[PLUG] Reliable storage on a shoestring

Eli Stair eli.stair at gmail.com
Sun Aug 7 15:34:34 UTC 2005


If you use a very current OS (Fedora4 or RHEL/CentOS4), the md
(Software RAID) drivers are configured by default to alert you (via
syslog/logwatch at least, possibly mail to root I forget).  As Aaron
mentioned, use smartd which also comes standard with all modern
distros.

Just as a note, I have seen 'silent' corruption in a couple of cases,
where you don't know your files are toast until things stop working
right, start segfaulting and you MD5 them.  I've seen this during some
filesystem testing and tripped an XFS bug, and also some straight up
3Ware issues.

/eli



On 8/6/05, Aaron Burt <aaron at bavariati.org> wrote:
> On Sat, Aug 06, 2005 at 09:11:58PM -0700, Evan Heidtmann wrote:
> > This leads to a more specific question: in what way are my drives likely
> > to fail? Will they throw up a red flag saying "I'm dead, replace me", or
> > will they be less straightforward about it and start flipping bits in my
> > data?
> 
> Drives almost always fail two ways: slow decay (more and more unreadable
> sectors) or catastrophic (won't do nuthin'.)  Either way, silent data
> corruption doesn't happen.
> 
> With RAID, unreadable sectors aren't a big problem; the system bitches
> about it in the syslog and gets the data from a good drive.  Same deal
> with a dead drive.
> 
> Modern IDE drives have a thingy called SMART for monitoring drive
> health.  You can install SMARTd, which will keep an eye on it.
> 
> 
> Aaron, who doesn't miss MFM, RLL or ESDI *at* *all*.
> _______________________________________________
> PLUG mailing list
> PLUG at lists.pdxlinux.org
> http://lists.pdxlinux.org/mailman/listinfo/plug
>



More information about the PLUG mailing list