RAID 5 Multiple Hard-drives failure
Chris Wright
linux-list at cwic-solutions.co.uk
Tue Mar 14 18:12:26 UTC 2006
Bob Chiodini wrote:
> On Tue, 2006-03-14 at 10:26 -0500, Reuben D. Budiardja wrote:
>> On Tuesday 14 March 2006 08:55, Bob Chiodini wrote:
>>> On Tue, 2006-03-14 at 08:21 -0500, Reuben D. Budiardja wrote:
>> <snip>
>>> Reuben,
>>>
>>> Have you checked the power supply?
>> I have not checked the power supply for the system. Any recommendation on how
>> to do so ?
>>
>>> Have all of your failures been in the same machine?
>> Yes, all in the same machine, that's why I suspect there's something else
>> physically wrong.
>>
>> Thank you.
>> RDB
>> --
>> Reuben D. Budiardja
>> Dept. Physics and Astronomy
>> University of Tennessee, Knoxville, TN
>>
>
> Reuben,
>
> Open up the case and find an unused drive power connector. Measure the
> voltage between the yellow and black, should be ~12VDC. The voltage
> between the red and black should be ~5VDC. I'm not sure what the
> tolerances are for your drives, maybe it's on their website. As a rule
> of thumb, I'd not let the 12V get below 11.9V or above 12.1V (about
> 10%). The 5V should be above 4.8V and below 5.1V. These are guidelines.
> The last time I opened a PC power supply the 12 and 5 volt supplies were
> not independently adjustable.
>
> Your BIOS may also tell you these voltages, along with various
> temperatures. lm_sensors might work as well.
>
> Bob...
>
One thing to try when testing the power supply as above is to make sure
you 'load' the system.
When the system is idle (or close to idling), the 12V for example, might
be 11.99V. But when you start spinning up multiple hard drives, this
puts a load on the power supply.
If there is fault with the power supply, it will cause the voltages to drop.
'Most' PC PSU's are meant to be constant voltage type, ie. they try to
maintain their voltage outputs as close to spec as possible
(+5v,-5v,+12v,-12v, 3v etc etc etc). If you took most PC PSU's out of
the case and measured their outputs with nothing connected, you would
read near exact values.
It is only when you place them under load that the outputs will vary or
indicate a fault.
So while measuring, try to run some sort of 'stress' test on select
items of kit / various items of kit.
Also worth pointing out, is that a lot of people these days are
inserting extra fans inside their machines, and these sometimes have
thermostatic controllers on them to switch them on at a certain temp.
Some fans are electrically noisy, and of course, that noise will also
only be present when the fan is running. Likewise, they will only
'load' the system when they too are running.
So in summary, make sure you are running as much of a stress test as you
can (without making it a destructive stress test).
Regards
Chris
More information about the fedora-list
mailing list