Constant idle disk usage

[cremor]

A few weeks ago I noticed that my NAS has a higher idle disk usage than it used to have. I can now hear the disk being accessed nearly constantly while this wasn’t the case before. With trial-and-error I found that Plex is the cause for this high disk usage.

I’m not fluid with Linux commands but with the command “iotop” I figured out that the process with the command “dockerd --config-file /var/packages/Docker/etc/dockerd.json” is the one that accesses the disk. It writes around 5 MB per minute. The writes happen every 5-10 seconds.
This only happens if the Plex container is running and also happens if the Plex container is the only running container. So it has to be Plex that is writing something here.

I first noticed this problem with PMS version 1.22.1.4228. But I can’t say which version exactly started it because I didn’t update for some time before (I definitely jumped some versions).
It still happens with the current version 1.22.3.4392.

And to be clear again: This happens even if Plex is idle. No client is connected to the server and the web interface isn’t open.

I quickly checked the PMS log files and found the following two lines which are repeated every 5 seconds in “Plex Media Server.log”:

Apr 25, 2021 16:57:41.132 [0x7f7d3a7fc700] DEBUG - Request: [[::1]:59944 (Loopback)] GET /identity (4 live) Signed-in
Apr 25, 2021 16:57:41.132 [0x7f7d6f07e700] DEBUG - Completed: [[::1]:59944] 200 GET /identity (4 live) 0ms 398 bytes (pipelined: 1)

What is this? Could this be the cause?

Host: Synology DS718+ with DSM 6.2.4-25556
Docker version 18.09.0-0519 (Synology package)
PMS version 1.22.3.4392
Docker image plexinc/pms-docker:latest
The container is running with host networking configured and with parameters according to your readme

[cremor]

Any help for this please?

[Volts]

There’s an aggressive healthcheck built into that docker image.

plexinc/pms-docker/blob/716ed27eca02783562d0f85ca97452b3ffd15b75/root/healthcheck.sh

PMS Plex Docker - Constant disk activities

Ok, I have a new update. It is the Plex container creating the writes, but not Plex itself but Docker and yes it is realted to the healthcheck. Apparently the check itself does not create the io but docker is updating the result of the healthcheck everytime in hostconfig.json and config.v2.json and since Plex is running the healthcheck every 5s it updates every 5s the files. Now that I have figured out where this is originating from I found others having the same issue: and somebody also file…

I’m not sure if the --no-healthcheck option suggested in that last post is a potential solution.

[pshanew]

@Volts

I’m not sure if the --no-healthcheck option suggested in that last post is a potential solution.

I tested here on my plexinc/pms-docker container; adding --no-healthcheck did indeed prevent the debug logging being written every five seconds by the health check. I’ve not tested it extensively, but it doesn’t appear to have negatively impacted the server.

Another option is to disable the server’s debug logging but, of course, this has the impact of limiting the usefulness of logging should a non-easily reproducible problem occur.

[Volts]

I imagine that the aggressive healthcheck would still trigger some amount of Docker logging, even if you disable logging within Plex itself.

[pshanew]

Ah, excellent point and it seems very likely.

[cremor]

@Volts Thanks for linking that thread! I’ll continue there.