Benchmark node performance
Opened this issue · 2 comments
Feature request
This will allow for better use of the nodes due as we'd know how well each system performs in doing the transcode tasks. If you had all nodes all with the same hardware then this wouldn't matter but if you have quite different hardware configurations then it becomes much more important. My systems are a good example.
Node 1 - I3-2120 (dual core 3.3Ghz)
Node 2 - i5-4250U (dual core 1.30Ghz)
Node 3 - FX-8350 (6 core 4.00Ghz)
After quite a bit of thinking (and not set in stone, I think the best solution is to use the built in benchmarking process in ffmpeg, which plex uses so will always be there).
I was about to put an example but for some reason my tests today were coming back with very similar results for all 3 systems which it didn't the other day.
But that's the thinking I had, obviously some systems may be more powerful than others to having some way of the system knowing that even if it's approx can help with it's load balancing decisions.
You can assign it to me if you like or collab
What does the command look like to benchmark?
cd /opt/plex/tmp/benchmark; rm -f *.ts ; prt_local -benchmark -ss 0 -i
That's a quick dirty script I did.
Takes the 1st 100 seconds of the file and converts to mpeg ts which is what plex uses for streaming, it outputs the time it took to do that. The lower the time, the better the performance.