[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [SLUG] Can hard disk behaviour be an indicator of a dodgy power supply
> While I can't say for sure in your instance, intermittant power fluctuations
> can produce surges in your low voltage computer power. Typically, such a
> fluctuation would need to be >0.2s, just enough for the caps in the power
> supply to partially discharge, but not enough for them to totally discharge
> (around 0.3s). As the power supply switches on again and "fills" the caps with
> current, this can produce a small spike on your 5V and 12V power rails,
> typically resetting hard drives. This reset can cause failure in the drive if
> the surge is of sufficient power or you experience a double-surge very quickly
> which resets and then resets the HDDs again before they have a chance to fully
> If your affected hard drives are a few years old, I would expect that this is
> what has happened and, unfortunately, there is not a lot to be done. A good UPS
> will usually solve the problem.
I dont think this is the case as this would be causing the PC to reset
at the same time - something that wasnt mentioned.
One of the outputs from the supply is a power good signal to "cope"
with problems such as these. The power good signal will go away
well before the 5/3v rails go significantly out of spec. The power
good signal changing state will cause the PC to do a cold reboot.
That said, if the power supply is old the caps on the 5 and 12v rails
will probably be suffed. Possibly enough that heavy loads (such as
multiple drives - you have 4?) will cause the rails to go noisy giving
data errors. This wouldnt cause physical damage to the drives
Probably worth throwing a new PSU in anyway as they are very
cheap (which is part of their problem anyway).
> You can quite easily kill your entire computer (literally), merely by
> flicking the power on and off 6-8 times in rapid succession. (BOFH Hint #1024)
Sigh, and so propagates yet another urban myth. It is possible, but
no more likely than waking up in an ice filled bath with no kidneys :-)
From the description posted of the symptoms I'd agree with another
poster on the list and have a good look at the airflow around the
HDD's. It could be that the drive position that keeps failing is getting
quite hot. Or that the drive type in question runs significantly hotter
than the other types in the machine.
With hard drives (and most mechanical bits) MTBF is inversely
proportional to temp.
Its quite easy to check for excessive heat with one of those
indoor/outdoor thermometers blu-tacked onto the drive metalwork.
Just make sure that the sensor is in contact with the drive body and
is NOT exposed to the air.
Above 50C on the drive case should be cause for concern.
IMHO - heat is the biggest killer of hard drives bar none - the lower
temp the better.
> On 17 Jun 01, at 23:24, Terry Collins wrote:
> > Has anyone had any experience with a dodgy power supply killing scsi
> > hard disks?
> > I've just rebuilt a file server a number of times as a succession of
> > scsi hard disks have failed. The pattern was fairly similar.
> > They would become musical - the ting ting ting of the "head" hitting the
> > stop, then about 24 hours later, there would be major problems with
> > large numbers of bus errors then total failure. On each occassion, it
> > was the first drive (/dev/sda) that failed, so this meant that Linux was
> > reinstalled each time.
> > As the drives were all the one brand and model IBM DCAS-34330 (4Gb
> > single ended) and they have been in used continuously for 4 years, I was
> > inclined to think dodgy model or dodgy batch. However, I heard musical
> > notes again tonight and all that remains are three drives of a different
> > brand and model. At the same time, the tape drive seemed to drop power
> > (which was kind of not appreciated as I was in the middle of a 20Gb
> > backup off three machines) but it recovered and carried on.
> > So I'm wondering if my problem is really a intermittently dodgy power
> > supply and asking if anyone has experienced anything similar.
> PGP Key Block available at:
> SLUG - Sydney Linux User Group Mailing List - http://slug.org.au/
> More Info: http://lists.slug.org.au/listinfo/slug
Hardware R&D Manager
TPG Network, Canberra
"bad or missing Coffee.sys - operator halted"
SLUG - Sydney Linux User Group Mailing List - http://slug.org.au/
More Info: http://lists.slug.org.au/listinfo/slug