Package: ganeti Version: 2.11.6-1~bpo70+1 Severity: important Dear Maintainer,
this at least happens with the extstorage interface and the bug from me yesterday, #782073, though it appears it may happen with all failures pre-migration: Ganeti does not clean up. In this case it does not undo the disk changes done (which led to the other bug), so the exported disk is still on the host it had selected when trying the last migration. At a minimum this is unclean - having the disks exported to two hosts all the time can lead to errors - but it also refuses to work with migrations later on: Should the hbal command select the same ganeti host for the VM where a pre-migration failed earlier on, even though the host is fine now to take the VM[1], this particular VM will always fail. As the disk is already exported, another attaching fails (as the attach sensibly just tells "here is the disk already", not redoing the attaching). IMO Ganeti should, on failed migrations, run all the commands neccessary to undo whatever actions it already took. At least up to the point where the actual migration (memory transfer & co) starts. [1] in this case no longer so many md devices around, and other VMs migrate easily to it -- bye Joerg -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org