hard drive burn-in

[Matthew Gillen]

John Abreau wrote:
>> Matthew Gillen wrote:
>>> I like to use SMART to do tests:
>>> The 'long' SMART test reads every block on the disk,
>>> and can take a very long time.
> ........
>>> I have a cron job that runs a 'long' test every month on my file
>>> server, so I guess it's not really a burn-in thing... :-)
> 
> 
> What sort of impact does the SMART long test have on a production
> server? Does the testing completely tie up the drive, or can the
> system continue to function normally during the test?

All the tests are able to be run during normal system operation (and are 
that way by default; you have to specify a a special option, called 
'captive', that ties up the drive).  I'm not sure what effect it has on 
read/write performance, but you can time it so that it shouldn't matter.

Turns out I was wrong, I don't use cron I use smartd, and I do it twice 
a month.  Here's a portion of my smartd.conf file for one of my drives 
(the -s part says " #on the 1st and 15th of the month, do a 'L'ong 
self-test at 3am):
/dev/sda -H -m root-bi+AKbBUZKY6gyzm1THtWbp2dZbC/Bob at public.gmane.org \
           -d sat \
           -s L/../(1|15)/./03 \
           -l error \
           -l selftest \
           -f \
           -p

Matt