Just had this happen for a third time this week.

Emby stops working, and if I check the server console, the shares are not there.

Last two times I got a diagnostics (attached).

a few days back it was doing a backup from an array to unassigned disk, plus moving some files around at the same time.

And a VM was running with a high itilisation task.

 

Today, it was parity check day, plus I was copying some stuff to an unassigned disk, plus I was deleting a whole load of files.  I stopped the parity check and stopped and started everything again.

 

I suspect it is unassigned disk related.......  I have a sata disk that is a backup for stuff, and with a load of photo  folder renaming, this is hundreds of GB or changes its trying to save.

 

Had to do a hard power off this time today though, as the unassigned disk would not unmount, even though I issued a stop of the user script writing to it.tower-diagnostics-20250301-1430.ziptower-diagnostics-20250227-1953.zip

 

Any help is appreciated.......  

 

Feb 27 19:44:31 Tower shfs: shfs: ../lib/fuse.c:1402: unlink_node: Assertion `node->nlookup > 1' failed.

 

This is the problem, and it's an old issue, are you using both user and disk shares paths with your config/usage?

1 hour ago, JorgeB said:

Feb 27 19:44:31 Tower shfs: shfs: ../lib/fuse.c:1402: unlink_node: Assertion `node->nlookup > 1' failed.

 

This is the problem, and it's an old issue, are you using both user and disk shares paths with your config/usage?

 

Can you explain me those ?  user shares I have loads.  Not sure what a disk share is though.......

 

If its this, then I am not :

 

21 minutes ago, vw-kombi said:

Can you explain me those ?

 

You can access a share by:

 

/mnt/user/<share name> and /mnt/<pool name or disk#>/share

 

Sometimes users use the latter with some container or apps.

20 minutes ago, JorgeB said:

 

You can access a share by:

 

/mnt/user/<share name> and /mnt/<pool name or disk#>/share

 

Sometimes users use the latter with some container or apps.

 

Hmmm - I dont think I every do that.  Is it related to the screenshot I added - showing no disk shares ?

It does seem to happen when I am using the unraid file copying thing that’s built in to the interface now.  And usually something else (like copying to an external unassigned disk.

1 hour ago, vw-kombi said:

Is it related to the screenshot I added - showing no disk shares ?

In part, but you can also use with containers, for example, are they all mapped to /mnt/user/something?

 

 

I just went through the containers and they are all /mnt/user/xxxxx showing on the dashboard docker tab. 

Go to Settings - Global Share Settings - Number of fuse File Descriptors

 

Set that to the maximum value shown on the right, then restart the array and retest.

14 minutes ago, JorgeB said:

Go to Settings - Global Share Settings - Number of fuse File Descriptors

 

Set that to the maximum value shown on the right, then restart the array and retest.

Thanks for that, I need to stop the array to change it, so I will plan for that.

14 hours ago, vw-kombi said:

Thanks for that, I need to stop the array to change it, so I will plan for that.

OK - I have done that, restarting array now.

Next time I am copying / moving I will place close attention.

Ideally an email alert from unraid when all the shares have gone would be a heads up on this.

May look at a user script for that.

Just to close this out, I completed a load of data IO things at the same time :

 

Import to immich

copy to external disk

move stuff around on unraid console.

 

No issues seen, so I suspect the solution given is a fix.

 

Is there an impact of this setting anywhere ?

If not, should it not be increased as a default in unraid at install time ?

8 hours ago, vw-kombi said:

Is there an impact of this setting anywhere ?

There shouldn't be, but also best to keep testing for a few more days to make sure this really helps, it appears to help in some cases, but not all.