[volunteergrid2-l] Redundancy

[Shawn Willden]

Well, for storage you donate to the grid, there's really no need to provide
redundancy.  The grid and your client configuration parameters provide
that.  My grid storage is on a RAID5 array, but that's just because it was
convenient to carve a logical volume for it from an existing RAID5-based
LVM volume group.

I don't personally see much value in hardware RAID.  In some situations
hardware RAID performance can be slightly better than software, but it's
unlikely to be enough better to matter -- and in some cases software raid
is faster.  For reliability, you have to remember that your RAID card
becomes a single point of failure, and if the card dies, you need to have
another one that will be able to understand the layout of the data on your
disks.  Some people choose to keep an identical spare card on hand to
mitigate this problem.  With software RAID you really don't have to worry
about it.  Even if you have to replace disk controllers and your drives end
up "moving around", MD can still figure out how to use them.  Even in cases
where things get completely hosed, you can often force MD to recover by
forcibly re-creating and re-starting the array with the disks in the order
they were before the failure.

I haven't used ZFS, but I hear it's a great option as well, especially with
the built-in integrity checking.

One thing I highly recommend is forcing a periodic read of the full content
of each disk.  Debian (and, I'm sure, other Linux distros) provides a tool
to do this.  I don't recall the name right now.  Another option is just to
have a cron job that runs "cat /dev/sda > /dev/null" every month or so,
being careful to stagger the runs against each disk.  The purpose of this
is to get the disk to identify any bad sectors and try to reallocate them
-- or to fail if it can't.  This is to flush out any latent failures
earlier rather than later, so you don't find yourself in a situation where
attempting to reconstruct your RAID array after a failure uncovers another
latent failure on another drive, and in so doing hoses all of your data.

For important data that isn't backed up elsewhere (though that's risky in
and of itself), I also recommend RAID6 or RAID10 over RAID5.  I almost lost
some irreplaceable and very important data once because it was on RAID5,
and because I wasn't scanning all of the disks periodically.  Now I put
that data on RAID6, scan it regularly, and back it up to VG2.

On Mon, Apr 23, 2012 at 11:24 AM, Samuel Willcocks
<samuel at willcockses.com>wrote:

>  Hi list, one of the new lurkers here.
> I'm looking into options for providing redundancy - what do you grid
> members use? Hard/software RAID? ZFS? I have looked briefly into RAID
> cards, can you recommend any over others for reliability/compatibility with
> Linux/etc?
> Cheers,
> Sam
>
>
> _______________________________________________
> volunteergrid2-l mailing list
> volunteergrid2-l at tahoe-lafs.org
> https://tahoe-lafs.org/cgi-bin/mailman/listinfo/volunteergrid2-l
> http://bigpig.org/twiki/bin/view/Main/WebHome
>



-- 
Shawn
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://tahoe-lafs.org/cgi-bin/mailman/private/volunteergrid2-l/attachments/20120423/23a9cb98/attachment.html>