Why am I switching to transcode in the middle of playback?

[puithove 208]

Server 4.7.5 (was on 4.7.2 when this started)

Android TV client 2.0.77g (Shield 2019 Pro)

Something strange started just in the past week or two.  I noticed that playback would just hang without giving the buffering spinner... then after a minute or so the buffering spinner would start, but playback just wouldn't resume (or I never waited long enough for it to).  So digging into the embyserver.txt log, I see now that it appears to be wanting to switch from DirectPlay to transcode.  It ends up erroring out due to not being able to start ffmpeg for some reason and it throws a 500.  Ignoring why it can't start ffmpeg, why is it trying to switch to transoce in the first place when DirectPlay has been chugging along fine?

 

In the attached log file, I had two Shields playing two different movie files, and they both did this right about the same time.

Client one:  2022-07-04 09:39:17.257

Client two: 2022-07-04 09:39:20.058

 

When I first spotted this, I changed the client-side max bitrate setting from Auto (thinking the client may have thought there was a network slowdown) to 1Gbps.  In the log at the times noted, they were both set to 1Gbps.  I'm not sure why else it would be wanting to switch mid-stream.

embyserver.txt

[ebr 14910]

Can you please play that item until the issue occurs and then follow the instructions to send a log from the app?  Send the log right from the menu on the OSD right after it starts trying to transcode.  Thanks.

 

[puithove 208]

Will do.

[puithove 208]

Crap - it just happened and I sent the logs from the app - playback hung, but I may have stopped it too quick.  I don't see the start of any transcoding in my server log (attached), just me trying to exit to get back to the home screen to send logs.  Happened about 8:15pm Eastern time, user "tv", watching Harry Potter and the Prisoner of Azkaban.

When it happens again, I'll wait a bit and confirm that I see the server log showing the transcoding, then send the logs.

embyserver.txt

[puithove 208]

I'm starting to think something is happening server-side that stops it from feeding the stream causing the client to request transcoding.  I'm going to have to dig in a bit deeper the next time it happens to see what is hung up on the server - either in Emby or at the OS level.

[puithove 208]

Ok, new logs sent with it happening again.  It does look like my nfs share is hanging, so Emby Server isn't able to continue accessing the file.   gotta figure that out.

embyserver.txt

[ebr 14910]

13 hours ago, puithove said:

It does look like my nfs share is hanging,

Yep, something is making the source unavailable:

	System.IO.FileNotFoundException: System.IO.FileNotFoundException: Unable to find the specified file.

Please keep us up with what you find.

Thanks.

[puithove 208]

Thanks for helping confirm that. I was able to figure out a way to consistently trigger it (heavy file reads with dd).  Other NFS clients were able to continue accessing the NFS server at the time of the hang.  From there I figured out that there was something in the networking of the guest VM where I'm running Emby server.  I haven't changed any configs on the guest VM, the VM host, or the NFS server, so I'm thinking that a recent update brought along something ugly.  I think I've cured it now by turning off multiqueue on the VM's network adapter - or at least I can no longer trigger it as I had been able to before.

Edited July 5, 2022 by puithove

[ebr 14910]

Good to hear.  Thanks for the update.