I’m running Netdata 1.35 and I can see that the worker thread for streamrcv are very busy. Often at 100%.
I’m running in a 100+ cluster with many pods and there are often gaps in data and I think the parent is having difficulty managing all the data.
Is there any way to increase the thread count or lower the amount of work they need to do?
Hi @data-smith !
A couple of questions, are you using SSL and/or compression in the stream settings?
I’m using the default for 1.35. I haven’t changed any of the configs for ssl or compression. I scaled it to 14 nodes and i’m still having issues. There’s never any free worker nodes for streamrc.
I allocated 13 cpus and the parent is using about 1/10 of 1 cpu. The worker threads are almost always 100% busy for some reason. I’d love to just configure something to allow it to spawn more threads.
Setting this seems to have solved the issue
web server threads = 12