diff options
| author | Hans Lindgren <hanlind@kth.se> | 2013-04-11 16:05:23 +0200 |
|---|---|---|
| committer | Hans Lindgren <hanlind@kth.se> | 2013-04-13 00:05:34 +0200 |
| commit | fd26589afa2852762575feaae95a0c9b0c091461 (patch) | |
| tree | 02ea5db94f010a3144bf3bb8f3851d4ff568a84f /doc | |
| parent | 05f5106c198a596f22adadab1ddf6929ab9c247a (diff) | |
| download | nova-fd26589afa2852762575feaae95a0c9b0c091461.tar.gz nova-fd26589afa2852762575feaae95a0c9b0c091461.tar.xz nova-fd26589afa2852762575feaae95a0c9b0c091461.zip | |
Make sure confirm_resize finishes before setting vm_state to ACTIVE
Setting it to ACTIVE to early will allow new resize or whatever
operations to start before the current confirm has finished which
might result in all sorts of conflicts.
Resolves bug 1161529 where an intermittent failure in a tempest test
happens because a confirm operation conflicts with a later resize
carried out on the same instance. As the new resize starts to move
around instance disks, the still not finished confirm operation might
fail at recalculating total disk usage due to a missing disk.
Other possible conflicts might be due to system_metadata and/or quota
not being up-to-date at the moment the instance becomes ACTIVE.
Change-Id: I079da4e271469739504fa87e23149d03652d98f0
Diffstat (limited to 'doc')
0 files changed, 0 insertions, 0 deletions