Hi,
Hello. Did not really know where I would have to go with my problem,
but I take a chance that anyone here maybe can help me with my
problem.
As the topic reveals, I get a lot of "hardware interrupts" on my
Debian based server. The problem arises as soon as any of my HLDS
processes use core 0 or 11.
Now I have assigned 10 HLDS installations to core 1-10. But as soon
as I assign HLDS number 11 to core 0 or 11 the Hardware interrupts
occurs.
This problem occurs even if I dont assign HLDS number 11 to any core,
and run it on auto and as soon as core 0 and 11 grabs process, I get
"hardware interrupts".
My guess, and it is just a guess, is that cores 0 and 11 are being used
by default to poll the network cards for packets (see Linux NAPI) rather
than waiting for interrupts.
When you put a HLDS process on that core you stop the polling from being
fast enough and so interrupts are needed.
Switching back to interrupts is less efficient, but not fatal.
A
_______________________________________________
To unsubscribe, edit your list preferences, or view the list archives, please
visit:
http://list.valvesoftware.com/mailman/listinfo/hlds_linux