Docker containers become unresponsive randomly

[downdrown]

Background

 

I am hosting some services using Docker for myself and my family. I always used my Synology DS 1815+ for that but I was never really happy about it because my NAS would always have "one foot in the open web". Recently I decided to invest into an upgrade and configured a small home server for that purpose. A friend of mine suggested to check out Unraid for that purpose, I found it quite interesting and gave it a shot. I bought the hardware, purchased a basic Unraid licence and moved all the docker containers and data to my new server.

 

Since I use the server only - well, as a server - I do not care about lots of storage. For this I still have a small Synology with HDDs. The server operates on a 3 1TB SSD Array setup because I mostly care about speed and availability.

 

The Problem

 

Unfortunately I kind of had issues right from the start. My Docker containers sometimes randomly become unresponsive and the CPU starts going nuts pushing some threads up to 100% usage. After I restart everything works nice again, until sometime later everything becomes unresponsive again. It's driving me crazy and I do not have any idea what could cause this problem. I first suspected the parity check to cause this problem, currently it doesn't seem like this is the only trigger. It happened again last night and I had to reboot the server this morning.

 

Since the only purpose of this machine is to reliably host Docker containers it's a huge issue for me if this doesn't work out.

 

I'd really like to fix this issue somehow, it's driving me crazy.

Would appreciate any hint & help.

bigboy-diagnostics-20230409-1111.zip bigboy-syslog-20230409-0912.zip

[Squid]

Not your complete problem, but you've got two SSDs (disk 1 & 2) and a spinner (parity)

 

Reads are going to be fast, but writes will be slow.  The docker image and appdata are effectively read and written to constantly, so the write penalty will be huge.  You're going to be better off setting up the SSDs as a cache pool (trim isn't supported in the array).  This will significantly increase your speed / response (orders of magnitude)

 

You've got tons of "invalid opcodes" being logged.  Could be simple program errors, but also could be bad memory.  Run memtest for at least a couple of passes.

 

[downdrown]

Thanks for the hint, Actually these are 3 SSDs (The parity Disk is also a SSD). I also thought about mem issues because this is happening randomly. I disabled the Plex container, that one seems to be causing most of the issues. I'll demount it from the rack and try to run a few memtests for now.