[geeks] Agan, I seek the wisdom of the Collective ...

[Phil Stracchino]

I've been observing how my expanded array is behaving, and I have some
theories.

First of all, the hardware:  minbar is a U30 with a single 250MHz CPU,
512MB of RAM, 18GB and 36GB internal disks, two framebuffers (basically
unused), two SCSI channels, two hmes (one onboard, one on a
SunSwiftPCI).  It's running Solaris 9.  The array is twelve IBM
ultrastar 36XP disks spread across two 711 cases, one on each controller.

Now, the behavior:  I don't seem to be actually experiencing any data
*loss*.  But when the array is being accessed hard, it has a nasty
tendency to fall down and drop offline.  Umount it, fsck it, remount it,
and it comes back online just fine, so far with all data intact as far
as I can tell.

Speculation, on which I seek advice and experience:  Is my problem
simply that a single USII-250 isn't enough CPU for RAID5 on that
hardware?  It's run a six-drive RAID0 for about four years without once
falling down like this.

If this is in fact the case, would moving the array to a dual-400MHz U2
with 2GB RAM help?  I've heard it mentioned (but never seen conclusive
documentation) that the Sun metadevice drivers are single-threaded, and
would not benefit appreciably from SMP.  Does anyone know if this is
correct?


-- 
 Phil Stracchino       phil.stracchino at speakeasy.net
    Renaissance Man, Unix generalist, Perl hacker
 Mobile: 603-216-7037         Landline: 603-886-3518