Docker and VMs broken Shares missing

[jmztaylor]

No major changes.  Just tried to start a VM and it failed to start. Upon reboot all shares are missing but data is on each disk correctly.  BTRFS came back clean on the pool.

tower-diagnostics-20220712-1139.zip

[JorgeB]

Jul 12 11:36:40 Tower emhttpd: error: share_luks_status, 6151: Operation not supported (95): getxattr: /mnt/user/lxc

This appears to be the problem, you should not create any mountpoint in /mnt/user, those should be reserved for Unraid array/pools only.

 

Also I believe ich777 recommendeds not using a FUSER path for LXC for performance reasons, use /mnt/cache/lxc or /mnt/disk#/lxc

[jmztaylor]

1 minute ago, JorgeB said:

Jul 12 11:36:40 Tower emhttpd: error: share_luks_status, 6151: Operation not supported (95): getxattr: /mnt/user/lxc

This appears to be the problem, you should not create any mountpoint in /mnt/user, those should be reserved for Unraid array/pools only.

 

Also I believe ich777 recommendeds not using a FUSER path for LXC for performance reasons, use /mnt/cache/lxc or /mnt/disk#/lxc

 

So I figured it out.  Somehow that folder got created on /mnt/user/  It wasn't the lxc plugin.  My lxc path is /mnt/cache/lxc.  Dunno what happened but nuked the folder /mnt/user/lxc. stopped the array and restarted it.  It all came back normal. Not sure what exactly happened with that folder.

[itimpi]

4 hours ago, jmztaylor said:

So I figured it out.  Somehow that folder got created on /mnt/user/  It wasn't the lxc plugin.  My lxc path is /mnt/cache/lxc.

It is probably not that simple.   The ‘mnt/cache/lxc’ location is part of the ‘/mnt/user/lxc’ share as the user share is a composite view that includes the ‘lxc’ folder on any array drive OR pool drive..

[trurl]

10 minutes ago, itimpi said:

The ‘mnt/cache/lxc’ location is part of the ‘/mnt/user/lxc’ share as the user share is a composite view that includes the ‘lxc’ folder on any array drive OR pool drive

And so if /mnt/cache/lxc exists, then /mnt/user/lxc exists.

 

/mnt/user/lxc is the lxc user share.

 

/mnt/cache/lxc is the portion of the lxc user share on cache.

[jmztaylor]

2 minutes ago, itimpi said:

It is probably not that simple.   The ‘mnt/cache/lxc’ location is part of the ‘/mnt/user/lxc’ share as the user share is a composite view that includes the ‘lxc’ folder on any array drive OR pool drive..

 

Yeah thats why it was confusing.  But the contents of /mnt/cache/lxc was not the same as /mnt/user/lxc.  /mnt/usr/lxc was emtpy and /mnt/cache/lxc was not.  I think the /mnt/user/lxc was created by a mistake or something.  And when I rebooted it couldn't link /mnt/cache to user and that caused everything to break.  

[Arbadacarba]

Would that happen if you had a share called lxc and had it set to the wrong "Use Cache Pool" setting?

[trurl]

15 hours ago, Arbadacarba said:

Would that happen if you had a share called lxc and had it set to the wrong "Use Cache Pool" setting?

That setting just controls which pool new files are written to, and which pool mover works on. All pools and all array disks are part of user shares.

[trurl]

16 hours ago, jmztaylor said:

Yeah thats why it was confusing.  But the contents of /mnt/cache/lxc was not the same as /mnt/user/lxc.  /mnt/usr/lxc was emtpy and /mnt/cache/lxc was not.  I think the /mnt/user/lxc was created by a mistake or something.  And when I rebooted it couldn't link /mnt/cache to user and that caused everything to break.  

If you had no pool named 'cache' (offline for some reason for example), and wrote to the path /mnt/cache, then a folder named 'cache' in /mnt would have been created, and that path would be in rootfs (RAM) and not part of user shares.