New node in a cluster can't use all of its storage

[larsen]

Hello,

we have been running Proxmox for some years on our first server and have now just started to use a second server. The cluster is set up and running, but I'm confused about how it works with storage now. PVE 6.1-8

Server 1:
1 TB disk space
"local" storage: 800 GB

Server 2:
8 TB disk space
"local" storage: only 94 GB

This has been configured automatically on installation. I noticed that server 2 has a lvm-thin pool "data" with ~7 TB, but this doesn't show up as a storage.

Could anyone please explain this to me?
Why only 94 GB for "local" and a lvm-thin pool, but no storage?
How would I migrate a VM that has a disk with more than 94 GB?


Lars

 

[tim]

We create a lvm-thin storage when you install without changing the defaults. Did you already add the new node to a cluster with the old one?
I guess that you didn''t check the storage configuration after clustering.

 

[larsen]

Yes, the new node has already been added to the cluster. The state I described in my initial post is after clustering.

 

[tim]

please post your /etc/pve/storage.cfg
I guess it has been overwritten during the cluster join, try to re-add the lvm-thin from your second node to the cluster storage.cfg. This can be done via GUI as well. Mark the storage as only available on your new node.

 

[larsen]

Via "Storage > Add > LVM-Thin"?
When I click on the dropdown for "Thin Pool", it's scanning, but doesn't find anything.

atl-vm03:~# cat /etc/pve/storage.cfg
dir: Backup
    path /var/lib/vz/dump
    content backup
    maxfiles 2

dir: local
    path /var/lib/vz
    content images,iso,vztmpl,rootdir
    maxfiles 0

 

[tim]

yes, where you connected to the right node?

 

[larsen]

No, I thought this wouldn't make a difference. Apparently it does in this case. It's added now.

I want to migrate a VM from node 1 (where it's using a local storage (type "directory")) to the new node. Will I be able to configure it to use the lvm-thin storage then? Or how would I continue?

 

[tim]

You can do a live migration with a live storage migration from local to target node with lvm-thin, this is as well possible via GUI.
Please make sure you have a backup, because your migration needs a storage conversion as well.

 

[larsen]

Would I start this migration on a specific node?

 

[tim]

No this should be proxied to the required nodes automatically, just select the VM from GUI and click migration. (VM should be running)

 

[larsen]

Migration was fine, but I get an error when I try to "Move disk" to switch from local storage to lvmthin (using GUI on first node; VM is on second node):

Virtual Environment 6.1-8
Search
Virtual Machine 108 (atl) on node 'atl-vm03'
Server View
()
scsi0
lvmthin
Raw disk image (raw)
create full clone of drive scsi0 (local:108/vm-108-disk-1.qcow2)
  Rounding up size to full physical extent <465.77 GiB
  Logical volume "vm-108-disk-0" created.
transferred: 0 bytes remaining: 500109934592 bytes total: 500109934592 bytes progression: 0.00 %
transferred: 500109934592 bytes remaining: 0 bytes total: 500109934592 bytes progression: 100.00 %
transferred: 500109934592 bytes remaining: 0 bytes total: 500109934592 bytes progression: 100.00 %
  WARNING: Device /dev/dm-6 not initialized in udev database even after waiting 10000000 microseconds.
  WARNING: Device /dev/dm-6 not initialized in udev database even after waiting 10000000 microseconds.
  Logical volume "vm-108-disk-0" successfully removed
  WARNING: Device /dev/dm-6 not initialized in udev database even after waiting 10000000 microseconds.
TASK ERROR: storage migration failed: command '/sbin/lvs --separator : --noheadings --units b --unbuffered --nosuffix --options lv_size /dev/pve/vm-108-disk-0' failed: got timeout

 

[larsen]

Will open another thread for the last problem as the original thread topic has been resolved with comment #7