User-Agent:       Mozilla/5.0 (X11; Linux x86_64; rv:36.0) Gecko/20100101
Firefox/36.0
Build Identifier: 

The mdcheck script runs a check on all RAID (in my case RAID5) arrays once a
month. This check is limited in time (6 hours in the default setup). Some of my
RAID5 arrays are large enough that the check doesn't finish in 6 hours, so the
check is stopped and resumed the next day. When the check is resumed, the count
of mismatches it had found on the previous day is set to zero again. So if
during day 1 sixteen mismatches were found and during day 2 zero mismatches
were found, the final count will be zero. This creates the false impression
that the RAID5 array is OK.

I did not receive an email reporting the mismatches to me. I am not sure
whether this is because the reporting is delayed until the full check is
completed (in which case it is crucial to get the total sum of mismatches over
all days right) or because the script doesn't report any mismatches it finds
(which would not make any sense to me, such a feature would be important to
have).

Maybe this doesn't belong here, but since we are on the subject... I have not
been able to find out how you can identify which physical disk in a RAID5 array
is responsible for the mismatches. Is it possible to get that information?

Reproducible: Always

Steps to Reproduce:
The mdcheck script is started automatically by cron.
Actual Results:  
No mail was sent, the mismatch count was 0 after the final installment of the
check even though the RAID5 array contained errors.

Expected Results:  
I would expect at least to get an email reporting the mismatches in the RAID5
array. Ideally the mismatch count should also reflect the total number of
mismatches on the array, not just the number from the last chunk of the check.