Annoying RAID behaviour on X4600

Hi All

Not sure if I was doing something wrong, so any pointers would be welcomed. I
would suspect others may have seen this or am I just unlucky?

Scenario

I attended site yesterday on a dead X4600M2 which as it turned out had a
faulty CPU and motherboard, so we swapped the chassis as simplest course of
action and transferred over the PCI cards and the disks (being careful to
retain the disk positions). This solved the dead issue with the system and all
looked well. Now onto the problem.....

The customer had setup the three disks as a mirror'd pair and hot spare, I
went into the LSI BIOS and resync'd the primary to the secondary which worked
fine. Then booted the system. It came to the initial splash screen for Solaris
10 and then proceeded to load, after a short while it came up with a root
filesystem not found error and crashed.........

At this point I should note the order of the disks in the system as per
original system,
0 - Hot Spare
1 - Primary
2 - Secondary

I assume originally that the admin who setup the RAID would have used 0 & 1
and that the hots spare would have been 2, but obviously over time the
original hot spare had been used and another installed in slot 0.

I booted a DVD and sure enough I could see the full O/S on the disk which
looked 100%. My next port of call was firmware versions, was this due to
differences between chassis, the replacement had the latest firmware, the
original some quite old firmware, I nearly got to the point of downgrading the
system, but then noticed a curious thing.

Please bear in mind there is just one RAID volume and that is all the LSI BIOS
reports etc. On boot the LSI discovery screen shows one Volume BUT sets its ID
to 1 (the Solaris installed expected 0 as normal) hence the crash on boot. So
why oh why did the LSI controller assign 1 to the volume? It is not
configurable from what I could tell in the LSI BIOS, it just does it!

The answer was to rejig all the disks so the primary is in slot 0. Am I right
that the LSI picks up the slot number of the primary as the ID on boot?

This is crazy for such a system with Hot Swap disk as if you are managing
remote and have no DVD in the drive,  you cannot get the system to boot and
you physically need someone to swap disks if the drive in slot 0 is not part
of the RAID set.

Any comments on this would be appreciated, especially if I have missed the
obvious config option, but given the wide use of this controller it may be
something that others need to know especially for remote management.

Peter
---------------------------
Peter Stokes
Ashlyn Computer Services
Tel: 01636 627990
Mbl: 07977 532320
---------------------------