Plex freezing a transcode no matter the client stating out of space

Nov 20, 2022 20:05:08.306 [0x7fb5dc2e1b38] ERROR - [Req#dc617a/Transcode/ep3v6sffxn5xtf1v74zyzvta/01a4e170-44dc-4530-baf1-4bdf798d4a21] Error while decoding stream #0:1: No space left on device

What “space” is it talking about here? I have monitored filesystems on both the server and client and nothing is running out of space.

Server Version#: 1.30.0.6406
Player Version#: 1.57.1.3358-5fd2a504

both of these are linux boxes, but have also duplicated the issue with Plex on roku 7.0 build 19. I have logs but only see the error with verbose logging turned on. An assistance would be greatly appreciated.

Can’t tell from the snippet

  1. Could be inotify slots
  2. Could be disk space on the transcoder temp directory.

Impossible to tell.

ChuckPa, I monitored the temp filesystem, as well as all others, and nothing approached being full. I am unfamiliar with inotify slots. Can you provide methodology to check max quantity and availability?

Here’s the FAQ about inotify.

ChuckPa, that link pointed to an article focused on synology. I am running an Ubuntu instance, so after searching max watches inotify, I found the control file, /proc/sys/fs/inotify/max_user_watches, managed by sysctl. The quantity was already at 65536, and even though I only have 15105 directories, I tried doubling the quantity. This did not help. As it only occurs during a transcode, I changed my temp transcode directory to an area with 100’s of gigs of free space, still no joy. The film in question is ‘Blue Thunder’. The audio track is dolby truehd which does seem to fit your description. The easiest to reproduce spot is at 1:42:27. If I play the ac3 - 2 channel track, it plays fine, but the truehd track locks up at that point. Is there any other pointers/suggestions you could provide, or is there something more than verbose mode I could enable to help diagnose and correct this?

Sorry about that. I have both Syno and Linux bookmarked side by side.

Would you please confirm DEBUG logging is on and recreate then Download the logs ZIP file and attach it?

This is the second case of truehd failing so there’s something happening.

If you’ve looked at your console / other logs, do you see Error 121 being printed ?

Debug is definitely enabled. Would you like verbose enabled as well? Also, just to confirm, I was looking in the correct place with max_user watches?

I don’t need VERBOSE. Verbose generates so much output that your logs last 2 minutes at most.

Problem was reproduced at 20:35 est.
Plex Media Server Logs_2022-11-21_20-36-20.zip (4.8 MB)

Could you confirm the max_user_watches as well?

Thanks.

With PMS running,

can you

  1. turn VERBOSE logging ON.
  2. Immediately repeat the failure (30 seconds at most)
  3. Turn VERBOSE OFF
  4. Download the logs zip again before it’s gone.

I’ll reach out to the dev about this now.

UPDATE:

Roll back (downgrade) to 1.29.2 while Engineering investigates please.

No need for the verbose log files.

ChuckPa, was wondering if this is still being investigated. I noticed and installed another 1.30 release and tried it, but the problem still persists. Downgraded back to 1.29.2 for the moment, but was hoping to see a resolution on this issue.

They won’t “re-release” 1.30.0 but will fix in an upcoming.

Can I get the logs as requested or is it too late for that?

I also have options to turn transcoder debugging levels higher and lower as needed

Not a problem. Will upgrade the version and reproduce in the morning.

I’m sorry, but your posting of Nov 21 at 9:04 led me to believe you didn’t want my logs, but will get them to you.

My family is currently viewing, but I get up at 5am, so can upgrade, reproduce, and have them posted by 7am at the latest.

I’m sorry for that.

Overnight (3am EST) I was helping a user dealing with an overly small root partition on Linux.

We customized the service (override.conf) and it fixed his.

If that is what you’re experiencing, I think we can do the same for you (?)

ChuckPa, here’s the logs you requested with verbose logging on.
Plex Media Server Logs_2022-12-09_05-33-26.zip (3.7 MB)

The error occurred at 5:33am and was replicated at the same spot in the movie ‘Blue Thunder’
at 1:42:26.

I hope this helps.

Chuck, I don’t think that’s my issue. I have 607G available in my root partition. Besides, it work’s perfectly fine under 1.29, just not 1.30.

Can we do the logs thing which captures it going wrong?

Clearly something got changed :frowning:

I turned on verbose logging and recreated the issue. I then performed a download logs. I posted the logs at 5:38am. Did they not capture the issue?