Fix strict-local volume restore stuck in attaching state #2049
Conversation
For a volume being restored, the spec.nodeID is empyty but the status.ownerID is set. Thus, set r.Spec.HardNodeAffinity to volume.status.ownerID instead. Longhorn 6239 Signed-off-by: Derek Su <derek.su@suse.com>
derekbit
force-pushed
the
strict-local-restore
branch
from
98fddf0
to
b46fb5e
Compare
|
@mergify backport v1.5.x |
🟠 Waiting for conditions to match
|
|
@mergify backport v1.4.x |
🟠 Waiting for conditions to match
|
| if v.Spec.NodeID == "" { | ||
| continue | ||
| if v.Status.RestoreRequired { | ||
| r.Spec.HardNodeAffinity = v.Status.OwnerID |
There was a problem hiding this comment.
@derekbit Can you explain where v.Status.OwnerID will be set when the volume is being restored? so it can be used here.
There was a problem hiding this comment.
For a volume being restored, the volume.spec.nodeID and volume.status.ownerID are empty, and the later one is determined in https://github.com/longhorn/longhorn-manager/blob/master/controller/volume_controller.go#L295.
There was a problem hiding this comment.
I think this has been fixed in the master and 1.5.x code base already as below. The node id will be set by vol.Status.OwnerID.
/controller/volume_restore_controller.go#L190-L194
if vol.Status.RestoreRequired {
createOrUpdateAttachmentTicket(va, restoringAttachmentTicketID, vol.Status.OwnerID, longhorn.TrueValue, longhorn.AttacherTypeVolumeRestoreController)
} else {
delete(va.Spec.AttachmentTickets, restoringAttachmentTicketID)
}The fix here is more for 1.4.x instead?
|
Ah, yes. You're right. I didn't notice the fix in v1.5.x and master. So, the fix targets v1.4.x. |
|
then, let's change the target branch is 1.4.x instead. 🙂 |
|
Target for v1.4.x only |
For a volume being restored, the spec.nodeID is empyty but the status.ownerID is set. Thus, set r.Spec.HardNodeAffinity to volume.status.ownerID instead.
longhorn/longhorn#6239