Server Version#: 1.25.1.5286
Player Version#: Null
I had 16GB and saw 50% nominal. Randomly (Every month or two) Plex would bloat and use all the RAM until it killed the host. I checked everything else that it could possibly be, now even with 48GB it almost did the same thing. I assumed it would get to a point where it had enough and would stop but that’s not the case it seems.
I have no idea about Linux or Docker in general.
But I do remember this:
If the /tmp folder of the container is not mapped to a real folder on a hard drive, Docker will put everything in tmp into a RAM drive.
If you have an operation going which produces high volume of temporary data (like sonic analysis, loudness analysis, intro detection etc.) this RAM drive can get filled up.
I didn’t realize those processes could be so RAM intensive. I’ll give disabling those options a try and see what happens.
Is there some way to restrict how much RAM it uses? I wasn’t expecting back to back crashes like this after I isolated it from all the other containers it was working with.
If this is indeed the /tmp folder which is growing, then Plex has no way to know that it is actually using RAM when it writes someting into /tmp…
Docker is preventing Plex from recognizing that.
You need to edit your dockercompose file (or whatever it is using) and map the container’s /tmp to a folder on a hard drive.
I’ll try this in the mean time I just realized that this had the knock on effect of corrupting my OS which lives also in RAM. Luckily it only took about an hour to diagnose and fix. Cloud saves saved my butt today.
I tried remapping ‘/tmp’ and It did it again unfortunately. This time I caught it at 20GB before it caused a problem.
It’s not the normal behaviour for plex, it happens only with the 1.25.x.xxxx versions, stick to the version 1.24.5.5173 there’s no problem with this one.
It’s not very rassuring that the same problem occurs in the last 3 versions of PMS and it’s not still fixed…
