[DRBD-user] drbd /proc/diskstats counter issue

[Justin Albstmeijer]

Hi,

I am monitoring the "# of I/Os currently in progress"
from /proc/diskstats for our drbd devices.
After upgrading from drbd-8.4.1 to drbd-8.4.3 this counter does not seem
to behave as advertised.

Field  9 -- # of I/Os currently in progress
    The only field that should go to zero. Incremented as requests are
    given to appropriate struct request_queue and decremented as they finish.

It starts at 0 but quickly grows to a value of +4294000000.
It then seems to stabilize on such high number and fluctuates lightly.

The expected behavior is to settle at 0 and lightly fluctuate.
The local underlying disks and dm-* devices show normal counter behavior.
It only seems to impact drbd device counters for which the server it self is primary. 

# while true; do cat /proc/diskstats |grep drbd21 | awk {'print $12'}; sleep 10; done
4294964312
4294964312
4294964312
4294964312
4294964312
4294964312
4294964312
4294964312
4294964312
4294964312
4294964313
4294964312
4294964312
4294964312
4294964312
4294964312
4294964312
4294964312
4294964935
4294964312

Any idea what is happening here?

Currently running:

Centos-6
kernel-2.6.32-358.14.1.el6.x86_64
drbd-kmdl-2.6.32-358.14.1.el6-8.4.3-33.el6.x86_64

Thank you,

Justin