Hello,

i had a hard server crash some days ago and since then i have many problems with my server.

The crash happened while backing up my most important files over night.

Backup procedure:

• SIA renterd providing a S3 bucket

• backrest backing up a zfs disk to SIA

Primarly shfs crashed on the following operations:

• Creating/Renaming/deleting files inside a zfs dataset over SMB

• mounting/rescanning my Media Library in plex

Maybe you can see any problems regarding corruption or other stuff in the diagnostics. I did not have these problems before the crash, so for me it seems its the most likely cause for this.

Maybe i have to recreate my Cache pools?

Right now im transferring files from the server to prevent loss of important data.

Heres a summary by codex after analyzing the logs:

I am seeing repeated /mnt/user failures caused by shfs crashing, not by an obvious disk, XFS, or ZFS I/O error.

Primary crash:

• Apr 12 12:43:38 Unraid shfs: shfs: ../lib/fuse.c:1402: unlink_node: Assertion 'node->nlookup > 1' failed.

Immediate context at the time of the crash:

• Apr 12 12:43:38 Unraid smbd: Could not close dir! fname=binhex-sabnzbdvpn/Neuer Ordner, fd=-1, err=107=Transport endpoint is not connected

Immediate follow-up errors:

• rsyslogd starts failing to write to /mnt/user/syslogs/... with Transport endpoint is not connected

• Samba starts failing for shares under /mnt/user, for example:

  • canonicalize_connect_path failed for service appdata, path /mnt/user/appdata

  • vfs_ChDir(/mnt/user/appdata) failed: Transport endpoint is not connected

• emhttpd later reports:

  • malloc_share_locations ... Operation not supported (95): getxattr: /mnt/user/syslogs

Later affected shares included:

• appdata

• Cryptomator

• previously also Documents, Walli, and others

This does not look like a one-off event. I found the same shfs assertion crash multiple times:

• Apr 9 19:11:10

• Apr 11 23:29:39

• Apr 12 00:35:15

• Apr 12 12:43:38

A previous crash at Apr 12 00:35:15 also happened during active file access over SMB and produced the same Transport endpoint is not connected fallout.

My current conclusion:

• The primary failure is shfs itself crashing

• /mnt/user then becomes broken/stale

• SMB and rsyslog errors are secondary effects

• In my case, SMB operations on paths under /mnt/user seem to be a likely trigger

Additional note:

• Array disks, cache, and pool mounts themselves appeared healthy in the logs

• I did not find matching primary XFS/ZFS/media I/O errors around the crash points

unraid-diagnostics-20260412-1300.zip

Have you done a memtest recently? RAM issues can cause that type of error.

7 minutes ago, itimpi said:

Have you done a memtest recently? RAM issues can cause that type of error.

Hey, thanks for reply. not yet. One thing to consider.

Can you see any hints regarding corruption?

I am a bit concerned about possible data loss, because since the crash I currently cannot reliably access my backups.

To reduce the risk, I already started making an inventory of where important data is already backed up, so I can verify that nothing critical is missing and identify gaps before I make bigger changes to the system.

At the moment my main concern is not confirmed filesystem corruption, but that the repeated /mnt/user / shfs failures make access unreliable enough that I do not want to assume my backup situation is fully safe until I verify it manually.

Already made mistakes, and i want to prevent more mistakes.

Going through memtest now, 40% no errors yet

1st pass Passed. Will leave it running through the night.

18 hours ago, itimpi said:

Have you done a memtest recently? RAM issues can cause that type of error.

So, Ram does not seem to be a problem here.

The only running containers the last time /mnt/user got dropped and access /mnt/user were

Plex, Sonarr, Radarr, Lidarr, Immich.

Stopped all other containers from autostarting.

20 hours ago, MrSliff said:

shfs: ../lib/fuse.c:1402: unlink_node: Assertion 'node->nlookup > 1' failed.

This is an old issue affecting some users if NFS is used with FUSE (user shares), and it can happen just by having the NFS service enabled, even if no shares are being exported.

Workarounds are disabling the NFS service if you can use just SMB, or using NFS with exclusive shares only.

Ok, thanks. I will try that out!