2011/3/20 Alexander Farber <alexander.farber@gmail.com>
Hello,

yesterday night I had a problem with
my server located at a hoster (strato.de).
I couldn't ssh to it and over the remote serial console
I saw "out of memory" errors (sorry, don't have the text).

Then I had reinstall CentOS 5.5/64 bit + all my setup (2h work),
because I have a contract with a social network and
they will shut down my little card game if it is not reponding.

Now the server seems to work ok,
but I'm worried about those /var/log/message:

 kernel: INFO: task md1_resync:9770 blocked for more than 120 seconds.
 kernel: "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables

My guess is that you only saw these messages while the RAID sync was still going on? You got those messages due to the system I/O being stressed, which hung the system in periods.

I wouldn't worry about it if your RAID is now in sync and you don't see the error messages anymore. You can lower the I/O stress of the system under a RAID-resync by setting a lower maximum kb/sek in /proc/sys/dev/raid/speed_limit_max (default is 200.000kb/sec ~ 200mb/sec).  This will of course also extend the time used to complete the sync (which also can be bad, as you want it back in sync as fast as possible).

Best regards
Kenni