[6.12.0-RC3] Docker directory mode creates massive slowdowns

This is not an Unraid issue:

 

https://stackoverflow.com/questions/52387244/how-to-clean-up-docker-zfs-legacy-shares

 

Also see rc3 release notes for more info about recommend practices for this.

Quote

It also sets up snapshots by default

Forgot this, Unriad doesn't enable snapshots, do you mean Docker you do have some zfs plugin installed?

42 minutes ago, JorgeB said:

This is not an Unraid issue:

 

https://stackoverflow.com/questions/52387244/how-to-clean-up-docker-zfs-legacy-shares

 

Also see rc3 release notes for more info about recommend practices for this.

Yes I set it up per the release notes, but looking through the link you posted, maybe this isnt a great idea, seems like the docker zfs storage isnt super ready for prime time...thank you!

Docker on zfs is fine if you use an image instead of a directory.

Ive upgraded from 6.11.5 to 6.12.0-RC3 and one of my 3 appdata 240gb drives has failed.  This is ok, I also have another 240gb and a 512 m2 ssd in.  Ive replaced the failed 240gb with another 512gb same model drive as the other.  

I mention this in case this is the reason for my issues....Any VMs or Docker containers that I have not got started when i click on them, they wont start.  However if I click start all button at the bottom of the pages then they all start and work as expected which is bizarre...?

44 minutes ago, Lunch said:

which is bizarre...?

It is, but looks unrelated, you should post a new thread on the general support forum with the diagnostics.

I waited for the balancing of the new appdata disk cache share to finish and then rebooted it again and still the same behaviour so I have downgraded back to 6.11.5 sorry.  If Im feeling risky Ill try again tomorrow when kids at daycare, but I think it must have been something in this RC(Not that ive tried thew 2 previous I should add) but I went from 6.10.3 early this morning to 6.11.5 stayed there for an hourish to make sure all was stable and 'normal' which it was which is why I then went up to the RC and then thats when it seemed to happen....the start button was there but couldnt be pressed so to speak, same behaviour on the home screen as the docker page 

Edited April 19, 2023 by Lunch

1 minute ago, Lunch said:

I waited for the balancing of the new appdata disk cache share to finish and then rebooted it again and still the same behaviour so I have downgraded back to 6.11.5 sorry.

So I assume you are not using zfs? In that case it has nothing to do with this.

No, so sorry I thought I had posted in general bugs for 6.12, so sorry 😬 disregard!

 

I hadn't noticed this issue, will look.  I've used docker directory on zfs for quite a while too, including up to RC2.  So we now have to use an image, with BTRFS on it?  That seems a bit wrong.

You don't have to use an image, just be aware of the legacy datasets and best practices, release notes has move info.

On 4/21/2023 at 2:21 AM, JorgeB said:

You don't have to use an image, just be aware of the legacy datasets and best practices, release notes has move info.

So I followed best practices, Ill mark this as "fixed" but for people reading this, I disabled snapshots on the docker dataset, that will not prevent snapshots when docker images are updated, and the release notes still dont say anything about managing them, but that info can be found in the link JorgeB posted above. Basically, dont auto snapshot docker images as they are legacy and lead to 5 minute plus slowdowns on the "Main" page when you reach 8000 snapshots or so. I worked around this by setting my default page at login to dashboard so if snapshots got out of hand I could still log in, I dont auto snapshot docker dataset, and I have a user script that automatically cleans up snapshots and docker images that arent cleaned up on its own. That leaves me with around 600 snapshots for around 30 docker applications, and a 8 second load time when I hit the Main page, which is acceptable. 

Changed Status to Closed