Interrupts from ethX sticking to one core

From: Roy Sigurd Karlsbakk
Date: Thu Feb 20 2014 - 06:07:19 EST


Hi all

I'm running a NAT44 gateway on Debian Wheezy, kernel 3.2, and timers from eth[01] sticks to cpu0 even though I've set cpu_affinity to 0f (4 cores). This is a VM on ESXi. Setting affinity to a specific core works, but the kernel doesn't balance interrupts dynamically as I would have hoped it should, even when CPU load is at 95% (mostly soft interrupts). I also tried to install irqbalance without this having any effect. Drive used is e1000 (for the moment, will move to vmxnet3, but afaics from testing on another host, this didn't solve the problem).

Is this a known issue? Any way I can fix this?

Please CC: to me as I'm not on the list

Vennlige hilsener / Best regards

roy
--
Roy Sigurd Karlsbakk
(+47) 98013356
roy@xxxxxxxxxxxxx
http://blogg.karlsbakk.net/
GPG Public key: http://karlsbakk.net/roysigurdkarlsbakk.pubkey.txt
--
I all pedagogikk er det essensielt at pensum presenteres intelligibelt. Det er et elementÃrt imperativ for alle pedagoger à unngà eksessiv anvendelse av idiomer med xenotyp etymologi. I de fleste tilfeller eksisterer adekvate og relevante synonymer pà norsk.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/