)]}'
{"nova/tests/unit/virt/libvirt/test_driver.py":[{"author":{"_account_id":11303,"name":"Markus Zoeller (markus_z)","email":"mzoeller@linux.vnet.ibm.com","username":"mzoeller"},"change_message_id":"c5cc517ef4483b8445ee2db22e9b0164e55ff38b","unresolved":false,"context_lines":[{"line_number":5205,"context_line":"        dest_topology \u003d objects.NUMATopology("},{"line_number":5206,"context_line":"            cells\u003d["},{"line_number":5207,"context_line":"                objects.NUMACell("},{"line_number":5208,"context_line":"                    id\u003d1, cpuset\u003dset([1, 2]), memory\u003d1024, mempages\u003d[],"},{"line_number":5209,"context_line":"                    )])"},{"line_number":5210,"context_line":"        drvr \u003d libvirt_driver.LibvirtDriver(fake.FakeVirtAPI(), False)"},{"line_number":5211,"context_line":"        src_compute_info \u003d {\u0027cpu_info\u0027: \u0027asdf\u0027, \u0027numa_topology\u0027: \u0027\"{src}\"\u0027}"}],"source_content_type":"text/x-python","patch_set":2,"id":"7aaa499b_b153eae1","line":5208,"updated":"2015-04-15 12:44:26.000000000","message":"Adding a test where the cpu-set of the target host is greater than the cpu-set of the source host would be great. I think I had the problem already when there is a difference, not only if the cpu-set of the target host is smaller the the cpu-set of the host (like in this unit test).","commit_id":"9ce4bb65f86a695a4fea41f8047d41259c9ab4b0"}],"nova/virt/libvirt/driver.py":[{"author":{"_account_id":11303,"name":"Markus Zoeller (markus_z)","email":"mzoeller@linux.vnet.ibm.com","username":"mzoeller"},"change_message_id":"c5cc517ef4483b8445ee2db22e9b0164e55ff38b","unresolved":false,"context_lines":[{"line_number":5008,"context_line":"            for dest_cell in dest_numa_obj.cells:"},{"line_number":5009,"context_line":"                host_cpuset \u003d host_cpuset.union(dest_cell.cpuset)"},{"line_number":5010,"context_line":""},{"line_number":5011,"context_line":"            if not guest_cpuset.issubset(host_cpuset):"},{"line_number":5012,"context_line":"                reason \u003d (_(\u0027Unable to migrate to destation host, cpuset \u0027"},{"line_number":5013,"context_line":"                            \u0027mismatch, guest cpuset: %(guest_cpuset)s, \u0027"},{"line_number":5014,"context_line":"                            \u0027host cpuset: %(host_cpuset)s.\u0027),"}],"source_content_type":"text/x-python","patch_set":2,"id":"7aaa499b_115a9e2a","line":5011,"updated":"2015-04-15 12:44:26.000000000","message":"What about the second option you mentioned in the bug description: \n\n   \"2 live change cpu set in the dom xml, to allow this\n   migration. we can do this by pre_live-migration in \n   dest host\"\n\nWhat would be the impact with that approach? If number of chosen vCPUs is smaller then the cpu-set of the target host, then we could cancel the live migration. If the number fits into the target host cpu-set, then we could safely live migrate, right?","commit_id":"9ce4bb65f86a695a4fea41f8047d41259c9ab4b0"},{"author":{"_account_id":12175,"name":"Eli Qiao","email":"qiaoliyong@gmail.com","username":"Eli"},"change_message_id":"30409c91493abbed3f2f49ed859065f25d72c208","unresolved":false,"context_lines":[{"line_number":5008,"context_line":"            for dest_cell in dest_numa_obj.cells:"},{"line_number":5009,"context_line":"                host_cpuset \u003d host_cpuset.union(dest_cell.cpuset)"},{"line_number":5010,"context_line":""},{"line_number":5011,"context_line":"            if not guest_cpuset.issubset(host_cpuset):"},{"line_number":5012,"context_line":"                reason \u003d (_(\u0027Unable to migrate to destation host, cpuset \u0027"},{"line_number":5013,"context_line":"                            \u0027mismatch, guest cpuset: %(guest_cpuset)s, \u0027"},{"line_number":5014,"context_line":"                            \u0027host cpuset: %(host_cpuset)s.\u0027),"}],"source_content_type":"text/x-python","patch_set":2,"id":"7aaa499b_82289a87","line":5011,"in_reply_to":"7aaa499b_115a9e2a","updated":"2015-04-15 13:44:12.000000000","message":"hi Markus\nthanks for your review, I should clarify why I did this change:\n\nif we have an instance which cpuset is [0-15] (by default if you don\u0027t have numa config, the cpuset will be from host cpuset), and we want to live migrate it to a host which only have cpuset[0-3], currently libvirt will deny this when do live migration in libvirt driver, if we manually change the  cpuset from [0-15] to [0-4] before livemigration, we need to change the domani xml, so it will bring inconsistent with instance (the domain xml will be generated from domain and host local configure), I am not sure this approach will be okay. I\u0027d like to discuss this with DP(libvirt driver core and libvirt maintainer). in the dev mailing list\n\nEli","commit_id":"9ce4bb65f86a695a4fea41f8047d41259c9ab4b0"},{"author":{"_account_id":11303,"name":"Markus Zoeller (markus_z)","email":"mzoeller@linux.vnet.ibm.com","username":"mzoeller"},"change_message_id":"1d2a0394d071159f56f1e0f04d72e65fafe2102e","unresolved":false,"context_lines":[{"line_number":5008,"context_line":"            for dest_cell in dest_numa_obj.cells:"},{"line_number":5009,"context_line":"                host_cpuset \u003d host_cpuset.union(dest_cell.cpuset)"},{"line_number":5010,"context_line":""},{"line_number":5011,"context_line":"            if not guest_cpuset.issubset(host_cpuset):"},{"line_number":5012,"context_line":"                reason \u003d (_(\u0027Unable to migrate to destation host, cpuset \u0027"},{"line_number":5013,"context_line":"                            \u0027mismatch, guest cpuset: %(guest_cpuset)s, \u0027"},{"line_number":5014,"context_line":"                            \u0027host cpuset: %(host_cpuset)s.\u0027),"}],"source_content_type":"text/x-python","patch_set":2,"id":"7aaa499b_a21b96c9","line":5011,"in_reply_to":"7aaa499b_82289a87","updated":"2015-04-15 13:55:39.000000000","message":"Thanks for you clarification Eli. I\u0027m part of the team which brought the support for the IBM system z platform (arch\u003ds390x) and we don\u0027t support NUMA so this change could affect us in a negative way (see this review too https://review.openstack.org/#/c/170780/ ). Please add a link to your mailing list discussion, I\u0027m very interested in this topic.","commit_id":"9ce4bb65f86a695a4fea41f8047d41259c9ab4b0"}]}