PMS v1.18.1.2019 crashed 2 times in 3 days. Win10 - 503 Service Unavailable

Server Version#: Plex Media Server v1.18.1.2019-c186313fe on Windows 10 Pro 1903 x64
Intel i7 7700, 16GB Ram, NVME SSD.

2 Times now since updating, my PMS has become unresponsive to any plex client, PMS is still running but not usable, i can see its still writing to its logs, but PMS is not in a usable state

last night i finished a episode around Nov 14, 2019 23:52 in the Plex Media Server.log
and then could not play any further episodes infinite loading circle on android tv.

checking what happened it this morning Plex is still running but '503 Sevice Unavailable`

please see logs and screenshot


Logs.zip (14.0 MB)

1 Like

I am having the same issue, as are a number of others
https://forums.plex.tv/t/pms-on-windows-10-keeps-crashing-after-v1-18-1-1973-update/

thanks for linking, but im im not sure from that thread its exactly the same as my issue, my PMS .exe is not actually crashed it is still running, but the plex service is not actually functional

theres no entry in event viewer showing it crashed as some of the posts

alot of the posts in that thread are lacking specifics on whats actually happening to them which doesnt help.

I actually am having the same issue as you where it becomes unresponsive, but is still open. I disabled he transcoding and reverted and now it is at least stable and up, for now. Honestly I couldn’t watch more than an hour without it becoming unresponsive.

were you hardware trancoding the video when it crashed?

Sometimes yes sometimes no.

Mine just went down the exact same way. Annoying…

1 Like

HW encoding crashes - is ĂŻt GPU related, graphics card related, or graphics driver related? I see lots of reports and logs, but the affected group could be narrower than we think.

both times it has crashed for me i was not hardware transcoding at the time, but it was transcoding audio

ping @sa2000 can you please assist with this and the other thread mentioned above regarding Windows 10 crashes

I need the dmp files and the corresponding server logs with debug enabled beforehand

dmps do not always get uploaded to our crash reporting systems and would remain on the server. Also if crash reporting is disabled - see advanced server general settings

Look in the "Crash Reports\…. release identifier … \PLEX-MEDIA-SERVER folder for dmp files
You can zip and send to me privately by PM - but I would need also logs leading to a crash

So best to get logs and dmp from new crash

and after the crash, restart the Plex Media Server and wait few minutes and then capture the logs zip

and then look into the “Crash Reports” subdirectories for any dmp files that failed to get uploaded

1 Like

Just realized there may be several issues here in this thread

The 503 Service Unavailable errors are different from crashes when hardware transcoding is enabled - in fact there would not be a dmp file and we would need to force one - but before that to get list of connections

See this post concerning diagnostics for deadlocks and 503 errors

there are no crash reports in that directory,

and debug logs are in the first post which i copied from when it was “503 Unavailable”, before i restarted PMS

at the moment i have downgraded my PMS to a earlier release so getting the dmp fill be hard unless it does the same thing on this version i installed, but hopefully others can as well

503 Service Unavailable responses start to appear after we exhaust the limit of number of queued requests . This happens after a deadlock eg one thread locking out all others - which could be due to two threads being in a deadly embrace state - having different locks and waiting for one to free up.

Investigation of deadlocks requires getting a list of connections at the time as well as a forced process dump to accompany the debug logs. The forum post I linked to gives the instructions for providing that

Just sent you a PM with a DMP, connection list and logs. Hopefully that helps!

Thank you. Yes it appears to be a lockout and requests started to queue up

How often does it arise ?

I see the server was up for 20 hours before the lockout started

A point of advice on providing logs - never zip the Logs folder in-situ. If you want to do the zipping yourself, copy out the directory you want to zip and then zip the copy. Otherwise we lose the current log files., I do not have for example the latest log file for Plex Media Server but have the previous one Plex Media Server.1.log - But it is ok in this case because the lockout happened before the last log file was created

I am referring this to our development team

This happens around 1-2x per day. The server is then unavailable anywhere from 30 seconds - 30 minutes, there doesn’t seem to be any rhyme or reason to it. The server always recovers eventually and becomes available again if I don’t intervene, or I can restart the host machine and it’s available immediately upon reboot.

Another thing worth noting is that I cannot terminate the process, either with Task Manager or Process Explorer. Trying to do so just doesn’t work the first time, then subsequently results in “Access Is Denied” messages. The only way to end the PlexMediaServer.exe process once locked up is to actually restart the system.

Also, I realized after zipping the logs up that I had made that mistake - so I actually manually copied the Plex Media Server.log (the in-progress file) into that zip before sending it to you. I figured the other log files probably weren’t necessary.

Having the same issue when I upgraded to version 2019. Following but also upgrade to beta a couple hours ago to see if that fixes the issue.