Description of problem: Please add the reason why vm migration fail, in case of inhibitted migration from rhel7.1 to rhel6.6. Version-Release number of selected component (if applicable): rhevm vt13.5 hosts: rhel6.6, rhel7.1 (or rhel7) How reproducible: 100% Actual results: Event: "Migration failed due to Error: Fatal error during migration (VM: <vmname>, Source: <rhel7.1 hostname>, Destination: <rhel6.6 hostname>10.35.117.20)." Expected results: Event should be somthing like: "Migration failed due to Error: Migration between rhel7.x to rhel6.x is inihibitted" (VM: <vmname>, Source: <rhel7.1 hostname>, Destination: <rhel6.6 hostname>10.35.117.20)."
That's maybe a misunderstanding of the cross cluster migration. Everything is allowed, it's not inhibited, it just fails at some point. So what you see is one of the "regular" migration failures
1. In general, if there is a migration failure, and engine has info on the root cause, it could be good if it is reflected to user in the same event, reporting on migration failed. 2. Specifically here, a migration done from rhel7.1 to rhel6.6, via REST, will fail (bug 1154631). Please consider, adding the reason of why migration failed.
well, a proper and accurate migration failure reason is the holy grail we are looking for for a long time:-) Again, it's not failing because we inhibit anything...we don't, hence you get a particular reason, most likely related to the exact qemu-kvm versions and VM configuration I agree "Error: Fatal error during migration" is not descriptive, and we have an existing bug for that. Although, we don't have a solution yet. This is a recurring problem. Francesco, maybe we can do finally something more with the exitcode? At least something like "libvirt call failed" or something like that...that would be the first step to be able to pull some more relevant details from libvirt in this case
Reply to Michal Skrivanek from comment #3) > well, a proper and accurate migration failure reason is the holy grail we > are looking for for a long time:-) > Again, it's not failing because we inhibit anything...we don't, hence you > get a particular reason, most likely related to the exact qemu-kvm versions > and VM configuration > > I agree "Error: Fatal error during migration" is not descriptive, and we > have an existing bug for that. Although, we don't have a solution yet. This > is a recurring problem. Francesco, maybe we can do finally something more > with the exitcode? At least something like "libvirt call failed" or > something like that...that would be the first step to be able to pull some > more relevant details from libvirt in this case I think yes, we could improve reporting leveraging the exitReason as you suggested, at least signaling and reporting a libvirt error. Taking the bug
This bug is flagged for 3.6, yet the milestone is for 4.0 version, therefore the milestone has been reset. Please set the correct milestone or add the flag.
I'd CLOSE-WONT fix this specific issue - unless this is a general issue for improving the migration failure reason reporting.
That's exactly how I understand this bug. Need to create mechanism of reporting detailed error messages about migration failures. Maybe create another bug for this or at least change the bug description?
Moving from 4.0 alpha to 4.0 beta since 4.0 alpha has been already released and bug is not ON_QA.
oVirt 4.0 beta has been released, moving to RC milestone.
this is no longer relevant in 4.0 as we do not support RHEL 6 hosts in any supported cluster level