125% iowait when mover is running

[MelodyKH3]

Hi; I have had a persistent issue with unraid. The mover causes extremely high iowait; here is some logs from swarmprom (grafana + friends)

 

This has persisted me resetting everything and creating a new array.

 

I have a ST8000DM004 (Seagate Barracuda Compute) as parity (though I think it may be SMR...) and ST8000VX0022 (SkyHawk Surveillance) as the only array drive. There is also a 256gb SSD as the cache from which the mover is moving files.

 

When the mover runs; the web ui somewhat works (the docker page doesn't load; or loads very slowly) but none of my containers can be reached due to timeouts. As far as I can tell my appdata folder (where all my containers run out of) should be on the cache drive and unaffected by this.

 

I tried adjusting the priority of the mover with mover tuner setting it to low and this seemed to help a little bit but still causes it to be unusable.

 

Does anyone have any tips for diagnosing this issue or making it less severe? I have attached the diagnostics.

sol-diagnostics-20200429-1218.zip

[JorgeB]

System share which contains the docker image exists on both cache and disk1, so can't see where the image is, there have been some reports that being on the array might cause issues, so confirm where it is, post the output of:

 

find /mnt -name docker.img

 

[MelodyKH3]

Thanks for your response. It was on the array; was a pain to move back (had to disable docker service; somehow the file docker.img was still open; ended up having to reboot the server because fuser couldn't work out what had the file open; then mount hte arrays, move the files; enable docker again)

It's now on /mnt/cache as you can see here (i also set system and appdata to be cache only)
 

root@Sol:~# find /mnt -name docker.img
/mnt/user/system/docker/docker.img
/mnt/cache/system/docker/docker.img

This mostly resolved the issues; I can now load emby etc while the mover is running; so no more unexpected down time mostly.
While investigating I noticed all IO seems to go through loop2, is this normal and would you be able to show me any documentation or explain what it is for me? I can't seem to find much on it sorry.

 

Thanks!

[JorgeB]

5 hours ago, MelodyKH3 said:

While investigating I noticed all IO seems to go through loop2, is this normal and would you be able to show me any documentation or explain what it is for me?

loop2 is usually where the docker image is mounted, you can confirm with:

 

df | grep docker