addendum: after backing up data for 30 hours off the degraded array, replacing sda and after another 50 hours resync the raid5 was healthy again. for about 8 hours when again it was syncing. and then I discovered at least a small part of the mystery: centos 5 runs a script /etc/cron.weekly/99-raid-check (once a week of course) which (since a few months) triggers a re-sync of the array (which then runs for another 50 hours at a system load of around 8.0). I always noticed that it resynced without a drive marked "faulty" in /proc/mdstat, but I never really took it seriously. And being away so often I never realized that it always started at sunday 4:20 in the night. rebooting during the resync stops it and the system behaves normally again. If I interpret the text in /etc/sysconfig/raid-check correctly, I'd better leave 99-raid-check running (for 50 hours ... :-( ), check the value of /sys/block/md0/md/mismatch_cnt and if it contains other than 0, I'd better get worried? and, catching up with a previous suggestion in this thread: is it safe to run a smrtctl long selftest on each disk while the raid is mounted & active? long selftest is supposed to take about 4 hours (per disk). tnx & cu -- To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
