Docker containers failing after enabling direct_io?

[Omnicrash]

Ok, so here's a weird one. I just upgraded my server hw, and everything was working fine.

Then I began tweaking stuff, and preparing to enable VMs.

I shut down docker, and moved my docker.img to /mnt/disk1/system, and reconfigured docker.

It started up fine, however most of my containers (Plex, PlexPy, Sonarr, ...) had disk I/O errors. I recreated my entire docker image two times with no effect.

 

I managed to fix it by changing all my containers' config files from /mnt/user/appdata to /mnt/disk1/appdata. However, before this was working fine, and this should be no problem according to this post: http://lime-technology.com/forum/index.php?topic=40937.msg466185#msg466185.

 

One thing I did while tweaking was turning on Tunable (enable Direct IO). Since the post I linked above mentioned FUSE, I think that might have something to do with it. I haven't tested disabling it again myself though. I've disabled it, and it did indeed fix the problem.

[CHBMB]

I've seen tunables causing some pretty funky issues in the past. Why do you want the docker.img and appdata on an array drive any way?

 

Sent from my LG-H815 using Tapatalk

 

 

[Omnicrash]

I don't have cache drives . Anyway, I just confirmed it and submitted a bug report. Turning off direct_io and setting it back to user shares worked just fine.

[Videodr0me]

You might try starting your dockers in "--privileged" mode in order to use path mapped to "mnt/user0/" or "mnt/user/" if direct i/o is turned on.