[SOLVED] Docker services extremely slow

[Turnspit]

For the last couple of weeks I've noticed that the overall docker services on my machine are extremely slow.

 

While the dockers itself run perfectly fine performance-wise, the updating process is atrociously slow.

 

The "checking" process runs in an endless loop, I've already let in run for 30+ minutes without any progress. Upon reloading the page, I see which dockers have updates available and which don't, so the process must've finished in the mean time.

Strangely, there's always exactly one random docker that doesn't seem to be able to retreive a status update and shows the "no update available" info (don't know the exact wording at the moment).

 

On top of that downloading updates for dockers takes ages, even packages as small as a few MBs take minutes to download. My overall download speed on other devices/services is perfectly fine though.

 

CA is up to date and runs perfectly fine and fast as well.

I'm not using Pi-hole as a DNS for UnRAID, which uses Cloudflares 1.1.1.1/1.0.0.1 instead. Changing the DNS to the likes of Google or OpenDNS didn't improve anything here.

 

Thanks in advance!

Edited November 15, 2021 by Turnspit

[Squid]

The link you posted returned a 404.  Try again posting to the forum.  Many users including myself do not particularly like clicking random links

[Turnspit]

Uploading working now - Diagnostics attached. 🙂

homeserver-diagnostics-20211114-1955.zip

[Squid]

Nov 14 03:05:23 Homeserver kernel: BTRFS warning (device dm-7): csum failed root 5 ino 8891848 off 495853568 csum 0x53264c22 expected csum 0x265b60b1 mirror 1
Nov 14 03:05:23 Homeserver kernel: BTRFS error (device dm-7): bdev /dev/mapper/sdl1 errs: wr 0, rd 0, flush 0, corrupt 10, gen 0
Nov 14 03:05:23 Homeserver kernel: BTRFS info (device dm-7): read error corrected: ino 8891848 off 495853568 (dev /dev/mapper/sdl1 sector 286184976)

You've got some corruption on your cache pool (ssd-cache)

 

Not sure how much that would affect it though.

 

Do you have a whack of containers that are displaying the question mark for the icon?

[Turnspit]

I've got 17 containers total, and on an "update all" it's always exactly 1 (random) container showing the question mark. Upon a forced update on that container it is then resolved though.

 

I just ran a btrfs-scrub including corrupted-block-repair, which finished successfully without any erros found.

 

Is there anything else I could do?

 

EDIT:

 

It seems that a btrfs-scrub + corrupted-block-repair has fixed the issue (for now).

Checking for updates takes about 10 seconds, download speed is back to normal and no more "?"-status reports.

 

Thanks a ton Squid! 🙂

Edited November 15, 2021 by Turnspit
Update

[JorgeB]

Btrfs is detecting data corruption, running Ryzen with above max officially supported speeds is known to cause data corruption is some cases, see here.

[Turnspit]

2 hours ago, JorgeB said:

Btrfs is detecting data corruption, running Ryzen with above max officially supported speeds is known to cause data corruption is some cases, see here.

 

I am indeed running XMP as well as the stock CPU overclock on my MSI mainboard. Will revert those to default and observe the behaviour. Thanks!