Well, I said I would report back on any progress...
I replaced the motherboard with a newer P31 based board and an Intel
Dual Core (Actually a core2duo with crippled cache) and 2GB RAM and
the issues have completely gone. I ditched the eSATA cards (Also 3512
based) and am using the 4 onboard motherboard ports plus the 2 port
sii 3512 card. The cables have mostly stayed the same, except for 3
disks which were connected using eSATA. I'm not too sure if the
supposed bug I found with multiple Silicon Image controller cards
actually exists or not and I don't really fancy risking my data -
however, if anyone is googling for sata_sil corruption lost interrupt
then they might find this message (And the backstory to it) :-)
The moral of the story here is: If you're going to be buying lots of
disks, do *not* skimp on cabling and controllers.
Thanks all for the helpful suggestions and "me too!" entries - my 6
disk raid5 is about a half hour away from completing resync and I've
had no nasties in dmesg. Win! :)
Twig
2008/5/8 berk walker <berk@xxxxxxxxx>:
> YES!! intermittent problems ALWAYS suck. WE need some really good diags
> which can be run as daemon, which can suspend I/O and associated times out,
> do pre-programmed things ie, notify of error, execute tests to the test
> cyl., and notify of correction which happened, if any, maybe
> possibly/probable thermal/physical/whatever tests to be run - AND have an
> associated pgm for the SU to run while executing these tests. (having come
> from industry) Maintenance engineering definitely sucks hind teat. I would
> think that many of our readers here, who support hundreds of instantiations
> like this would/could cause a remedy. If 10,000 disks are purchased from
> one vendor BECAUSE this is offered. I'll bet a hundred bux (U$, not worth
> much) that someone will sit up and take notice. Somehow, I think that the
> "who owns the market" has been ignored.
>
> YOU - the BIG guys own the market. And what you say holds sway, and we
> should support you, because, in the end we all ('specially us little guys)
> will benefit.
>
> b-
>
> Twigathy wrote:
>>
>> hm, well I have a couple more eSATA -> SATA cables on order, so I'll
>> report back once they arrive and I can test my disks with them.
>> Intermittent problems totally suck :-(
>>
>> 2008/5/8 Greg Cormier <gcormier@xxxxxxxxx>:
>>>
>>> I had scary messages in dmesg like yours above. Spent forever
>>> troubleshooting everything. New SATA cables and they disappeared.
>>>
>>> Greg
>>>
>>>
>>>
>>> On Wed, May 7, 2008 at 10:52 AM, Twigathy <twigathy@xxxxxxxxx> wrote:
>>> > PSU is a relatively reliable, if slightly old now, 380W thing. It has
>>> > enough power on the 12v lines to power the disks and CPU, but you may
>>> > be onto something there - all disks reading/writing at once = power
>>> > levels higher.
>>> >
>>> > The new kit I've ordered for this server includes a 450W PSU, so
>>> > perhaps that'll solve things.
>>> >
>>> > The trouble with debugging a problem like this is that there are so
>>> > many variables!
>>> >
>>> > 2008/5/7 Maurice Hilarius <maurice@xxxxxxxxxxxx>:
>>> >>
>>> >> Twigathy wrote:
>>> >> It could be cables, although that wouldn't explain the disks working
>>> >> perfectly well (Maxed out, too in the case of doing a badblocks test)
>>> >> when they are by themselves.
>>> >>
>>> >> Sounds like a current delivery issue.
>>> >>
>>> >> How is your power supply?
>>> >>
>>> >>
>>> >> --
>>> >> With our best regards,
>>> >>
>>> >> Maurice W. Hilarius Telephone: 01-780-456-9771
>>> >> Hard Data Ltd. FAX: 01-780-456-9772
>>> >> 11060 - 166 Avenue email:maurice@xxxxxxxxxxxx
>>> >> Edmonton, AB, Canada http://www.harddata.com/
>>> >> T5X 1Y3
>>> >>
>>>
>>>
>>>> --
>>>
>>> > To unsubscribe from this list: send the line "unsubscribe linux-raid"
>>> in
>>> > the body of a message to majordomo@xxxxxxxxxxxxxxx
>>> > More majordomo info at http://vger.kernel.org/majordomo-info.html
>>> >
>>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-raid" in
>> the body of a message to majordomo@xxxxxxxxxxxxxxx
>> More majordomo info at http://vger.kernel.org/majordomo-info.html
>>
>
>
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
