Hi.

I've read a lot about the mnt/user problem here in the forum and that you should upload your diagnostics. I found out the following. my shares disappear when i try to delete directories that still contain files. I try to delete them from my PC, but then I get the message: "Delete not possible, directory not empty." The files have been deleted anyway, but the empty directories are still there. If I try again now, the directory is either deleted and everything continues to run normally, or I get the same error again. If that happens, then I can see in the unraid gui that all shares have disappeared. only a restart will help.
what i also noticed is that it doesn't matter whether the directories are on the array or on a single cache_pool hard disk. it seems to happen more often on the cache pool hard drive. I hope I was able to contribute a bit to finding a solution and you can help me. I installed tdarr but never had any problems with it, because of a thread here I uninstalled it, but that didn't solve my problem. I thank you in advance for any support.

diagnostics-20221206-1142.zip

If you don't need it disable NFS, or if used see if you can switch to SMB.

thx for your reply. switch to smb is the solution ? - i use NFS. and i love NFS. 😭

Is there another solution for this issue maybe ? i never had problems before and i use NFS for years already in unraid.

Edited December 6, 20223 yr by joghurt

1 hour ago, joghurt said:

switch to smb is the solution ?

NFS is known to be one of the things that cause this, and it appears to be the reason in your case since the call trace mentions it.

i faced problems with smb shares and honestly spoken i am not familiar with the options to put in my fstab file.
my Client is MX Linux.
Could you maybe please let me know the entry in my fstab ? i mean the options. I would very appreciate it. Thx

Try using the UD plugin to mount the remote shares.

Same problem for me several times now since 6.11.5

Without diags not much we can add.

It happened again. All shares gone. This time I have diagnostics.tower-diagnostics-20221224-0008.zip

Have you tried the suggestions already in this thread? I assume not since you don't even have Unassigned Devices plugin installed. And your syslog has NFS dumps

I don't understand how Unassigned Devices plugin would help with this problem.

"Unassigned Devices allows you to mount and share disk drives that are not managed as part of the array."

 

My NFS shares IS a part of array. I and use them for linux clients. It worked without problem before 6.11.5.

 

I tried use SMB shares for linux clients but performance is not as good as NFS. Works well for Windows clients.
I also tried 9p and virtofs but NFS is faster.

 

The question is why NFS dumps?

 

We can try disabling hard links if not needed (Settings -> Global Share Settings), it helps for some users.

Ok, I will try that

Could it be better to export nfs mount to /mnt/user0/share? even if cache is not used in share? default when enabling nfs on share unraid uses /mnt/user/share. I read something about cache and nfs perhaps could cause problems.

It might be worth trying.

It has been stable now with hard links disabled and nfs exports to /mnt/user0/share. 

When will this BUG finally be fixed? - I tried this workaround from @ryddan. But this does not solve the problem completely. Once in a while it happens again and again and again.... IF you now ask for diagnostic file, I would ask you why this happens since the version where you established the cache pools ?

You tell me to use SMB, but I don't want to use SMB, I want to use NFS, as it is provided by the system.

 

So please when will it finally be fixed? - On other tasks you received already a lot of diags regarding this problem. Thank for qualified answer.

 

 

I ran into this issue as well.  An interesting tidbit that might help others.  

When the /mnt/user disappeared, I always ended up rebooting because neither docker or kvm would work properly after bringing the array off/online.  Upon reboot, the parity check would be triggered but the rate was very slow.  Seeing > 15 MB/sec was rare.  All the benchmarks on the drives showed over 150 MB/sec or higher for each drive so it was rather confusing.  

Disabling NFS sharing on each of the shares and manually putting the nfs shares into /etc/exports using /mnt/user0/... resulted in an increase of parity check performance to 200 MB/sec.  Something is definitely amiss with what the GUI is doing to the shares with NFS.  

YMMV as always