[wellylug] Hardware raid and Debian

Ewen McNeill wellylug at ewen.mcneill.gen.nz
Tue Feb 8 16:44:17 NZDT 2005


In message <4208258F.1010507 at mowgli.net.nz>, Rob Stockley writes:
>I have an Adaptec PM2654U2 raid card running a four disk raid-5 array. 
>The array was created using the card's BIOS utility and is accessed 
>under Debian Woody through the standard dpt_i2c kernel module.
>
>Does anyone know of any Linux tools for accessing this card in a running 
>system? 

The "raidutils" package (in Debian Testing and above) will support these
cards I believe (I think it's the same DPT-derived interface as some of
the other cards I've managed).  From memory there's also a Redhat RPM
available from the Adaptec site with the same tools which can be
converted (eg, alien) to do the job.

>The card has a warning tone on it that sometimes goes off. There 
>must be something

As you've discovered it sounds like it's failed one of the disks in the
array.  You should figure out which disk it is, and then pull it, and
replace it, then force the array to rebuild.

You want something like:

raidutil -L physical
raidutil -L logical

to figure out what is there, and which one is failed.  Then something
like:

raidutil -a rebuild LOGICALDRIVE

to force it to rebuild _once_you've_changed_the_disk_.  (Don't rebuild
onto the failed disk; it might work for a while, but it might make
things worse later.  Rebuild onto another disk and then test the failed
disk -- eg SMART tests, badblocks -w tests, etc, and either get it
replaced or put it into your spares with a note that it's been failed
once.)

IIRC there's also a way to acknowledge the alarm and silence it for now,
which is very useful for ones sanity, but I don't remember the exact
command for it right now.

Good luck.

Ewen




More information about the wellylug mailing list