Tvheadend

A bit more information. I've seen a lot of people post that this happens to them while the epg grabber is running.
It doesn't look ike that's the case for me. For me, it happens when two things are true. When I'm watching or recording HD channels and using a transcoding streamm.
If I use the pass stream, the problem goes away. If I watch a SD channel, the problems goes away.
That makes me think it's running out of resources, but the vm (and host) appear to have plenty of resources availanle.

The machine can't keep up with transcoding, it's as simple as that. There's no point in increasing the buffer as it would just cause even more backlog and the stream would still be unwatchable.

Thanks for the reply. I agree with you that it must be running out of a resource, but I can't work out which one. There's plenty of CPU, memory and IOPS available.
Is the transcoding multi threaded? If it's only using a core per stream that might explain it.

The transcoding container I'm using is MPEG-TS/av-lib and the codec is webtv-h264.

The reason I was looking at increasing the buffer is, almost all my stuff is recorded and watched later, if it takes a little longer to transcode it, it shouldn't bother me too much.

The vm has 28 cores or CPU, it can't possibly be using all of them.

It's running out of CPU.

Just use a Postprocessing command, then it's recorded normally and transcoded afterwards. Or use hardware acceleration, that'll also help.

Ok thanks for the tip. I'll give post processing a go.

Just out of curiosity though, is it actually possible to increase the buffer?

Of course that's possible, but you need to figure out where the size is set and if there are other important changes you need to make.