If your external USB drives as raid starts, you'll know that it can sometimes take days before a rebuild or check the raid is complete: md4: active raid5 sdc sdf [0] [ 3] sde [2] sdd [1] 2930287488 blocks level 5, 64k chunk, algorithm 2 [4 / 4] [UUUU] [>.................... ] check = 0.9% (8861912 / 976762496) finish = 1374.5min speed = 11735K/sec If ye then wonder why the response on a Sunday night raid on a Debian system is very slow and 50% of the CPU by a thread with named md4_raid5 occupied , I recommend you the time of the cronjob mdadm to see. This is the first Sunday of each month and reviewed the raid for consistency. On the one hand, quite useful but it can also be annoying - but as I said, it takes relatively long time. There are also in Debian at some installations, a bug, so that the cron job runs every Sunday. / usr / share / mdadm / check array-x - all
The cronjob is defined in the / etc / cron.d / mdadm .
disablecomplete is also ugly, because it prevents the detection of errors on the raid. In Mirror Raids is fatal, the failure of a disk in a Raid 5 as well, because then there are also no reduntenten information to restore.
0 comments:
Post a Comment