Plex Deadlocking with to many hung connections issues for last 5 months Now!. (This ended up giving me anxiety attacks!) (Edited title for clarification)

Plex Media Server Desktops & Laptops
206

Has anyone else here put their Plex server behind an nginx proxy (using subdomains)? and then subsequently stopped having this problem? If your answer is no I suggest you try it out to see if things improve.

207

Do you have any instructions for that? Iā€™m willing to try anything to be stable again.

208

Are you using Linux?

209

Iā€™m using unRAID with Plex running in docker.

210

Yea, I dunno about unRAID. Iā€™ve never used it. But if you can get nginx installed I will post the config.

211

And people wonder about Plex support?

An issue that is originally and still is tagged as Windows and yet there is hardly a post about Windows since. Now weā€™re on which variation of Linux/Docker and unraid.
Hold your head in shame Plex. :joy:

Q. What are logs.
A. That thing that someone else can be bothered to provide on the relevant OS

212

I was having the same issue on CentOS until I 1) got rid of all Android clients and 2) put my servers behind nginx. So what may have originally been about windows is more widespread than it.

213

Yeah and I havenā€™t read the whole thread to be honest.
To be fair anything involving Android clients seems seriously screwed as a random forum browser.
I didnā€™t (but may have missed) the correlation.

I guess itā€™s why my Shield (next to the Apple TV) needs dusting downā€¦ well plugging in actually.

214

Hi

Hi - I am still waiting for you to have a look at my post

Plex Deadlocking with to many hung connections issues for last 5 months Now!. (This ended up giving me anxiety attacks!) (Edited title for clarification) - #195 by sa2000 and look at the windows event log for any disk issues and also doing a surface read test and chkdsk on the drive

215

It would be a kill -SEGV <pid> where <pid> is the process id of the PlexMediaServer process

But our dump uploads do not always succeed and you may need to search for the dump file after the attempt to upload - search within the docker files/directories

This is an example of me doing this in docker under windows wsl (as a test)

sudo ps aux

USER       PID %CPU %MEM    VSZ   RSS TTY      STAT START   TIME COMMAND
sa       13518  0.4  0.2  99340 60160 ?        Ssl  Jul26  37:36 /usr/lib/plexmediaserver/Plex Media Server

The Plex Crash Uploader logs would have the dump identifier after a process is killed with a -SEGV

I found a copy of the dump file in my case in this path here

 sa@DESKTOP-87HJ4EP:~$ sudo ls -ail /var/lib/docker/overlay2/68ffa8d2b5fd18fce4f6a48b943dd1fc9f33fd6a72ec7dccd9ba979c51e54230/diff/tmp
total 424
101508 drwxrwxrwx 4 root root   4096 Jul 26 09:53 .
101382 drwxr-xr-x 6 root root   4096 Jul 26 02:00 ..
103433 -rw------- 1 sa   sa   415768 Jul 26 09:53 2492d061-aa66-4549-1bcafe94-ed01d014
101510 drwxr-xr-x 2 sa   sa     4096 Jul 26 02:00 pms-6a06576e-557d-4db5-b3ee-2817c99066ac
103428 drwxr-xr-x 2 sa   sa     4096 Jul 26 09:53 pms-703d7052-e34b-410a-9262-3e466fc93e70

The dump file was 2492d061-aa66-4549-1bcafe94-ed01d014 1

as an example

1 Like
216

Experiencing deadlocks tonight with 300+ connections. Tried killing the pid inside the docker console and got this error

# ps aux
USER       PID %CPU %MEM    VSZ   RSS TTY      STAT START   TIME COMMAND
root         1  0.0  0.0    204     4 ?        Ss   13:39   0:00 s6-svscan -t0 /var/run/s6/services
root        38  0.0  0.0    204     4 ?        S    13:39   0:00 s6-supervise s6-fdholderd
root       338  0.0  0.0    204     4 ?        S    13:39   0:00 s6-supervise plex
abc       5419 95.7  0.3 323896 201848 ?       Ssl  19:59  99:15 /usr/lib/plexmediaserver/Plex Media Server
abc       5454  0.1  0.0  62184 41740 ?        SNl  19:59   0:07 Plex Plug-in [com.plexapp.system] /usr/lib/plexmediaserver/R
abc       5510  0.0  0.0  42236 10384 ?        Sl   19:59   0:04 /usr/lib/plexmediaserver/Plex Tuner Service /usr/lib/plexmed
abc       6216  1.6  0.0   2564  1172 ?        S    20:20   1:23 Plex EAE Service
root      9683  0.0  0.0   2616   608 pts/0    Ss   21:42   0:00 sh
root      9689  0.0  0.0   8900  3272 pts/0    R+   21:43   0:00 ps aux
# kill -SEVG 5419
sh: 2: kill: Illegal option -S
# kill -SEVG 5419
sh: 3: kill: Illegal option -S
# kill 5419
217

Itā€™s -SEGV not -SEVG. I also see the CPU was spiking while the deadlock happened. Bet this has something to do with database querys. For some reasonā€¦

218

@sa2000 Iā€™m still adding crash reports and dmp+logs+connections into that Onedrive for 1.24.1.4931-1a38e63c6

219

Is this somehow related to the deadlocks?

New BETA PMS Version Available - 1.24.2.4973-2b1b51db9

  • (Library) The server could become unresponsive if certain HTTP requests were made simultaneously (#12844)
220

passably dose sound interesting

221

It is the fix for the issue i said we discovered - so i recommend running that version. There was no clear indication from the ones I have seen in this thread that this was the cause - but best to run with that fix

222

SEGV not SEVG

223

@Lo-ki ping - please see my earlier response

224

Best to hold fire on adding more. I would like to see more examples from different users

225

Is this issue fixed for anyone in the latest release?