The problem
At least on Debian, md arrays are routinely checked monthly by /etc/cron.d/mdadm. During the check, the nagios plugin check_md_raid generates critical alerts.
The solution
In /usr/lib/nagios/plugins/check_md_raid, change
Code:
re_clean = re.compile('^clean(, no-errors)? $')
to
Code:
re_clean = re.compile('^clean(, (checking|no-errors))? $')