Closed cr1ll3 closed 2 years ago
hm, that code is executed in the plugin in alloc_image()
, where LINSTOR actually creates the DRBD resource. Looking back in the trace PVE takes a lock and gives the command a fixed amount of 60 seconds to return. There is not that much that happens in that time: creating the DRBD backing devices, writing the *.res files, starting initial sync.
So I would say you have a pretty slow LINSTOR cluster and/or cluster in general and/or storage pools and/or $anything that makes that operation that slow. Usually these 60 seconds are ways more than LINSTOR needs to create the DRBD resource. Otherwise we would have a lot more complaints. Unfortunately I think this is a fixed timeout (Cluster.pm:cfs_lock()
) not under our control.
Sometimes, when attempting to clone an image I get the following error message;
I'm not sure if this is necessarily related to Linstor, but I haven't seen this error with any other storage with proxmox before.
This also leads to volumes being left behind, to cause name collisions and whatnot on the next clone-attempt.