Hi,
On 14-04-2015 13:04, Johan Kooijman wrote:
Hi,
Today I migrated a server form C6 to C7. The machine is dedicated for a couple of redis daemons. given the traffic etc, we see cpu0 being 100% in use immediately. No problem, had the same thing on the C6 server.
So I pinned certain processes/irq's to other CPU's and made irqbalance was monitoring things. That did solve the issue in C6, but not this time.
The links below contain the output of /proc/interrupts. As you can see CPU0 doesn't seem that busy http://pastebin.centos.org/23361/ http://pastebin.centos.org/23366/ The 2nd paste was about 60 seconds after the first one.
Am I missing something here? This is a brand new C7 installation:
[root@redis1 ~]# uname -a Linux redis1.gamebasics.lan 3.10.0-229.1.2.el7.x86_64 #1 SMP Fri Mar 27 03:04:26 UTC 2015 x86_64 x86_64 x86_64 GNU/Linux
Hardware is a brand new supermicro X9DRW board with onboard igb NIC's in use.
Hope someone can help me!
Is it really just because of networking? The counters didn't seem very high. You may watch /proc/softirqs for other sirqs too.
Marcelo