Transcoding errors starting 2-3 days ago

Server Version#: 1.19.1.2645
OS: Windows 10 Pro 1909 X64
CPU: AMD Ryzen 3600
GPU: Nvidia Quadro P2000 (Driver 26.21.14.4112 from 10/24/2019 (newest from Windows Update))
SSD: Samsung EVO 970 Plus 250GB (stores OS, Plex Install, and Plex database).
Video Storage: All video is stored on secondary servers connected by SMB shares from 10G connected Windows 2012R2 file servers.
Player Version#: See below

Note: I don’t use any special antiviruses or similar on this system, just whatever comes with Windows 10.

Everything has been working perfect for months, but starting about 2-3 days ago, I started getting complaints from family members about errors on their clients while watching streams that are being transcoded. The errors being “Conversion failed. The transcoder exited due to an error.” and “Playback has stopped due to multiple playback errors”. My family members use a plethora of players including Rokus, Fire TV sticks, Vizio TVs, etc. I recently saw that Plex Media Server updated from 1.18.X to 1.19.X, but I’m not sure if that is part of the problem or just a coincidence. I leave my server on auto-update on the Public channel. I did move my temporary transcode location to C:\Plex as I noticed some people try in the forums, but that didn’t seem to help (it has full permissions to everyone as well). Everything else is in its default locations.

As recommended, I turned on debug logging and replicated the issue locally using my Roku Ultra (4660RW) with firmware 9.2.0 build 4807-46 and Plex Version 6.5 build 1. The issue happened just minutes before I pulled the logs, but I didn’t grab the exact time. The error this time was “Playback has stopped due to multiple playback errors.”

I’ve done the simple “reboot everything” and the issue still stands. Help Please!

Plex Media Server Logs_2020-04-10_11-18-44.zip (2.7 MB)

Thinking that it couldn’t be a coincidence that updating to 1.19.X happened at about the same time as all of my issues starting, I downgraded back to 1.18.9.2578 (was on 1.19.1.2645) before last night’s rush and I did not hear any complaints from my family at all. I will update again in a few days if the downgrade did actually resolve my issue.

I have the same issue after upgrading the server to 1.19.1.2645.
Downgraded to 1.18.9.2578 and everything runs fine.

I’m getting the same on my end for clients

1 Like

It is the same issue as I am investigating with diagnostics from a few other users

We are failing to get a dynamic tcp port
The transcoder failed at this time

Apr 10, 2020 10:57:09.075 [8920] ERROR - [Transcoder] [tcp @ 14713600] Connection to tcp://127.0.0.1:32400 failed: Error number -138 occurred

But the scanner got an error earlier

Apr 10, 2020 10:56:56.547 [1584] WARN - HTTP error requesting GET http://127.0.0.1:32400/:/metadata/updateProgressMessage?message=Scanning%20The%20Dark%20Knight%20%282008%29 (7, Couldn't connect to server) (Failed to connect to 127.0.0.1 port 32400: Address already in use)

I have not established yet if there was a change in windows recently to do with this - but I am aware that windows did have bugs in this area and this was already acknowledged in some windows updates - but not recent

For one user the problem mysteriously went away - Recording failures after update to 1.19.1 beta - #16 by sa2000

For another user the problem did not go after windows update to a later version 1909 OS build Transcoder crashes on access errors - #16 by technaholic

So we need to establish if there has been a recent windows change that impacted this or it is only down to Plex Media Server changes

The scanner logs @Subrosian in your zip only go back to Apr 10, 2020 10:56:24
Would be good to capture the first scanner error and to see after how many request (from reboot) does it arise

It may be necessary to increase number of logs - to say 20 - See registry settings for Plex Media Server and how to increase logs using LogNumFiles
https://support.plex.tv/articles/201105343-advanced-hidden-server-settings/

First need to work out how to reproduce the error -
With 2600 movies - how many scans of the Movies library after a reboot does it need to give the error. There is a 5 minute wait time in windows before a dynamic port can be re-allocated. - With logs covering the first fail and covering whole period since reboot - I can establish how many requests in a given time period would lead to the error

So with a method to reproduce the error after a reboot, I would like to see if downgrading to an older version of Plex Media Server does not give the error.

1.18.9.2578 was mentioned by others and that would be one to try and if it definitely does not fail - then would like to see logs + wireshark pcap file

I have this same error from every user that tries to watch from my server. It happens EVERY time, usually after about 10 or 15 minutes of watching. It happens on TV shows & movies. It’s been happening for months and across at least two different version upgrades. It’s so bad that nobody even watches. Unfortunately I’m not good with this. Is this so uncommon that you have no idea why it’s happening? Unfortunately I don’t even know how to get my logs for this. I downloaded the log files & there’s so many I’ve no idea what you need. I can’t even open some of them because they aren’t text files. I can’t imagine that there are only a few people with this problem.

Hardware transcoding hasn’t worked reliably for me since June 2019… Tried two different cards (Quadro RTX 4000 and Quadro P5000)

So it would be nice to get to the bottom of this. My observation seems to point to the decoding side of the transcoder.

Do you mean you see this error in the Plex Media Server logs ?

Connection to tcp://127.0.0.1:32400 failed: Error number -138 occurred

The problem actually arises some time before this erorr starts

Best way to know when the issue arise is to open the Windows Event Log with event viewer

Start / Run / eventvwr.exe
Expand Windows Logs
Click on System
Click on Filter Current Log
Overwrite filter <All Event IDs> with 4227
That will show when the problem arises
You can change the filter to 4227,6005
and that will show reboots and the tcp error

I am coming to think that this was a recent windows change - but what I need to establish is how many library scans it takes to get this and how big are the libraries - (a reboot would clear the issue)

You can save copies of the 3 event logs and zip and send me and I will see if 4227’s are occurring

Application
System
Update

Server logs after a 4227 is logged by windows would show what was being done at the time

I started getting transcoding error often with the newer version, I downgraded to 1.18.9.2578, things are working fine again. This time I am staying with this version as long as possible. The most recent versions seem unstable. I setup three test servers with the newest version, clean install, windows fully updated, standard configuration, all generated transcoding errors. Tested with version 1.18.9.2578, no errors.

Did you save any server logs ? Would need that to start an investigation

Please help us fix this error. I have people using my plex server that get this conversion failed error while playing files during my library scanning.

I don’t think it happens unless my library is scanning. I don’t like having to turn off my library scanning. I like the fact that it could always detect recent media and add it to plex almost immediately.

Have you tested while scanning all your libraries and playing video at the same time?

I can only comment on issues for which I have seen logs for

If you have large libraries and the scanning goes very fast then it could be the port exhaustion issue but I am not going to assume it is

Sorry, I have tried uploading logs all over.

Thanks for responding. I can happily upload my logs.

My library is massive and the conversion fails happen within a minute or so of scanning.
these are from my most recent rollback to version 1.18.8.2527.
Plex Media Server Logs_2020-05-06_06-15-29.zip (6.6 MB)

these below here are from the very latest plex as of this past week.

Plex Media Server Logs_2020-05-04_06-03-40.zip (6.4 MB)

i have windows 10. someone else told me it could be related to part of how curl (a very basic part of the scanner) that uses ports. It uses about 4-6 ports per movie which is why it eats up all of the ports causing the transcoder issue.

I would appreciate if you could look at my logs and maybe help me.

I don’t mean to be a bother, but I really would like to get this fix. I have been a plex pass member for some time and this is the only time I have ever reached out for support.

i am on windows 10, all recent updates.

Sorry, I didn’t ignore you. I had provided the info in the other thread to give a bit more information to find correlations with that user’s problems and my own. Since the Covid-19 issue started, I’ve been putting in 80+ hour weeks in my job and haven’t had the time or energy to even touch the server. If the issue isn’t resolved by the time my life goes back to normal, I will re-engage you to continue figuring out the issue. Right now, turning off scheduled library scans and letting the server only update when it finds changes and staying on 1.18.9.2578 has lessened the issue enough that users aren’t complaining anymore. Anyways, thank you for being so active on the forums. :slight_smile:

Thanks - For every case of 4227 Windows Event Log entry seen during scanning, I would like to collate info on the sizes of the libraries - how many movies and if the issue arises after scanning a movie library.

Once a 4227 event shows up where tcp ports are exhausted - the safest option to recover is to reboot windows.

Thank you - very good diagnostics

It looks like a change in the version of curl used within Plex Media Server that is probably the cause. We are now using up all ports after about 4100 movies

I have added your diagnostics to the issue that is open. We are waiting a response to an existing curl issue.

In the meantime - you could consider breaking up your movies library into multiple libraries

Ports appear to start getting reused after 3.5 minutes +

1 Like

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.