Disable/spread regular HDD integrity checks

[Nathan Driver]

Hey, the last time my Helios4's PSU blew I suspect was due to a power overload, with the PSU not being able to deal with all 4 HDDs doing active work (this also killed one of the HDDs in my array). Based on the system log, it looked like the system was doing a regular integrity checks on all 4 HDDs at once (!).

 

I would like to either disable the integrity check altogether, or more ideally, spread it out. I have two RAID-1 arrays, and there is no reason for all four drives to be checked simultaneously. For example, I can check one of the arrays (two HDDs) and then once that is done, the second array (the other two HDDs) after.

 

Does anyone know where the code/script for the integrity check lives? The closest I found was /etc/cron.daily/mdadm which runs

mdadm --monitor --scan --oneshot

but from the manpage description it is unclear to me whether that is superficial check or that may actually invoke a deep and lengthy integrity check.

[djurny]

Hi Nathan,

This is something that indeed is done by mdadm configuration. You need to check /etc/cron.d/mdadm:

#
# cron.d/mdadm -- schedules periodic redundancy checks of MD devices
#
# Copyright © martin f. krafft <madduck@madduck.net>
# distributed under the terms of the Artistic Licence 2.0
#

# By default, run at 00:57 on every Sunday, but do nothing unless the day of
# the month is less than or equal to 7. Thus, only run on the first Sunday of
# each month. crontab(5) sucks, unfortunately, in this regard; therefore this
# hack (see #380425).
57 0 * * 0 root if [ -x /usr/share/mdadm/checkarray ] && [ $(date +\%d) -le 7 ]; then /usr/share/mdadm/checkarray --cron --all --idle --quiet; fi

You can either disable this cronjob by commenting out the actual cron entry, or by moving the file 'mdadm' out of the '/etc/cron.d' folder.

 

Easiest would be to write a script that will iterare all mdadm devices, start the redundancy check, wait for the check to complete. then move to the next mdadm device.

 

Something like:

#!/bin/bash

case "$( /usr/bin/id -u )" in
'0')
	;;
*)
	echo "Please run as root user."
	exit 1
	;;
esac

for MD in /dev/md[0-9]*
do
	SYNC_ACTION="/sys/block/${MD:?}/md/sync_action"
	( echo 'check' > "${SYNC_ACTION:?}" ) || exit 1

	while true
	do
		case "$( /usr/bin/cat "${SYNC_ACTION:?}" )" in
		'')
			exit
			;;
		'idle') 
			break
			;;
		esac
		sleep 10
	done
done

# EOF

 

Note that that code snippet was not tested, but should give you direction to your solution.

 

Hope that helps,

Groetjes,