ZFS Snapshot destroy - dataset is busy - Docker

Hi everyone,

i dont know if this is a Unraid, Docker or Borgmatic problem. (Using RC8)

I want todo the following.

• Create ZFS Snapshot of dataset appdata for borgmatic
  • cache-mirror/appdata
  • zfs snapshot cache-mirror/appdata@borgmatic
• Mount the complete cache-mirror pool into borgmatic container
  • /mnt/cache-mirror to /mnt/cache-mirror
• Try to backup /mnt/cache-mirror/appdata/.zfs/snapshot/borgmatic
  • Problem: If Container is started BEFORE the snapshot is created, i recieve this error when i want to access the folder:
    • dir_open: [Errno 40] Symbolic link loop: 'borgmatic'
      • this happens with the 'ls' command and with borgmatic
  • Solution: If Container is started AFTER the Snapshot is created, it works fine
• Next problem: Every time the container is started AFTER the Snapshot is created (Solution), i cant delete the snapshot afterwards. It doesnt matter if the borgmatic container is stopped or not.
  • cannot destroy snapshot cache-mirror/appdata@borgmatic: dataset is busy
    • Sometimes (couldnt reproduce it every time) when the container is started BEFORE the snapshot is created (when access in the container is not working to the snapshot (.zfs folder), a snapshot destroy is possible.

It is possible i use the snapshot command wrong or should do this another way, but because its ZFS maybe there is another problem in the RCs

 @JorgeB sorry for pinging you directly but we communicated already in the RC7 thread and maybe this is related to the other dataset is busy problems.

Edit:

Can you tell me how i could force remove the snapshot or unmount? I do every time a reboot if the error happens because the force unmount or force delete snapshot commands dont work for me.