Additional information and upstream patch can be found on ubuntu bugzilla: https://bugs.launchpad.net/ubuntu/+source/mdadm/+bug/946758 This is a very serious bug, btw. Monitoring just silently dies at random, so that administrator can notice this only by chance. Reproducible: Always
See also http://git.neil.brown.name/git?p=mdadm.git;a=commit;h=8453f8d0df0111cfd25e984afb7a64153b04bc27 (upstream commit fixing said bug)
Aw, come on, it is at least severity:major! This bug is highly likely to result in loss of data, if administrator relies on mdadm monitoring capabilities for raid health check. I stress that mdadm --monitor just silently segfaults if after array check mismatch_cnt occurs to be high. Without monitoring, it is likely to get a faulty drive and just not notice that! And if mismatch_cnt IS high, faulty drive is all more likely, which means monitoring is completely broken: it works while everything's OK and suddenly stops working when things go awry, without reporting anything.
fixed in 3.2.6 if not earlier
