Re: Notification of RAIDframe failures

[mlelstv%serpens.de@localhost (Michael van Elst)]

manu%NetBSD.org@localhost (Emmanuel Dreyfus) writes:

>On Sun, Sep 08, 2013 at 02:19:37PM +0200, Edgar Fuß wrote:
>> Currently, /etc/daily sends notification about RAIDframe component failures.
>> I would like to be notified earlier. Is there a more intelligent way doing 
>> this other than performing the /etc/daily-type polling more often? I.e., 
>> is there some kind of notification scheme I can subscribe to?

>I wrote a nagios plugin for that, it is in pkgsrc.


raidframe does no disk scrubbing, it is possible and likely that
a defective disk will not be recognized.

I'm running a small script as a cronjob that reads the disks for a short
time. Error messages can be mailed to root, or something like your
nagios plugin will see the error when running raidctl.

ftp.netbsd.org:/pub/NetBSD/misc/mlelstv/readdisk.sh

Usage is:

readdisk.sh seconds opre num disk1 disk2 ...

seconds = number of seconds to scan each disk
opre    = the file $opre.$disk is used to store the scan offset of a disk
num     = I/O is done in num Megabytes

The crontab entry looks like:

*/5 1-2,4-6 * * * /usr/local/etc/readdisk.sh 2>&1 >/dev/null 20 
/var/log/readdisk.offset 64 wd0d wd1d