[rescue] Re: Small RAID array setup
Curtis H. Wilbar Jr.
rescue at hawkmountain.net
Mon Jun 30 22:53:34 CDT 2003
>From: "Rip Loomis" <rip at flagon.com>
>> The one thing to keep in mind about using hardware raid of any sort,
>> is keeping a spare controller.... as if your controller dies, it
>> stands directly between you and your data and no amount of RAID 0,
>> RAID 0+1, or RAID 1+0 will help.
>Actually, it's worse than that AFAIK: One not only needs another
>RAID controller handy (for hardware RAID) but also a readable backup
>of the RAID config. If my primary DAC960 crashed then I have another,
>but I don't think I have a backup of the RAID config from the primary.
>If I try to initialize a new DAC960 with the same 4 drives that already
>have data on them, it will want to reformat them as a new (!) array.
>This wouldn't be quite what I wanted at that point...
> From what I can figure out in the Mylex docs (and other RAID
>controllers I've played with are similar) that's the whole point of
>allowing the RAID controller config to be backed up to a floppy.
>Just a thought for anyone who's concerned enough to be looking at
>getting a duplicate hardware RAID controller.
I have not looked into how to do this yet with my AMI MegaRAID cards
as I haven't gotten close to even begin the next generation of my
home server which will utilize one..... but it is something I'll
figure out when I get there :-)... thanks for pointing this out...
(again, kind of obvious, but not necessarily something immediately on
>So given the *BSD, Linux, and Solaris software RAID implementations
>out there, has anyone had any luck with recovering such a system
>after failure of the boot device (IOW, having to reconfigure software
>RAID on the same/a different system and trying to recover data that
>was in the software RAID array)? I'm curious to know...
I haven't had a catastrophic failure that way... however, with Sun's
DiskSuite, the MetaDB stores all this information. I alsways make way
more MetaDB duplicates than are needed (usually one per drive) and in
their own small slice. This will make recovery easier as I use the same
slice # on all drives... I only need one good drive to have access to
the MetaDB config of the array(s).
>Rip Loomis, CISSP
>Brainbench.com MVP for Internet Security | http://www.brainbench.com/
More information about the rescue