Hello,

I'm trying to use an Unraid SMB share as a storage for ISOs, backups, etc. The issue is I cannot set a cache drive to the share. It will immediately kick an error out in Proxmox saying it cannot find the SMB share, even though it can list it.

 

Here is a view of Proxmox showing details about the SMB share but unable to write to it

 

INFO: starting new backup job: vzdump 100 --prune-backups 'keep-last=2,keep-weekly=1' --all 0 --node PVE-SOL --storage UNRAID-SMB --mailnotification always --notes-template '{{guestname}}' --mode snapshot --compress zstd
ERROR: Backup of VM 100 failed - unable to create temporary directory '/mnt/pve/UNRAID-SMB/dump/vzdump-qemu-100-2023_01_16-18_19_28.tmp' at /usr/share/perl5/PVE/VZDump.pm line 930.
INFO: Failed at 2023-01-16 18:19:28
INFO: Backup job finished with errors
TASK ERROR: job errors

 

If I remove the cache drive from the share it works perfectly fine.

 

 

This may be related, but I tried to use NFS instead and ran into similar results, I didn't try it without the cache drive though.

 

This may also not be related, but I turned on SMB log level to 1 and the console is being spammed with this error:

I am having this issue with NFS and digging in to it as well. Ill give the no cache drive a try and see if anything changes.

What I find odd in my case is that I can add the NFS share to my cluster, see it on both nodes, set up a backup job, run said backup job and everything works great. Leave it for a bit and come back and the storage on both nodes has a question mark and inaccessible. I have another NFS share mounted from another NAS and that has been solid.

Edited January 17, 20233 yr by jtp755

17 minutes ago, jtp755 said:

I am having this issue with NFS and digging in to it as well. Ill give the no cache drive a try and see if anything changes.

What I find odd in my case is that I can add the NFS share to my cluster, see it on both nodes, set up a backup job, run said backup job and everything works great. Leave it for a bit and come back and the storage on both nodes has a question mark and inaccessible. I have another NFS share mounted from another NAS and that has been solid.

That is the exact issue I'm having with NFS. Like I said, I haven't tried it since then without the cache.

 

I am using another SMB share with a cached drive no problem mounted in an Ubuntu VM. So not sure what the hang up is with Proxmox.

21 hours ago, WarlockSyno said:

That is the exact issue I'm having with NFS. Like I said, I haven't tried it since then without the cache.

 

I am using another SMB share with a cached drive no problem mounted in an Ubuntu VM. So not sure what the hang up is with Proxmox.

I have been running for 24hrs with cache turned off on my backup share in Unraid and it has not even blipped in Proxmox. All snapshots have written fine and I can still browse the attached storage from Proxmox. Seems to definitely be something with cache-enabled.

 

I did see quite a few posts yesterday when I was looking in to it add the NFS share in /etc/fstab for automount and then in Proxmox use local filesystem to add the storage. I have not tried this yet though.

I also checked and my backups/snapshots haven't had an issue with cache turned off.

 

As stated before, I have another share with cache enabled that is working fine. In fact, I moved 500GB to it yesterday without an issue.

12 hours ago, WarlockSyno said:

I also checked and my backups/snapshots haven't had an issue with cache turned off.

 

As stated before, I have another share with cache enabled that is working fine. In fact, I moved 500GB to it yesterday without an issue.

So the other share you have mounted in the Ubuntu VM - the VM is running on Proxmox? Im betting that would work as its using the OS which is good news.

 

I am going to change mine tomorrow to have cache enabled but mount the NFS via /etc/fstab and see how that goes. Didnt get to it today.

Yeah I have the other share mounted in a VM on Proxmox, that one has had 0 issues.

I have not found a solution other than disabling cache on the pool in order to get the storage stable and usable in Proxmox host (not specific VMs or LXCs) - just adding it as storage.

Edited January 20, 20233 yr by jtp755

Finally there is a working solution for that problem. You just have to add the following line to the corresponding mount in /etc/pve/storage.cfg

 

options noserverino

 

That's all. I hadn't had an error since 3 days. I just use the share for backup with active cache. I'm on Proxmox 8.2.7

Edited November 2, 20241 yr by mattjes04

On 10/29/2024 at 10:38 PM, mattjes04 said:

Finally there is a working solution for that problem. You just have to add the following line to the corresponding mount in /etc/pve/storage.cfg

 

options noserverino

 

That's all. I hadn't had an error since 3 days. I just use the share for backup with active cache. I'm on Proxmox 8.2.7

I knew this was an issue that I ran into in the past and resolving it was to just change the share to only write to array.

 

But today I was wondering if this was resolved because I'm tired of my Proxmox performing backups at a slow rate because the share is only going to an 'Array'.

 

You sir are my savior today lol. This resolved the issue where I can use a cache drive in Proxmox. Tried to see if rebooting keeps the setting and Proxmox connects to the share just fine. I also performed backups and what a relief, I can definitely tell its faster.