (like NUMATopologyFilter) even when a host is forced\nSee \n\nhttps://bugs.launchpad.net/nova/+bug/1427772","commit_id":"83681fe8b7f1f3a77d3aa5e718d86d1376d83869"},{"author":{"_account_id":8175,"name":"Rafael Folco","email":"rfolco@redhat.com","username":"rfolco"},"change_message_id":"cab910c2ec01fd3c9451b724a932cac155f48ef1","unresolved":false,"context_lines":[{"line_number":15,"context_line":"        (ram_allocation_ratio and cpu_allocation_ratio)"},{"line_number":16,"context_line":"OR"},{"line_number":17,"context_line":"    b) --availability-zone zone:host is set (force_hosts)"},{"line_number":18,"context_line":"        force_hosts skips filter logic on scheduler"},{"line_number":19,"context_line":"AND"},{"line_number":20,"context_line":"    c) cpu pinning is disabled (hw:cpu_policy unset)"},{"line_number":21,"context_line":"AND"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"dada55a8_39789b8f","line":18,"in_reply_to":"dada55a8_0e90eb94","updated":"2016-07-22 17:52:04.000000000","message":"There is no consensus on what the behavior should be for the force_hosts scenario :(\n\nThe idea was just to fall back to cell limits if the oversubscription limits are not set. \nThe 2 bugs mentioned should be separately fixed anyway.","commit_id":"83681fe8b7f1f3a77d3aa5e718d86d1376d83869"}]}