Quote:
Originally Posted by jlcpremier
I am using distributed encoding and recently upgraded to v1.26 on all machines. I've noticed that on just about every job, a slave machine or two hangs when it finishes encoding that chunk. It will stop at either :02, :01 or :00 seconds left. If I manually disable that node by clicking the network button, it will re-encode that chunk on another node, or I can go into the temp folder and usually the chunk is there so then I just have to run the CombineAllChunks.cmd
We are running a full 16 server distributed encoding farm. I've noticed this problem mainly when using the avisynth filters for Mdegrain2 and KNLMeansCL. Each server is an HP EliteDesk 800G1 w/i7 and 8GB RAM on SSD, solely being used with CPU encoding (no GPU)
Is there any settings I can check or change to make this more reliable? You can see an example of how a job looks when it hangs below
|
Next time when you get this problem please also show me what encoding server reports in main TAB and in tcp communication TAB. IT looks like string containing "encoded ..." Was not sent back to encoding client.