Hi,
after upgrading to 3.9.2 (from 3.2.1) we have noticed million sized
spikes in Function call interrupts (CAL) graphics in Munin monitoring.
After some investigation we have found that it is caused by completely
borked numbers in /proc/interrupts
for example current output is:
CPU0 CPU1 CPU2 CPU3 CPU4 CPU5 CPU6
CPU7 CPU8 CPU9 CPU10 CPU11 CPU12
CPU13 CPU14 CPU15 C6 CPU17 CPU18 CPU19
CPU20 CPU21 CPU22 CPU23
CAL: 4294967272 5513065 56 5567608 21 5513363
50 5583946 42 5431365 32 5681459
100 2613904 95 2494755 95 2554234
122 2434635 74 2472960 63 2504140 Function
call interrupts
As you can see CPU1 output has really crazy value (FFFFFFE8), but we
have seen some other values on other cpus as well. It changes suddenly
from valid values, to nearly FFFFFFFF and then might return back after
some time.
There are no spikes in cpu usage, no interrupts, context switches,
just /proc/interrupts is wrong. it went stright from ~50 to FFFFFFE8
in one step.
System is dual Xeon X5650 NUMA box. ( kinda only interesting options
are THP with defrag off and that new Numa memory placement aware
scheduler is off ).
Anyone got any clue why it is happening?
Best regards,
Darius.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
