What I was able to observe:  

• Tried to install a new container (beets, 100ish MB) 
• Container downloaded quickly (gigabit connection) 
• 10 minutes passed to get to this: 
• A few minutes later, it finally finished.   During all of this, my server webui was super unresponsive.   It's been happening a lot lately, I'm wondering if it's anything to do with recent Unraid OS updates (as I don't remember this issue before them, but that is pretty subjective)
• The container shows up in the gui as not knowing if it's up to date or not.  If I force update it, it then figures it out.  Force update happens pretty quickly as (I'm guessing) all downloaded/extracted items are cached. 

 

Diagnostic logs attached.   I'm kinda at a loss here. 

 

nimbus-diagnostics-20240316-1046.zip

Edited March 16, 20242 yr by tiny-e

I'm seeing this as well since updating to 6.12.8. Looks like at least some of my containers are considerably more slow/unresponsive as well. And like you the unraid webui is unresponsive whenever updating a container and takes longer time then usual.

Yeah, I'm not sure what's going on.

Nearly a month goes by....   The behavior still exists.  As my system is unresponsive as I try to pull a docker container now... 

 

Check your Docker image is really on an SSD.

Thanks for the reply -- It is:

...

 

You know, the cache was originally part of a ZFS pool, and one drive died and I've yet to replace it.   Could that be the issue?   If so, can I un-pool the cache temporarily until I have a replacement in place? 

Edited April 5, 20242 yr by tiny-e

Wouldn't know. Is the cache working as expected otherwise?

Also it seems you're using an XFS Docker image, can't make use of the features of the default btrfs, might be contributing. Was there a specific reason to change the default?

Problem still exists.  6.12.10.

Adding, updating, or re-pulling a container make my system unresponsive.   I've tried various things from making sure (I think) that appdata lives only on the cache, switching to file mode in docker (folders also on cache drive), etc.

 

Getting frustrated.   Can anyone help?  I can upload new diagnostics once my server responds to my requests again... 

I had the same issue, turns out something was wrong with my cache pool. Very slow performance. I moved the docker image and appdata to a different ssd pool and the issue was resolved. I'm about to upgrade my hardware and replacing the two ssd's in my cache pool is on the list. When I've done that I will move the docker image and appdata back to the cache poo. Then I'll secure wipe the old ssd's and check if the performance is back to normal and put them to use again. 

50 minutes ago, tiny-e said:

Problem still exists.  6.12.10.

Adding, updating, or re-pulling a container make my system unresponsive.   I've tried various things from making sure (I think) that appdata lives only on the cache, switching to file mode in docker (folders also on cache drive), etc.

 

Getting frustrated.   Can anyone help?  I can upload new diagnostics once my server responds to my requests again... 

on the shares tab, click calculate next to the system share.   If it’s existing on the array then the docker image / files may be significantly slower if they are on the array instead of your pool

Thanks -- I Think I'm good, correct? 

Edited June 20, 20242 yr by tiny-e

Seeing Lots of iowait.. going to calculate appdata (it was set to Cache--->Array, but switched to cache only yesterday, I ran mover after making the change, but maybe there's an issue? 

 

I've been stuck here for like a half an hour:  

1 hour ago, tiny-e said:

Thanks -- I Think I'm good, correct? 

You should make it cache only (no secondary storage) and make sure exclusive shares are enabled in global share settings so that it looks like this, same for appdata.

 

 

  

2 hours ago, strike said:

Then I'll secure wipe the old ssd's and check if the performance is back to normal and put them to use again. 

That's worked wonders for me on a number of occasions lately.

Edited June 20, 20242 yr by Kilrah

Here's where I'm at currently.   Issue still persists.