Red Hat Bugzilla – Bug 892465
"expected downtime" in the output of "info migrate" is always 0
Last modified: 2013-12-18 10:03:26 EST
Description of problem: We have "expected downtime" item in the output of "info migrate"(human monitor) or "query-migrate" (qmp) commands. That is a very good feature for user to know that how long the downtime has to be to finish migration. But currently the "expected downtime" is always 0, that does not provide the expected downtime information. Version-Release number of selected component (if applicable): qemu-kvm-1.3.0-3.el7.x86_64 How reproducible: Always Steps to Reproduce: 1. Boot a guest on src host and also dst host with listening mode. 2. Implement migration. 3. (qemu) info migrate or {"execute":"query-migrate"} Actual results: (qemu) info migrate capabilities: xbzrle: off Migration status: active total time: 4081 milliseconds expected downtime: 0 milliseconds transferred ram: 127790 kbytes remaining ram: 444488 kbytes total ram: 2105728 kbytes duplicate: 383457 pages normal: 31854 pages normal bytes: 127416 kbytes Expected results: "expected downtime" should provide the correct time. Additional info:
Hi, Juan The "expected downtime" will be 0 at the beginning, and when the 'remaining ram' is small, the "expected downtime begins to change. Is this expected? Could you help confirm and close it if not a bug.
it is 0 at the start. But there were a bug upstream (and in RHEL7). Patches are posted upstream to fix this. We are thinking about getting the "initial" downtime value as the one set with migration_max_downtime (much better value than zero).