I've attached diags. As it stands right now, its been hung trying to stop the array for a few minutes now. It says its trying to 'sync filesystems'. What have I done wrong?

 

 

 

Edited December 19, 20223 yr by flyize

Share cache full doesn't mean cache is full, that will be logged as "cache disk full", share cache full means a share is hitting the minimum free space set and it's now copying directly to the array.

 

34 minutes ago, flyize said:

As it stands right now, its been hung trying to stop the array for a few minutes now. It says its trying to 'sync filesystems'.

Make sure there's nothing open on a disk, ssh session will be enough.

I was able to get it to reboot, but it seems Unraid thought it was an unclean shutdown, as parity check is running.

 

Thanks, your description of cache disk full makes sense. Do you see any obvious errors that would prevent me from rebooting the server in a normal amount of time. For a bit more info, I'm having some performance issues (slow/no downloads) with my nzbget container that I'm troubleshooting.

Try lowering the RAM cache size, you can do that with the tips and tweaks plugin, using a disk share instead of user share whenever possible should also help.

So I think I now have more questions...

 

I don't see anything in the tips and tweaks plugin about RAM cache. The only thing I see is Disk Cache settings.

 

And while I promise I read your links, I'm not sure I understand how disk shares work. Does that get rid of the fuse filesystem thing that makes cache + array look like one thing? What do I *lose* by switching to disk shares?

Set

vm.dirty_ratio=2
vm.dirty_background_ratio=1

 

13 hours ago, flyize said:

Does that get rid of the fuse filesystem thing that makes cache + array look like one thing?

Yep.

 

13 hours ago, flyize said:

What do I *lose* by switching to disk shares?

You can only use that if the data you are accessing is in a single disk or pool.