Jun 24 04:24:31 MediaServer kernel: ffprobe[371562]: segfault at 330 ip 000014bcf74eb822 sp 00007fff6e17d8e0 error 6 in libavcodec.so.60.31.102[2eb822,14bcf7211000+b1a000] likely on CPU 4 (core 2, socket 0)
Jun 24 06:52:20 MediaServer kernel: ffprobe[595425]: segfault at 330 ip 0000152f8daeb822 sp 00007ffe5f2ef870 error 6 in libavcodec.so.60.31.102[2eb822,152f8d811000+b1a000] likely on CPU 6 (core 3, socket 0)

I have been having this error populate regularly for the last month or so. Usually it is CPU 6, but core and socket change.

I don't know if there is correlation, but all Plex users have been having playback issues for approximately the same time.

Perhaps also related, the syslog has been slowly filling up, I feel like every time I chase down a fix, it ends up being nothing.

Have performed memtest86 and passed with 0 issues.

I will provide a diagnostics once it is finished downloading, but is taking a long time.

I have tried searching all over for a fix or a cause, but nothing has come up. Any help would be greatly appreciated.

run mem test. this looks like memory contoelr issues on the CPU hopefuly just early signs of ram failure.

segfaults are in ram / memory

47 minutes ago, bmartino1 said:

run mem test. this looks like memory contoelr issues on the CPU hopefuly just early signs of ram failure.

segfaults are in ram / memory

Nah, especially since they've already run memtest

16 hours ago, Tanker said:

I don't know if there is correlation, but all Plex users have been having playback issues for approximately the same time

looks like a broken file crashing the scanner/transcoder, especially if it's the same ffprobe process causing the segfaults. turn off plex for a few days or so and see if it goes away - you can also try deleting the codecs folder in your plex container so it re-builds. if it happens only when plex is running, you'll have to find the file causing the problem

It's plex related for sure. I sometimes see these errors too usually when a transcode goes haywire.

6 hours ago, bmartino1 said:

run mem test. this looks like memory contoelr issues on the CPU hopefuly just early signs of ram failure.

segfaults are in ram / memory

I ran memtest with 3 passes and no issues, I was hoping it would just be some ram sticks to replace and not the CPU.

5 hours ago, Michael_P said:

Nah, especially since they've already run memtest

looks like a broken file crashing the scanner/transcoder, especially if it's the same ffprobe process causing the segfaults. turn off plex for a few days or so and see if it goes away - you can also try deleting the codecs folder in your plex container so it re-builds. if it happens only when plex is running, you'll have to find the file causing the problem

I can try the codecs folder first and hopefully that can work, any ideas on how to quickly track down the problem file? I have nearly 70,000 videos in my library.

If memetest is passing then its not the cpu memory controler / ram hardware its software...

For some reason... forum downloads apear to be broken for me... would lik to have taken a look at your diag file...

If plex is happen befre the seg faults if could be a memory overflow issue...
it where you mount your config folder..

Plex config = /mnt/user/appdata/plex... / config

/Library/Application Support/Plex Media Server/Codecs

stop the plex docker and delete the contents of this folder. Plex will redownload the codecs again when needed.

start plex

usualy the code folder is tied to GPU transcode and needs flushed with chagne of hardware as teh codec for the hardware that was downloaded is not the same file sand codec for another...

11 hours ago, Tanker said:

I can try the codecs folder first and hopefully that can work, any ideas on how to quickly track down the problem file? I have nearly 70,000 videos in my library.

Carve it into sections, start with files that have been added around the time the issue started and move them out of your library in blocks until the problem goes away then narrow down in that block and so on.

On 6/25/2025 at 6:56 PM, bmartino1 said:

If memetest is passing then its not the cpu memory controler / ram hardware its software...

For some reason... forum downloads apear to be broken for me... would lik to have taken a look at your diag file...

If plex is happen befre the seg faults if could be a memory overflow issue...
it where you mount your config folder..

Plex config = /mnt/user/appdata/plex... / config

/Library/Application Support/Plex Media Server/Codecs

stop the plex docker and delete the contents of this folder. Plex will redownload the codecs again when needed.

start plex

usualy the code folder is tied to GPU transcode and needs flushed with chagne of hardware as teh codec for the hardware that was downloaded is not the same file sand codec for another...

diag files didnt want to download for some reason but I can try again.

Plex app data was being weird and I moved it into the cache but I think the issues with it are just correlation not causation.

On 6/25/2025 at 1:12 PM, Michael_P said:

Nah, especially since they've already run memtest

looks like a broken file crashing the scanner/transcoder, especially if it's the same ffprobe process causing the segfaults. turn off plex for a few days or so and see if it goes away - you can also try deleting the codecs folder in your plex container so it re-builds. if it happens only when plex is running, you'll have to find the file causing the problem

rebooted with plex off and waited, still getting the error. Only other container I can imagine using ffprobe is Tdarr so ill try that next.

Diags finally captured. Hopefully this can help.

mediaserver-diagnostics-20250629-1136.zip

Yeah Tdarr also uses ffmpeg. Mine does it during handbrake encoding if I abruptly stop the encode.

Just to add insult to injury I went into Tdarr and requeued some tasks, was met with the following errors in system logs.

Jul  9 08:48:31 MediaServer kernel: HandBrakeCLI[44062]: segfault at 147f40d52570 ip 0000147f40d52570 sp 0000147fa5bfd518 error 15 likely on CPU 3 (core 1, socket 0)
Jul  9 08:54:42 MediaServer kernel: HandBrakeCLI[149407]: segfault at 1521b02fea30 ip 00001521b02fea30 sp 00001521d85fd518 error 15 likely on CPU 9 (core 4, socket 0)
Jul  9 08:55:28 MediaServer kernel: HandBrakeCLI[156443]: segfault at 14c6003027b0 ip 000014c6003027b0 sp 000014c629dfd518 error 15 likely on CPU 10 (core 5, socket 0)

Makes me more suspicious that the CPU is either failing or somehow isn't seated properly.

29 minutes ago, Tanker said:

Just to add insult to injury I went into Tdarr and requeued some tasks, was met with the following errors in system logs.

Jul  9 08:48:31 MediaServer kernel: HandBrakeCLI[44062]: segfault at 147f40d52570 ip 0000147f40d52570 sp 0000147fa5bfd518 error 15 likely on CPU 3 (core 1, socket 0)
Jul  9 08:54:42 MediaServer kernel: HandBrakeCLI[149407]: segfault at 1521b02fea30 ip 00001521b02fea30 sp 00001521d85fd518 error 15 likely on CPU 9 (core 4, socket 0)
Jul  9 08:55:28 MediaServer kernel: HandBrakeCLI[156443]: segfault at 14c6003027b0 ip 000014c6003027b0 sp 000014c629dfd518 error 15 likely on CPU 10 (core 5, socket 0)

Makes me more suspicious that the CPU is either failing or somehow isn't seated properly.

Probably not the CPU since it's a 15. If you have multiple RAM sticks, try running with 1 and then the other to see if it continues to rule out bad RAM

On 7/9/2025 at 9:32 AM, Michael_P said:

Probably not the CPU since it's a 15. If you have multiple RAM sticks, try running with 1 and then the other to see if it continues to rule out bad RAM

I have 4 sticks total, 2 from the initial build and added 2 not long before the issue started.

removed the new 2 and rebooted, no error until I put plex back on then it popped up again.

swapped the old for the new and same results, so back to it likely being some plex issue.

I did see one other post with similar issues, but waiting to see if they ever found a fix.

After lots of testing and messing with chat GPT, I seem to have identified the issue and fixed it.

I had a script to ffprobe all my media which flagged 8 files that were bad, but the issue persisted even after removing those.

Then had a script to watch the ffprobe segfaults and determine which container was sending reports.

Bazarr was the container sending the segfaults and it looks to have stemmed from an older version of ffmpeg with older codecs.

installed a separate ffmpeg and mapped a path for Bazarr and so far it seems to be working with no errors.

Hopefully this helps someone in the future if they ever get a similar issue.

Edited August 2, 2025Aug 2 by Tanker