I recently setup a linstor 3 node cluster on Proxmox 8.3. I followed these instructions. The linstore drive appeared as expected on all three nodes. However, when I rebooted the 3 nodes the linstor drive disappeared. The controller and satellites are all running. I see the 3 nodes are online. However in the Proxmox GUI I cannot see the linstore drives.
How can I troubleshoot this?
Do you see any resources listed in linstor resource list?
Apparently not. What would cause me to lose the resource and what is the best way to get it back?
Thank you for your help!
How is your LINSTOR controller configured and persisted? Are you using DRBD and DRBD Reactor to make it highly available as suggested in the LINSTOR User Guide?
right now I’m in the initial setup and testing phase and have not made it HA yet. DRBD and DRBD Reactor are NOT setup yet. Once I can get it running correctly, I will make it HA next.
I followed the instructions in the guide provided. I have 3 nodes:
All three nodes are identical and connected over 25g NICs. I assume the setting in the single node controller persist when you reboot it. Is that not the case?
Also, I did not setup any VMs before power cycling all nodes so I assume we should not see anything in the resource list.
It does, LINSTOR stores its database (by default) in /var/lib/linstor/.
I apologize, I was reading this post as though, “Linstor Drives […] Vanish after Reboot”, was referring to virtual disks for VMs, which would be LINSTOR resources.
You’re saying Proxmox UI isn’t listing the LINSTOR storage, correct?
In that case, I would look to see if /etc/pve/storage.cfg was copied (by Proxmox) to each of the PVE nodes, and looks like this:
drbd: linstor_storage
content images, rootdir
controller 192.168.200.4
resourcegroup pve-rg
Maybe check that the pvesm is running.
Seems like it would be, but can you confirm that the configuration is as you would expect? Does restarting the PVE services change anything?
systemctl restart pve-cluster pvedaemon pvestatd pveproxy pve-ha-lrm
Yes I have run that a few times on all nodes. I think I’m going to put this project on hold. It’s becoming clear to me the overhead required to setup and administer a linstor cluster will require more time than I can dedicate to it.
I can see that setting up the next step of this (enabling the HA cluster and managing it in the future) without dedicated support will not be feasible.
Thank you for trying to help me out.