[ale] who is eating my drive

Greg Clifton gccfof5 at gmail.com
Tue May 31 17:49:10 EDT 2011


Obviously, if any of the remaining RAID member disks had a non-recoverable
read, the swap would be hosed, but I was just thinking that if it was
marginally readable and the disc had to make several read attempts to get a
good read, the time required for that process could cause a time out, even
though the data could be recovered with DD or such.

On Tue, May 31, 2011 at 5:38 PM, Jim Kinney <jim.kinney at gmail.com> wrote:

>
>
> On Tue, May 31, 2011 at 4:13 PM, Greg Clifton <gccfof5 at gmail.com> wrote:
>
>> I have no direct experience in the matter, but I would expect that there
>> could well be a problem during the rebuild when (not if) there is a block of
>> data that is difficult, but still possible, to read on one of the RAID
>> members. Seems the retries could cause the no swap panic. No?
>
>
> definate maybe. The failed drive was out so it was rebuilding off teh
> remaining 2. A second drive could have lost a block as well.
>
>>
>> GC
>>
>>
>> On Tue, May 31, 2011 at 4:00 PM, Jim Kinney <jim.kinney at gmail.com> wrote:
>>
>>>
>>>
>>> On Tue, May 31, 2011 at 3:45 PM, Pat Regan <thehead at patshead.com> wrote:
>>>
>>>> On Tue, 31 May 2011 15:34:05 -0400
>>>> Jim Kinney <jim.kinney at gmail.com> wrote:
>>>>
>>>> > On Tue, May 31, 2011 at 3:04 PM, Pat Regan <thehead at patshead.com>
>>>> > wrote:
>>>> >
>>>> > > On Tue, 31 May 2011 14:49:53 -0400
>>>> > > Jim Kinney <jim.kinney at gmail.com> wrote:
>>>> > >
>>>> > > > swap notes: NEVER put swap on a raid5. If you must park swap on a
>>>> > > > RAID rig, mirrored partitions are OK. If swap gets hosed, the box
>>>> > > > crashes. The RAID5/6 writes for swap are performance killers. Swap
>>>> > > > will do it own striping so don't bother doing a raid10. multiple
>>>> > > > swap parts on multiple drive spindles is a good thing. Say 8G RAM
>>>> > > > and 4 1TB drives, set up a 4G swap on each of the drives.
>>>> > >
>>>> > > Putting swap on RAID also means that your server won't go down due
>>>> > > to a disk failure.
>>>> > >
>>>> >
>>>> > Swap on RAID5/6 means during a rebuild the disk IO times out on swap
>>>> > retrieval and the kernel panics thinking the swap is down.
>>>>
>>>> I have never had this happen, although I can only recall one time that
>>>> I've ever had swap on a RAID 5.  It was a hardware RAID 5 and was the
>>>> only disk available.
>>>>
>>>> Is this only a problem with software RAID?  I'd be surprised if it is...
>>>>
>>>> In your experience with this situation, where exactly is the timeout
>>>> happening?
>>>>
>>>
>>> Saw this fail once a while back. swap on a hardware raid5 that lost a
>>> drive. the last line on the screen before the lockup was
>>> kernel panic swap error
>>> after the dead drive was replaced (hot swap setup) and the system was
>>> still running during the recovery.
>>>
>>> OK. So maybe the hardware could detect I'm no fan of RAID5 in any form
>>> and dislike hardware in most forms and just decided to take a dump because I
>>> was standing there.
>>>
>>>>
>>>> Pat
>>>> _______________________________________________
>>>> Ale mailing list
>>>> Ale at ale.org
>>>> http://mail.ale.org/mailman/listinfo/ale
>>>> See JOBS, ANNOUNCE and SCHOOLS lists at
>>>> http://mail.ale.org/mailman/listinfo
>>>>
>>>
>>>
>>>
>>> --
>>> --
>>> James P. Kinney III
>>>
>>> As long as the general population is passive, apathetic, diverted to
>>> consumerism or hatred of the vulnerable, then the powerful can do as they
>>> please, and those who survive will be left to contemplate the outcome.
>>> - *2011 Noam Chomsky*
>>>
>>> _______________________________________________
>>> Ale mailing list
>>> Ale at ale.org
>>> http://mail.ale.org/mailman/listinfo/ale
>>> See JOBS, ANNOUNCE and SCHOOLS lists at
>>> http://mail.ale.org/mailman/listinfo
>>>
>>>
>>
>> _______________________________________________
>> Ale mailing list
>> Ale at ale.org
>> http://mail.ale.org/mailman/listinfo/ale
>> See JOBS, ANNOUNCE and SCHOOLS lists at
>> http://mail.ale.org/mailman/listinfo
>>
>>
>
>
> --
> --
> James P. Kinney III
>
> As long as the general population is passive, apathetic, diverted to
> consumerism or hatred of the vulnerable, then the powerful can do as they
> please, and those who survive will be left to contemplate the outcome.
> - *2011 Noam Chomsky*
>
> _______________________________________________
> Ale mailing list
> Ale at ale.org
> http://mail.ale.org/mailman/listinfo/ale
> See JOBS, ANNOUNCE and SCHOOLS lists at
> http://mail.ale.org/mailman/listinfo
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.ale.org/pipermail/ale/attachments/20110531/676b880e/attachment-0001.html 


More information about the Ale mailing list