Loading wheel of death

server-docker

#1

Hi!

I have a problem that some files will stop playing after about 20min on some devices, and the only solution that works is restart of PMS.
It doesnt give me error messages, it just never starts playing. The transcoder spins up so i know that pms is trying to stream the file.

PMS is running the latest plex pass version on docker on a debian host. The image is the official plex image
Clients that i have tested are the android app on a Samsung Galaxy TAB S and the web client in Chrome on a windows machine.

This doesnt affect all my media files, but those that are, are always the same files. To add to this, the same file that might not play on 1 device might work on the other.

I have tried recreate the pms image, removed the whole pms database, reinstall the apps nothing seems to solve this.

I cant see any correlation between the media files either, they are a mix of mkv and mp4 containers with h264 and hevc video streams.


#2

That seems similar with what happens on my dedicated server on internet.
Yours is at home so its all on the same network?
I can play well using the plex.tv page and have the spin when trying to do directly.


#3

Reading on posts below ours seems that my laptop safary and probably your chrome don't play well with the last version yet. I tried with Firefox and can play directly with no problems.


#4

Yes, its all inhouse streaming.
The device that has the most problem is the android device.
I think i started notice these problems after the pms v1.7 or 1.8 update. Didnt have these problems before that.


#5

If you are using the app in android cant help there as i dont use smartphones.
One of my friends has changed from chrome to firefox on his windows laptop and also went well. On chrome seems to play some minutes and drop the connection.
I keep reading comments since day 19 and most of the problems apear to be the players, so if you can try firefox it may be a way until specific player problems are solved.


#6

May I see some logfiles captured immediately after experiencing this problem? (Please make certain Verbose logging is not enabled)

Settings - Server - Help - Download Logs

I'll need the ZIP file it presents you


#7

Took a while to recreate the error i got.
These two logs are after a failed playback.


#8

Your transcoding has the streaming SHAKA errors (which are fixed in 1.9.3)

I also see this error

Sep 28, 2017 01:45:14.335 [0x7f7d823fe700] DEBUG - Completed: [10.0.0.154:52141] 200 GET /status/sessions (10 live) TLS GZIP 5ms 1580 bytes (pipelined: 4)
Sep 28, 2017 01:45:16.001 [0x7f7d72bff700] ERROR - [Transcoder] [eac3_eae @ 0x2a2fdc0] EAE timeout! EAE not running, or wrong folder? Could not read '/tmp/pms-910b574c-4dd5-400c-b35c-802ff978f0b2/EasyAudioEncoder/Convert to WAV (to 8ch or less)/de59f25bc51e3c2-com-plexapp-android_15191-1-3.wav'
Sep 28, 2017 01:45:16.003 [0x7f7d5e3fd700] ERROR - [Transcoder] [eac3_eae @ 0x2a2fdc0] error reading output
Sep 28, 2017 01:45:16.004 [0x7f7d72bff700] ERROR - [Transcoder] Error while decoding stream #0:1: Input/output error

Where is the transcoder temp directory? Based on the errors, it looks like you have it on the LAN.


#9

the transcoder temp folder is locally linked to the docker container

This is my config for the container:
docker create \
--name=pon \
--network=macvlan0 \
--ip=10.0.0.22 \
-h pon \
-e PGID=0 -e PUID=0 \
-e TZ="Europe/Stockholm" \
-e PLEX_CLAIM="" \
-e PLEX_UID=0 -e PLEX_GID=0 \
-v /data/cdb/plex/db:/config \
-v /data/cdb/plex/transcode:/transcode \
-v /mnt/storage/anime:/media/anime \
-v /mnt/storage/movies:/media/movies \
-v /mnt/storage/music:/media/music \
-v /mnt/storage/shows:/media/shows \
plexinc/pms-docker:plexpass


#10

Please explain a few things?

Am I correct you specified /transcode as the transcoder temp directory?
Is it a local or remote filesystem ?

Out of pure curiosity, why are you running PMS in a Docker container when you can run it native on Linux with a lot less hassle ?


#11

'transcode' folder is on a local ext4 fs.

The host is used for more than just PMS, therefore its much cleaner and "easier" to run all the applications in containers.
If i for example wants to reinstall PMS, its just like 3 commands and i have a complete fresh install without any residual files from the previous install which helps with keeping the system "clean".
Im using your official docker image so hopefully there isnt an inherent issue with it.


#12

The only issue is my limited success and skill with Docker.

VM's I understand well. schroot containers are another matter. ;)

I unfortunately need to move this thread to the Docker forum where you can get better support.


#13

It's okay.
There seems to be other people with a similar issue in that sub-forum


#14

I had similar issue and the fix was (is) to avoid using /tmp locally to the container...
You might try this:
docker create \
--name=pon \
--network=macvlan0 \
--ip=10.0.0.22 \
-h pon \
-e PGID=0 -e PUID=0 \
-e TZ="Europe/Stockholm" \
-e PLEX_CLAIM="" \
-e PLEX_UID=0 -e PLEX_GID=0 \
-v /data/cdb/plex/db:/config \
-v /data/cdb/plex/transcode:/transcode \
-v /mnt/storage/anime:/media/anime \
-v /mnt/storage/movies:/media/movies \
-v /mnt/storage/music:/media/music \
-v /mnt/storage/shows:/media/shows \
*-v /tmp:/tmp \*
plexinc/pms-docker:plexpass


#15

@ofthesun9 said:

I actually figured that out too, through testing.
What i did is that i changed the whole transcode folder to tmp.
Which i guess just gives the same result.

Plex should update their readme for the docker image with this information, until they fix the issue with the transcoder.
i/e add another volume parameter like you described.