ERROR - [Transcoder] [ac3_eae @ 0x2fd4f40] EAE timeout! EAE not running, or wrong folder ....

server-docker

#7

Seems there's another cause for he issue. I was using the default mount /transcode as the transcoding folder but EasyAudioEncode was still looking under /tmp. I fixed the issue by mounting /tmp -> /tmp and changing the transcoding folder from /transcode to /tmp.

Hope the develops look at this thread and push a permanent fix for it.


#8

I'm having the same issue:

https://forums.plex.tv/discussion/268624/eae-timeout-eae-not-running-or-wrong-folder#latest

Anything after 1.5.1 seems to have the issue for me.

I'll try to use a kubernetes emptyDir to pull /tmp outside of the container and see if that helps.


#9

I'm consolidating a couple of threads around this issue to this thread, so I thought I would post the workarounds that have been mentioned so far:

The two attempted work arounds seem to be:
1) Increase max_watches:

sysctl -w fs.inotify.max_user_watches=100000

2) Mount the docker container /tmp folder to a host folder (pulling it outside of the docker overlay filesystem)

Alternative Path:
I've also locked my own images at 1.5.1 for now (doesn't seem to have the issue) using the ARG TAG option in a docker file:

ARG TAG="1.5.1.3520-ed60c70d6"
ARG URL="https://downloads.plex.tv/plex-media-server/1.5.1.3520-ed60c70d6/plexmediaserver_1.5.1.3520-ed60c70d6_amd64.deb"

This forces the installBinary.sh script to use a specific version of Plex. Make sure to set both ARGs or the build won't complete.


#10

I'm facing the same issue. What is the status on this issue?
This occured while transcoding AC3 audio to AAC audio when requesting a .mkv video on Android with an AC3 audio track.

Attached my logs where Plex Media Server.log is the relevant one


#11

I am on Plex for Synology version 1.8.1.4139-c789b3fbb


#12

@josomers said:
I'm facing the same issue. What is the status on this issue?
This occured while transcoding AC3 audio to AAC audio when requesting a .mkv video on Android with an AC3 audio track.

Attached my logs where Plex Media Server.log is the relevant one

You have exceeded the system limit on Number of directories that are added to the watch list
```
Sep 03, 2017 01:10:31.827 [0x7f20fbcb0700] DEBUG - [Notify] Now watching /volume1/Music/JT/iTunes/iTunes Media/Music/Soul Creation ft. Kenny Bobien

Sep 03, 2017 01:10:31.827 [0x7f20fbcb0700] ERROR - [Notify] Failed to add watch for /volume1/Music/JT/iTunes/iTunes Media/Music/Soul Creation ft. Kenny Bobien/Sunset Sessions (Deluxe Volume 2) disc 2 (28: No space left on device)

Sep 03, 2017 01:10:31.828 [0x7f20fbcb0700] ERROR - [Notify] Failed to add watch for /volume1/Music/JT/iTunes/iTunes Media/Music/Soul Migrantz (28: No space left on device)

Sep 03, 2017 01:10:31.828 [0x7f20fbcb0700] ERROR - [Notify] Failed to add watch for /volume1/Music/JT/iTunes/iTunes Media/Music/Soul Migrantz/Calling Chicago - EP (28: No space left on device)

Sep 03, 2017 01:10:31.828 [0x7f20fbcb0700] ERROR - [Notify] Failed to add watch for /volume1/Music/JT/iTunes/iTunes Media/Music/Soulfulicious (28: No space left on device)
etc
```

Easy Audio Encoder path would need to be added to the watch list and the limit was reached

See this as example of one thread with info on how to increase the limit
http://forums.plex.tv/discussion/comment/1419397/#Comment_1419397

Edit: actually it is in this thread as well - the post before yours


#13

@josomers said:
I'm facing the same issue. What is the status on this issue?
This occured while transcoding AC3 audio to AAC audio when requesting a .mkv video on Android with an AC3 audio track.

Attached my logs where Plex Media Server.log is the relevant one

Just to give you more info

You have successfully setup watch for 8192 directories and you have about 1000 that failed to get added


#14

If I may augment here? If you need more than 8,192 directories for PMS to watch,

in /etc/sysctl.conf create the kernel tuning entry.

#
# For more information, see sysctl.conf(5) and sysctl.d(5).
fs.inotify.max_user_watches=16384

After making the entry, Reboot.

The above increases from 8192 to 16384.

Caution is advised on systems with < 4GB of RAM. This is because iNotify entries increase the size of the kernel. If too large, nothing will run and DSM will need to be fully reset.


#15

@sa2000 said:

@josomers said:
I'm facing the same issue. What is the status on this issue?
This occured while transcoding AC3 audio to AAC audio when requesting a .mkv video on Android with an AC3 audio track.

Attached my logs where Plex Media Server.log is the relevant one

Just to give you more info

You have successfully setup watch for 8192 directories and you have about 1000 that failed to get added

Very useful feedback! Thank you!


#16

Just so everyone is aware, this issue is still occurring on version 1.9.7.4460. I'm running the official Plex docker container on Ubuntu 17.10. I can't sync anything to my mobile devices that has an EAC3 5.1 audio stream.

Logs attached.

I guess I'll try to mount /tmp somewhere on my local file system when I get home from vacation in a week and report back if it works.


#17

@Mogul345 said:
Just so everyone is aware, this issue is still occurring on version 1.9.7.4460. I'm running the official Plex docker container on Ubuntu 17.10. I can't sync anything to my mobile devices that has an EAC3 5.1 audio stream.

Logs attached.

I guess I'll try to mount /tmp somewhere on my local file system when I get home from vacation in a week and report back if it works.

Is eae process running?

Nov 26, 2017 22:21:23.000 [0x7f16837fb700] ERROR - [Transcoder] [eac3_eae @ 0x18ac180] EAE timeout! EAE not running, or wrong folder? Could not read '/tmp/pms-8f54e2af-71f7-47b2-b9d3-e38a3d987449/EasyAudioEncoder/Convert to WAV (to 8ch or less)/a15c484b-ac3f-4da7-a2fc-e42836700251_7570-1-0.wav'

Best to get logs and test it after restart - that way other issues like number of watched directories could be checked for being over the limit - also can see the attempt to start eae


#18

@sa2000 said:

Is eae process running?

Nov 26, 2017 22:21:23.000 [0x7f16837fb700] ERROR - [Transcoder] [eac3_eae @ 0x18ac180] EAE timeout! EAE not running, or wrong folder? Could not read '/tmp/pms-8f54e2af-71f7-47b2-b9d3-e38a3d987449/EasyAudioEncoder/Convert to WAV (to 8ch or less)/a15c484b-ac3f-4da7-a2fc-e42836700251_7570-1-0.wav'

Best to get logs and test it after restart - that way other issues like number of watched directories could be checked for being over the limit - also can see the attempt to start eae

So I didn’t think about checking for the EAE process before updating the docker image and replacing the container (since I was out of date). After updating, I’m running v1.10.1.4561.

Everything seems to have started up correctly. I tested a couple files with EAC3 and they worked. And in htop I see “Plex EAE Service” running. So that’s weird.

I’m still attaching the new log from restart. Guidance as to how to figure out if I’m over the file watch limit (even though I do not think I am) would be helpful. Also, I’m a linux noob, any help in detecting if the EAE Service is dead and making the container auto-restart would be great, in case this gets flaky. The behavior manifests itself to the roku client as getting stuck at 13% buffering, so if that happens again I’ll check to see if the EAE Service is running to see if I can repro this somehow.


#19

@Mogul345 said:

@sa2000 said:

Is eae process running?

Nov 26, 2017 22:21:23.000 [0x7f16837fb700] ERROR - [Transcoder] [eac3_eae @ 0x18ac180] EAE timeout! EAE not running, or wrong folder? Could not read '/tmp/pms-8f54e2af-71f7-47b2-b9d3-e38a3d987449/EasyAudioEncoder/Convert to WAV (to 8ch or less)/a15c484b-ac3f-4da7-a2fc-e42836700251_7570-1-0.wav'

Best to get logs and test it after restart - that way other issues like number of watched directories could be checked for being over the limit - also can see the attempt to start eae

So I didn’t think about checking for the EAE process before updating the docker image and replacing the container (since I was out of date). After updating, I’m running v1.10.1.4561.

Everything seems to have started up correctly. I tested a couple files with EAC3 and they worked. And in htop I see “Plex EAE Service” running. So that’s weird.

I’m still attaching the new log from restart. Guidance as to how to figure out if I’m over the file watch limit (even though I do not think I am) would be helpful. Also, I’m a linux noob, any help in detecting if the EAE Service is dead and making the container auto-restart would be great, in case this gets flaky. The behavior manifests itself to the roku client as getting stuck at 13% buffering, so if that happens again I’ll check to see if the EAE Service is running to see if I can repro this somehow.

If there is a problem with the transcoder starting then the roku 13% may show. Would need logs for that and when it is arising - perhaps with logs from both sides - PMS and Roku app

With regards to the notifier system limit - that should not be the case for you as you appear not to have automatic library updates enabled

Dec 07, 2017 18:55:55.152 [0x7f92f5ffe700] INFO - LibraryUpdateManager path watching is disabled

#20

Ok, it happened again today. Tried to watch a .mkv with EAC3 5.1 using my Roku and it got stuck at 13%.

I checked in htop and the Plex EAE Service is running. Log was again full of errors about EAE timeout errors, and unable to read a .wav file in the /tmp directory.

@sa2000 - Attached are the PMS and Roku app logs.

After I collected the logs, I restarted the container, and it started working again. Weird!


#21

@Mogul345 said:
Ok, it happened again today. Tried to watch a .mkv with EAC3 5.1 using my Roku and it got stuck at 13%.

I checked in htop and the Plex EAE Service is running. Log was again full of errors about EAE timeout errors, and unable to read a .wav file in the /tmp directory.

@sa2000 - Attached are the PMS and Roku app logs.

After I collected the logs, I restarted the container, and it started working again. Weird!

Will have a look at the logs - but may be you do not have an installation as per the documented setup.

Just noticed you have tcp port 32401 for local network IP - how did that happen ?
192.168.1.230:32401
Are you using two routes to the server?
local ports 32400 and 32401 are reserved for use by plex media server

Can you please check your setup and make sure it is a setup as per the guidelines here


#22

With regards to EAE - when you do the streaming that needs EAE for audio - can you see if the directories are actually being written to and files being created by the EAE process

The directory structure would be below
/tmp/pms-7ffc530d-1e2c-446e-8db7-4f33ca8e8ded/EasyAudioEncoder/
eg

/tmp/pms-7ffc530d-1e2c-446e-8db7-4f33ca8e8ded/EasyAudioEncoder/Convert to WAV (to 8ch or less)/8abb22c9cb126688832b0734995d6e8a_25754-1-0.wav

This is an example - it may be different subdirectory below EasyAudioEncoder

and is /tmp accessible to the docker process ?


#23

I’m running a multi-server environment right now, as I have my original PMS install on my windows machine running alongside my PMS install in docker that is having the EAE issue. The windows PMS server is running on 32400 public, so for my docker install, I’ve set the public port to 32401.

According to the section about manual port forwards in the Troubleshooting Remote Access guide, this should be a valid external port to use. The internal port should still be set to 32400. You can check me, here’s the command I use to create my docker container:

docker run --restart=always \
        -d \
        --name plex \
        -p 32401:32400/tcp \
        -p 3005:3005/tcp \
        -p 8324:8324/tcp \
        -p 32469:32469/tcp \
        -p 1900:1900/udp \
        -p 32410:32410/udp \
        -p 32412:32412/udp \
        -p 32413:32413/udp \
        -p 32414:32414/udp \
        -e TZ="America/Los_Angeles" \
        -e PLEX_CLAIM=$CLAIMTOKEN \
        -e ADVERTISE_IP="http://192.168.1.230:32401/" \
        -e PLEX_GID=$GROUPID \
        -e PLEX_UID=$USERID \
        -h DOCKER-PLEX \
        -v $CONFIGDIR:/config \
        -v $TRANSCODEDIR:/transcode \
        -v /mnt/data:/data \
        plexinc/pms-docker:plexpass

I am a docker newb, but if I understand how docker networking works, the line -p 32401:32400/tcp should be the equivalent of mapping the external port 32401 to the internal port 32400 in the port forwarding documentation. On my router, I have 32401 forwarded to my docker host (which has a static IP set to 192.168.1.230). I’m manually setting my external port to 32401 in PMS settings as well.

Is this not a valid setup? Do I have to make my router map 32401 -> 32400 on my docker host, and then map 32400 into 32400 on the container?


#24

You have misread the documentation

Two errors above

-p 32401:32400/tcp \

and

-e ADVERTISE_IP="http://192.168.1.230:32401/" 

These should be

-p 32400:32400/tcp \

and

-e ADVERTISE_IP="http://192.168.1.230:32400/" 

and separately from this

If you wish your Plex Media Server to be accessible outside of your home network, you must manually setup port forwarding on your router to forward to the ADVERTISE_IP specified above. By default you can forward port 32400, but if you choose to use a different external port, be sure you configure this in Plex Media Server’s Remote Access settings. With this type of docker networking, the Plex Media Server is essentially behind two routers and it cannot automatically setup port forwarding on its own.

So in your router public / wan port 32401 is port forwarded to port 32400 on 192.168.1.230
(note some very basic routers do not allow you to have different wan port from lan port - if this is so, you need to look for alternative router)
and in Remote Access server setting, tick manually specify port and enter 32401 as the public port


#25

Roger. For simplicity’s sake, I just fully decommissioned my old server, created a new container using the updated the docker run command as you indicated, and updated the port forward in my router to 32400 external -> 32400 internal.

After doing this, I managed to break things one more time by playing videos with EAC3 audio streams (last night, I figured out I could reliably break things by playing a video with an EAC3 audio stream, and seeking after play started. Once the Roku got stuck at 13%, the whole server is down, and no client would be able to connect to it until the container was restarted). I was tailing the server log at the time, though, and the EAE timeout error wasn’t occuring. So I restarted the container and have tried to break it again, and haven’t been successful. So for now, I guess I’ll call this fixed. Thanks for the help @sa2000.


#26

Ok @sa2000 it started happening again today. Here’s the command I used to run the container:

docker run --restart=always \
	-d \
	--name plex \
	-p 32400:32400/tcp \
	-p 3005:3005/tcp \
	-p 8324:8324/tcp \
	-p 32469:32469/tcp \
	-p 1900:1900/udp \
	-p 32410:32410/udp \
	-p 32412:32412/udp \
	-p 32413:32413/udp \
	-p 32414:32414/udp \
	-e TZ="America/Los_Angeles" \
	-e PLEX_CLAIM=$CLAIMTOKEN \
	-e ADVERTISE_IP="http://192.168.1.230:32400/" \
	-e PLEX_GID=$GROUPID \
	-e PLEX_UID=$USERID \
	-h DOCKER-PLEX \
	-v $CONFIGDIR:/config \
	-v $TRANSCODEDIR:/transcode \
	-v /mnt/data:/data \
	plexinc/pms-docker:plexpass

Attached is the PMS logs. I didn’t get the Roku logs this time. Restarting the container fixed everything, as usual. I checked in htop and the Plex EAE Service was running.