Hi there, this is technically unraid 7.0.0-rc.1, and I asked in discord, and was told to make a forum thread.

I recently (in 6.12.13) migrated all my data from my array to a ZFS pool (raidz2) within unRAID. It works beautifully, but I'm unable to stop the array. When I do, it gets hung up in multiple places.

 

1) Datasets are busy

2) Pool is busy.

 

```

Dec  5 09:39:28 voyager emhttpd: shcmd (806): rmdir /mnt/user
Dec  5 09:39:28 voyager root: rmdir: failed to remove '/mnt/user': Device or resource busy
Dec  5 09:39:28 voyager emhttpd: shcmd (806): exit status: 1
Dec  5 09:39:28 voyager emhttpd: shcmd (809): /usr/local/sbin/update_cron
Dec  5 09:39:29 voyager emhttpd: Retry unmounting user share(s)...
Dec  5 09:39:34 voyager emhttpd: shcmd (810): /usr/sbin/zfs unmount -a
Dec  5 09:39:34 voyager root: cannot unmount '/mnt/akademiya/kubernetes': pool or dataset is busy
Dec  5 09:39:34 voyager root: cannot unmount '/mnt/akademiya/data': pool or dataset is busy
Dec  5 09:39:34 voyager root: cannot unmount '/mnt/akademiya': pool or dataset is busy
Dec  5 09:39:34 voyager emhttpd: shcmd (810): exit status: 1
Dec  5 09:39:34 voyager emhttpd: shcmd (811): umount /mnt/user
Dec  5 09:39:34 voyager root: umount: /mnt/user: target is busy.
Dec  5 09:39:34 voyager emhttpd: shcmd (811): exit status: 32

```

 

If I ssh into the server, and run a `umount -l` on the datasets, they'll stop. I also shut down all NFS/SMB consumers, and lsof, fuser, and such all show there's nothing using the files. The pool just won't export, no matter what I do.

 

If I go in and run a zpool import before the array (and services) start, I can freely export it with no issues. It's only after starting the array, and any services, that it has issues.

 

Eventually I just opt to restart the server, and it eventually uncleanly stops the array, which won't automatically restart because of an unclean shutdown.

 

Reboot in safe mode, disable Docker and VM services, start the array, stop the array, see if that works.

 

 

Array Stopping•Retry unmounting user share(s)...System running in safe mode

 

Dec  5 11:03:08 voyager emhttpd: Retry unmounting user share(s)...
Dec  5 11:03:13 voyager emhttpd: shcmd (202): /usr/sbin/zfs unmount -a
Dec  5 11:03:14 voyager root: cannot unmount '/mnt/akademiya/data': pool or dataset is busy
Dec  5 11:03:14 voyager root: cannot unmount '/mnt/akademiya': pool or dataset is busy
Dec  5 11:03:14 voyager emhttpd: shcmd (202): exit status: 1
Dec  5 11:03:14 voyager emhttpd: shcmd (203): umount /mnt/user
Dec  5 11:03:14 voyager root: umount: /mnt/user: target is busy.
Dec  5 11:03:14 voyager emhttpd: shcmd (203): exit status: 32
Dec  5 11:03:14 voyager emhttpd: shcmd (204): rmdir /mnt/user
Dec  5 11:03:14 voyager root: rmdir: failed to remove '/mnt/user': Device or resource busy
Dec  5 11:03:14 voyager emhttpd: shcmd (204): exit status: 1
Dec  5 11:03:14 voyager emhttpd: shcmd (207): /usr/local/sbin/update_cron

 

Issue still occurs.

and is this done?

1 hour ago, JorgeB said:

disable Docker and VM services

 

I missed that the first time around. Yes, that seems to be working. Disabling docker + vms and rebooting into safe mode seems to let me stop the array. I'm just going to try and reverse this one at a time.

The odd part is that all my docker containers/vms use the cache drive/pool, not the spinning rust.

 

Testing these one at a time, rebooting, but not in safe mode, I can still safely stop the array.

 

I'll continue, and see if it's something in my VMs or my Docker config, or both.

So I have a VM that has the vdisk on `/mnt/user/domains/<dir>` and that seems to stop me from stopping the array.

I also have docker containers that reference `/mnt/user/docker` and that also seems to stop me from stopping the array.

So if either VM or docker service are enabled, you can't stop the array?

That's right. if either are enabled, I can't stop the array.

That's not normal, try moving the VM or Docker image to another storage pool/device and retest.

Everythins is on `/mnt/cache`, and nothing is on `/mnt/akademiya`. I can move them... to Akademiya?

 

Is it better to reference `/mnt/user` or `/mnt/cache`?

If the docker and VM services are using cache they should not cause issues with the other pool, something strange is going on there, 

 

27 minutes ago, LilDrunkenSmurf said:

Is it better to reference `/mnt/user` or `/mnt/cache`?

Try /mnt/cache, but should be the same for this.

I'm just going to stream of consciousness here for a sanity check/rubber ducky:

 

Docker Settings:

Docker vDisk location: /mnt/user/system/docker/docker.img

Default appdata storage location: /mnt/user/docker/

 

VM Settings:

Libvirt storage location: /mnt/user/system/libvirt/libvirt.img

Default VM storage path: /mnt/user/domains/

Default ISO storage path: /mnt/user/isos/

 

I have a docker container (minio) that does access the pool.

I rebooted into safe mode, with VMs and Docker disabled, and I can start the array, but I can't stop it again.

 

Dec  6 14:29:58 voyager emhttpd: shcmd (279): /usr/sbin/zfs unmount -a
Dec  6 14:29:58 voyager root: cannot unmount '/mnt/akademiya/data': pool or dataset is busy
Dec  6 14:29:58 voyager root: cannot unmount '/mnt/akademiya': pool or dataset is busy
Dec  6 14:29:58 voyager emhttpd: shcmd (279): exit status: 1
Dec  6 14:29:58 voyager emhttpd: shcmd (280): umount /mnt/user
Dec  6 14:29:58 voyager root: umount: /mnt/user: target is busy.
Dec  6 14:29:58 voyager emhttpd: shcmd (280): exit status: 32
Dec  6 14:29:58 voyager emhttpd: shcmd (281): rmdir /mnt/user
Dec  6 14:29:58 voyager root: rmdir: failed to remove '/mnt/user': Device or resource busy
Dec  6 14:29:58 voyager emhttpd: shcmd (281): exit status: 1
Dec  6 14:29:58 voyager emhttpd: shcmd (284): /usr/local/sbin/update_cron
Dec  6 14:29:58 voyager emhttpd: Retry unmounting user share(s)...

 

I'm confused, didn't that work before?

It sure did. I'm also very confused.

Something is using that dataset, you can try the open files plugin to see if you can find out what.