PMS Version 1.30.1.6562 - this update stopped my transcodes from working

Plex Logs.zip (1.0 MB)
Server Version#:Version 1.30.1.6562
Player Version#:web

Edit 4: I’ve been away from for while. Attempting some of a later versions from time to time, but having to rollback to previous versions. But now it seems to be working with the latest beta: 1.31.3.6819-2ef591a4c_amd64.deb… I’ll add a new post at the bottom of this thread :wink:

Edit 3: My transcode problems started with 1.30.1.6562, remote Roku boxes would fail. I have no access to them, so I relied on using Chrome Browser to test transcodes (via changing bandwidth/mbps settings to for a transcode)… With me rolling back my PMS version, the Roku users were once again happy. BUT in doing the Chrome testing I’ve now found a 2nd, non-related problem: Switching between different bandwidths causes PMS to screw up (even on 1.29.): When I first play a file, it will work (either DirectPlay, or if it’s remember to play it at a lower mbps), then when I switch bandwidths it will fail , if I switch again it’ll work, and so on and so on… Looking at the PMS logs, for hours, PMS is closing/killing the transcode process itself (it’s in my log dissection Logs ) even though the client is still trying to use it. To the point where there are GET commands met with 404 and the transcode session folder has been deleted.

Edit 2: I downgraded to PMS 1.29.2, made sure I was running Nvidia drivers v515, and everything is back to how it was. Transcoding is working just as good as it was. Though I didn’t get any helpful feedback from within my own post, I had to read others and then read between the lines to get things working.

Edit 1: I previous had my PMS version one, as I’m not running PMS BETA. So I’ve updated to the correct version no. now.

I updated my Ubuntu PMS, from the previous update, to PMS Version 1.30.1.6562. After that I found that transcoding had stopped working all together. I also saw plugin errors for the EPG?

TL;DR – I ended up having to restore an entire backup of the server to get it up and running agian!

Longer version : I had read the release notes on Nvidia driver versions, and knew that I was on v515, so wasn’t worried. But once the update completed, transcoding just would not start. I had multiple remote Roku boxes just drop the stream and not recover.

I went about upgrading the Ubuntu drivers for Nvidia, rebooted and still the same problem. I then went about trying to downgrade my version of PMS. This process wasn’t too hard; I stopped the service, downloaded the previous .deb file and installed it. But I still had problems where transcoding would not start, but the PMS console just stopped working at this point.

After that I just went about restoring a copy of my Ubuntu VM. Far easier than trying to figure out further what happened. Everything is now working normally agian.

I’m just posting this up in-case it helps someone else, or at least signposts someone saying “erm, I think there is a problem with this update”.

You don’t have debug logging enabled so it’s hard to tell what’s going on. I do see a bunch of errors related to hardware decoding, but again, can’t tell what’s wrong without the full debug info.

1 Like

Well that’s a shame… I’ll see if I can upgrade again and create more havoc for myself.

@anon18523487 Here’s the logs. I re-updated PMS and I’m back to having issues, debug logs should now be enabled.

Current Nvidia driver : NVIDIA-SMI 515.86.01 Driver Version: 515.86.01 CUDA Version: 11.7

@Minxster

I turn the transcoder debug on in these cases.

It might help you here. (This shell script assumes native Linux host install)

[chuck@lizum ~.2000]$ cat bin/set-transcoder-debug 
#!/bin/sh
Token="$(cat "/var/lib/plexmediaserver/Library/Application Support/Plex Media Server/Preferences.xml" | tail -1 | sed -e 's/.*PlexOnlineToken="//' | sed -e 's/".*//')"
curl -X PUT http://127.0.0.1:32400/:/prefs/set\?TranscoderLogLevel=debug\&X-Plex-Token=$Token
echo Token=$Token
[chuck@lizum ~.2001]$ 

@ChuckPa I’m on Ubuntu, but am no expert, but think I understand what the script is doing. Where will these transcode-debug logs be placed? In the ‘Plex Media Server.log’? Also, when I’m finished, what do I set “TranscoderLogLevel” to (what is it’s default?)?

FYI, I have no issue transcoding normally, haven’t done for years. All I did in my recent test is load up a know working film in Chrome/Web, then force the Mbps to drop to 10, thus causing a transcode. Again, normally none of this is an issue.

@ChuckPa Nice dice on the script, got it to run but got this back (see image), the token seemed populated though. I attempted to manually string together the info in a URL call from a browser, but I just got 404 error.

I find info about making preference changes (https://support.plex.tv/articles/201105343-advanced-hidden-server-settings/, so I carefully added this transcode debug entry into it and restarted the plex service.

I ran another quick test, same as before. But for clarity, I start a film, which is fine, then force it to 10Mbps which then should force a transcode. All I see in my browser is a blank screen, no caching or progression update/info. It just sits there.

What are you doing here???

Your logs:

Jan 11, 2023 22:52:18.368 [0x7f10ac9bdb38] INFO - Plex Media Server v1.30.1.6562-915986d62 - Ubuntu PC x86_64 - build: linux-x86_64 debian - GMT 00:00
Jan 11, 2023 22:52:18.368 [0x7f10ac9bdb38] INFO - Linux version: 20.04.5 LTS (Focal Fossa), language: en-US
Jan 11, 2023 22:52:18.368 [0x7f10ac9bdb38] INFO - Processor: 6-core Intel(R) Xeon(R) CPU E5-2690 v4 @ 2.60GHz
Jan 11, 2023 22:52:18.368 [0x7f10ac9bdb38] INFO - Compiler is - Clang 11.0.1 (https://plex.tv 9b997da8e5b47bdb4a9425b3a3b290be393b4b1f)
Jan 11, 2023 22:52:18.368 [0x7f10ac9bdb38] INFO - /usr/lib/plexmediaserver/Plex Media Server

My logs – Same CPU

Jan 11, 2023 15:29:20.067 [0x7ff57b832b38] INFO - Plex Media Server v1.30.1.6562-915986d62 - Ubuntu PC x86_64 - build: linux-x86_64 debian - GMT -05:00
Jan 11, 2023 15:29:20.067 [0x7ff57b832b38] INFO - Linux version: 20.04.5 LTS (Focal Fossa), language: en-US
Jan 11, 2023 15:29:20.067 [0x7ff57b832b38] INFO - Processor: 28-core Intel(R) Xeon(R) CPU E5-2690 v4 @ 2.60GHz
Jan 11, 2023 15:29:20.067 [0x7ff57b832b38] INFO - Compiler is - Clang 11.0.1 (https://plex.tv 9b997da8e5b47bdb4a9425b3a3b290be393b4b1f)
Jan 11, 2023 15:29:20.067 [0x7ff57b832b38] INFO - /usr/lib/plexmediaserver/Plex Media Server

Clearly your environment is differerent. How is it different?

Got it. I am seeing an error. Looking into it.

1 Like

I take that back. The errors I see are correct. You have a Xeon processor without QuickSync. The errors I see are PMS trying to use QuickSync which will fail so the error is correct. The transcoder ends up using NVDEC and NVENC for your P2200, so things look ok on the server side.

I see that you are using the TrueHD track. Can you try a different audio track and see if you get the same failure. There was an issue with TrueHD that we thought was fixed in this version, but maybe we are still missing something.

It’s only different in that I’m using VMWare, so when I restored the VM I just pinned more resources to the server, but it was also a fresh VM (some time/GMT needed to be changed at some point). All because of the restore function I’m doing, it’s a lot of work to switch it back to normal… Other than that it’s 100% the same PMS VM.

Just in regards to QuickSync, is there a way of telling Plex not to use it? Doing install/updates it knows there is no QuickSync, but it still tries to find/use it when a transcode starts.

I’ll re-update PMS, as I restore it from a backup once again late lastnight, and see if the audio change makes a difference. But I know this still won’t work as a normal 265 encoded TV show would not transcode to a Roku, which I’m not able to test and show as it’s at a remote site.

Screenshot of PMS upgrade:
Untitled

In the 2nd ZIP I uploaded, at line 11767 there are a bunch of “404 GET” messages. I’m currently running some tests to try and capture more data/info, but I’m seeing these types of message come up when the web play seems to be stuck (during transcode)… Is this helping you to spot any problems?

@anon18523487 I have no idea what’s going on… I must have 2 errors… On the older version of PMS I get 404 GET errors, that I mentioned above. Which is odd as transcoding is set to /dev/shm… But it seems ok to pull files before I ask it to transcode.

Anywho, after upgrading to latest PMS again, I saw a load of CUDA errors in the Console. So while I batter the 404 GET… I though I’d upload the CUDA errors I’m now seeing with latest PMS?

Just to add, my 404 GET errors that I’m getting on the slightly older version of PMS (playing/transcoding problems) go away if I turn on hardware transcoding and just use the CPU.

But again, as soon as I update I can see the CUDA errors in the last set of logs I posted.

(See my other posts before reading this)… So after more messing around and looking at different posts, my 404 GET error can be stopped if I browse my local PMS via https://app.plex.tv/desktop/. So I can deal with that, though I do not understand why it’s being a PITA… Is there something I can do to stop this from happening?

As far as upgraded PMS to the latest, I’ll have to try this again tonight after all the users are offline.

Well I thought I had solve some form of a problem with using app.plex.tv… But nope, this has not worked. I still get the 404 GET error message.

I think I need to fix this problem before I move onto upgrading PMS… Any ideas for the logs I’ve uploaded or things I can try?

Those 404 errors are correct. The transcoder is failing so there is nothing to return.

Jan 12, 2023 00:08:56.766 [0x7f2e1be9cb38] DEBUG - Jobs: ‘/usr/lib/plexmediaserver/Plex Transcoder’ exit code for process 13445 is -9 (signal: Killed)

I’m not familiar with using VMware so I can’t help there, but you may have a configuration issue with your GPU. PMS does not natively support HW transcoding in a VM, that will be up to you to get the configuration correct so PMS can utilize the GPU. If turning off the option to use HW transcoding allows the system to work, that almost definitely points at the GPU being the issue.

If you’re going to use a GPU inside a VM through VMware or XenServer etc you’ll need to pass the card to the VM. This will render it unavailable to any other VM while it’s assigned to a specific VM.

First your CPU needs to support VT-d/IOMMU. If it doesn’t you’re out of luck in that regard and will not be able to pass it through.

Otherwise you need to research and set up vGPU instead.

First step in all of this is to make sure your GPU is passed to the VM correctly.

Does it show up via lspci command? If yes then we need to make sure the user PMS runs under has the correct permissions to access the GPU.

As for telling Plex not to use QuickSync you could possibly give the VM a different CPU profile that emulates it to be a different model that doesn’t support QuickSync. Then when PMS goes to check for that instruction set in the CPU it won’t be there.

1 Like

That won’t help. PMS will still check for QuickSync everytime. There isn’t a way to stop this.

1 Like