Hit in /proc/interrupts

[Equipe_reseau]

Hello team,
I have one question about SND:

Our SND on core 0 and has never autobalanced, all in core 0

[Expert@XXXXXX:0]# sim affinity -l
eth0 : 0
eth3 : 0
eth1 : 0
eth2 : 0

But why the number of hits on other CPUs are appeared (CPU1,CPU2) ?

[Expert@XXXXXX:0]#cat /proc/interrupts
CPU0 CPU1 CPU2 CPU3
67: 412471638 613 0 0 IO-APIC-level eth0, eth3
75: 909964861 0 12 0 IO-APIC-level eth1
83: 954314010 0 115 0 IO-APIC-level eth2

Thanks

[_Val_]

Version, amount of CPUs?

[Equipe_reseau]

R80.30, 4 CPU, CPU 0 for SND and others for firewall workers

[_Val_]

The data from proc/interrupts is weird. Reboot, check again. All cards should be served by CPU0. 

[Equipe_reseau]

Hello _Val_,

Thanks for replying to me. We are using 4CPUs.

[Expert@XXXX:0]# fw ctl affinity -l -r
CPU 0: eth0 eth1 eth2 eth3
CPU 1: fw_2
mpdaemon lpd vpnd fwd in.acapd in.asessiond cpd cprid
CPU 2: fw_1
mpdaemon lpd vpnd fwd in.acapd in.asessiond cpd cprid
CPU 3: fw_0
mpdaemon lpd vpnd fwd in.acapd in.asessiond cpd cprid
All:

BTW, thanks for your suggestion again. I'll try to reboot and check it again.

Thanks

[Timothy_Hall]

It is possible those wayward interrupts on CPU1 and CPU2 happened very early in the boot process, before all the interfaces could be affined to CPU0.  As Val said try rebooting, if my theory is correct there will still be some wayward interrupts on CPUs other than 0.  As long as the overwhelming majority of interrupts are on CPU0 in your split config I don't think it is a problem.