[plug] harddrive becomes read-only

Gavin Chester gavin.chester at gmail.com
Sun Sep 7 15:52:34 WST 2008


On Sun, 2008-09-07 at 12:56 +0800, Paul Antoine wrote:
> Gavin,
> 
> I've taken to running a memtest86 for a few hours whenever a system is a 
> little random in its behaviour.  The odd flakey bit in a bank of memory 
> can wreak such havoc and has proved the most common failure aside from 
> disk drive death.

Okay, good tip. I've got memtest but have never run it. Will do so in an
idle moment, probably overnight :-)

> Also, how old are the drives?  Of course with the two drives in an LVM 
> you have effectively doubled your statistical likelihood of a volume 
> failure as you have to add the MTBF's of the drives, so ageing may be a 
> real issue despite each drive being quite young.
> 
> How old is the whole system?
 
It's a Dell Precision 450 workstation, circa 2002, bought 2nd hand a few
years back. Some bits (like the scsi controller card) have been replaced
with newer, 2nd hand items. So really, it's of unknown provenance and
duty in those critical areas :-/

Yeh, I don't really get turned on by LVM for the reasons you state, but
the system defaulted to it on recent upgrade install, and I thought
"what the heck, give it a shot for a bit :-)" I always keep /home on a
separate drive, and in this case it's on a different controller, too.

Gavin.    

> P.
> 
> Gavin Chester wrote:
> > I suspect my system's drive controller is failing. Can anyone use these
> > symptoms (below) to confirm my suspicion? I know it's a long shot to
> > give definitive diagnosis, given many possible variables:
> >
> > Setup:
> > Two 36Gb SCSI drives setup as LVM giving 72Gb as one volume, running
> > from a PCI-X 64bit soclet using an LSI logic 2-channel U320 controller
> > through terminated internal 68-pin ribbon cable. 
> >
> > Symptoms:
> > After running for different, random lengths of uptime the system may
> > show odd behaviour in the way it's running its apps. Running 'top' as a
> > quick process check always reports "input/output error" at this point.
> > Or, same as above except it can be in that state as a cold system right
> > from booting. Forced reboot (once or even twice, occasionally) sees the
> > system recover it's journal and be okay for days on end.
> >
> > I had a similar problem a few years back with similar setup. In that
> > case I finally realised that it always coincided with very hot ambient
> > temps and the discs overheating. I'm since using an identical setup but
> > with different controller card and mb. I made a point of adding a fan
> > blowing directly on the drives - and anyway ambient temps are still
> > quite low in my workspace.
> >
> > Any ideas of what to test for?
> >
> > Gavin      
> >
> > _______________________________________________
> > PLUG discussion list: plug at plug.org.au
> > http://www.plug.org.au/mailman/listinfo/plug
> > Committee e-mail: committee at plug.linux.org.au
> >   
> _______________________________________________
> PLUG discussion list: plug at plug.org.au
> http://www.plug.org.au/mailman/listinfo/plug
> Committee e-mail: committee at plug.linux.org.au




More information about the plug mailing list