Hi @Squid, maybe this is one you might be able to answer. Take a look at the screenshots attached. I get a whole whack of php errors on the plugins page and the a couple on the main page. Diagnostics attached.

zeus-diagnostics-20190828-0858.zip

Strange - nothing stood out in the diagnostics indicating a problem.    Are you sure these diagnostics covered the period when the problem was occurring?

 

The error message in the screen shots suggest that the plugin files that are loaded into RAM during system boot have somehow disappeared.  I would be tempted to just reboot the server to see if the problem goes away as that would reload all plugins as part of the booting process.

Looks like somehow you deleted /tmp or /tmp/plugins

Sent from my NSA monitored device

On 8/28/2019 at 10:01 PM, Squid said:

Looks like somehow you deleted /tmp or /tmp/plugins

Sent from my NSA monitored device
 

Mmmm..... No, not manually. But, I have my Plex transcode dir set to /tmp, but at the time of logs & screenshots nothing was playing so binhex-plexpas resource usage was little to nothing. Confirmed because I just got the same results on the plugins and main page and there is nothing playing at all because I've stopped the docker. 

 

I'm not a linux expert but my understanding is /tmp will write to disk (swap) if it RAM is exhausted right? so some paging but not complete loss of data right?

Edited August 29, 20196 yr by dgwharrison

9 minutes ago, dgwharrison said:

I'm not a linux expert but my understanding is /tmp will write to disk (swap) if it RAM is exhausted right?

Not the way it's set up on unraid.

Ok, so how do I fix it? And can I not use /tmp for plex transcoding in RAM on UnRAID?

12 hours ago, dgwharrison said:

Ok, so how do I fix it? And can I not use /tmp for plex transcoding in RAM on UnRAID?

I would have thought /tmp would work, but /tmp/plex would at least make sure that plex would not interfere with anything under /tmp.    If you do not have enough RAM to handle these transcodes then you will have to give up on using RAM for this purpose and map it to a physical drive (probably something like /mnt/cache/appdata/plex/transcode).

On 8/30/2019 at 10:16 PM, itimpi said:

I would have thought /tmp would work, but /tmp/plex would at least make sure that plex would not interfere with anything under /tmp.    If you do not have enough RAM to handle these transcodes then you will have to give up on using RAM for this purpose and map it to a physical drive (probably something like /mnt/cache/appdata/plex/transcode).

Hi @itimpiIt's never been a problem before. I'll change it to /tmp/plex for now and later if indeed it is a lack of capacity of RAM issue I'll move it to the SSD - just trying to prolong the life of the SSDs if possible.

 

Thing is though, even after a reboot and plex not running the plugins are all still broken - almost 25GB in available RAM. Do you or @Squid know how to fix them?

The files under /tmp/plugins are created each time a update check is performed.

This sounds like you run out of space in /tmp and file creation wasn't possible.

What is the output of

df -h /tmp

 

The files are created at that point, but the folder isn't.  Those errors only appear when /tmp/plugins doesn't exist.  I would surmise that due to a misconfiguration of the Plex template or Plex itself that Plex deleted the plugins folder.  And the diagnostics show that /tmp/plugins does not exist.

 

Change Plex to not autostart in the docker tab and reboot.  This will confirm that it is Plex that is causing the problem.

For some reason when I saved the docker path mapping of /tmp to /tmp/plex it didn't stick because when I checked it again it was back to /tmp /tmp.

 

So anyway I just edited the docker mapping again to /tmp /tmp/plex and rebooted, leaving autostart on for the first run, and hey presto it was fine.

 

So solution is definitely if you want plex to transcode in ram make sure it's doing it in /tmp/plex. 

Actually, ignore last, docker mapping /tmp to /tmp/plex doesn't work on a reboot because the dir doesn't exist. So I fixed it by adding the following to the /boot/config/go file:

# Create /tmp/plex for plex in-RAM transcoding
mkdir /tmp/plex

And also I changed the docker mapping from /tmp/plex to just /tmp and configured plex to use /tmp/plex in transcoder settings.