Odd behaviour with Docker Containers

[charibdis]

I've been attempting to troubleshoot odd behavior related to Docker. I'm not discounting the possibility that the issue may lie elsewhere. I had a few orphaned images, the drive on which the dockers were installed is getting up there in age, Therefore I simply assumed it to be an issue with the drive. As I wasn't worried about losing any data for any of the dockers, I removed the Docker image and cleared all saved data so that I could have a clean start...

 

However, I keep having issues with installing any Docker. It will either take 15+ minutes sitting on a blank browser window with no status/details. I seem to be able to open the web UI on a secondary PC during this operation and navigate. But the PC on which the install operation was triggered will not display any other UI page. When it finally finishes, I'll get a partially loaded UI page (the formatting/styles don't seem to get applied. See screenshot) If I click on the Done button it will reload the last page I was on and everything seems to work normally.   

 

Installs usually take a few minutes at most. I've reviewed my network performance to see if anything was routing incorrectly, (this server runs 1 of 2 Pihole dockers, but neither of my 2 unraid boxes are set to use these).

 

I attempted going through the logs, but nothing jumped out at me. 

charybdys-unrai-diagnostics-20221219-1601.zip

[trurl]

Looks like an actual disk problem with cache, which has resulted in corrupt docker.img.

 

Why are you using HDD for this? If it were me I would just put those shares on your other pool and forget about using that HDD for a pool disk.

[trurl]

1 minute ago, trurl said:

your other pool

Sorry I see your other pool is also HDD. What are you using the nvme for?

[charibdis]

I'm using HDD cause it's what I have/had. And I'm moving the docker to a different cache pool to eliminate that disk. I just don't have SSD's in it. (hoping to pick something decent up in boxing day sales). 

 

There also seem to be timeouts reading the interface/css from the server based on logs. Not sure why that is, likely just random unrelated. 

[charibdis]

2 minutes ago, trurl said:

Sorry I see your other pool is also HDD. What are you using the nvme for?

The NVME is passthrough to a VM.

[trurl]

7 minutes ago, trurl said:

Looks like an actual disk problem with cache

 

19 minutes ago, charibdis said:

drive on which the dockers were installed is getting up there in age

13+ years!

 

SMART attributes don't look bad, but it looks like it's been a while since it passed an extended SMART test. You could try that if you want but it might be better to just move those shares to the other pool as I suggested until you can replace that disk with SSD.

 

You will have to recreate docker.img

 

 

[charibdis]

1 minute ago, trurl said:

 

13+ years!

 

SMART attributes don't look bad, but it looks like it's been a while since it passed an extended SMART test. You could try that if you want but it might be better to just move those shares to the other pool as I suggested until you can replace that disk with SSD.

 

You will have to recreate docker.img

 

 

LOL getting my money's worth out of that drive. It's lasted longer than many larger capacity drives I've had. 

 

I've deleted the docker image and set it to use to the cache drive. I've installed 2 dockers and it's behaving much more as expected. I'll see about moving everything else off the drive in question and if I can run an extended SMART test on it. At the very least if it does go, it won't take much with it. The other cache pool is still 2 HDD's but it's better than nothing for the time being. 

 

I do have physical room for a second NVME in this machine. Should have upgraded my keys when that had the rebate a few months ago cause I'm pretty much maxed out on devices. Hindsight... I don't expect this one to be the issue though the other server I'd expect to have issues, as it runs hot, is purring along... 

 

Thanks for your help.