From: Mike Tomlinson on
In article <tSFDTLAI$mlLFw6+(a)none.invalid>, Mike Tomlinson
<mike(a)none.invalid> writes

>I don't think this is the problem here. The box has two SCSI cards -
>the first one is attached to a RAID with 16 drives. This contains a lot
>of important data.

I forgot to add:

The second (card) has two smaller RAIDs and a tape drive attached. It
is the two smaller RAIDs that I am trying to combine and use as a backup

Mike Tomlinson
From: Mike Tomlinson on
In article <+L3X1jAbmPjLFw47(a)none.invalid>, Mike Tomlinson
<mike(a)none.invalid> writes

[writing to SCSI array fails and causes the _source_ drive to go read-
only during heavy disk activity]

Just a follow-up. It turned out not to be a filesystem size issue at
all, but an obscure interaction between kernel 2.6.18 and the LSI Logic
'mpt' SCSI driver.

This is on a CentOS 5.2 box. The release notes for 5.4 indicate a lot
of work has been done on the SCSI layer, and the LSI mpt driver is newer
and fixes several issues.

"The MPT fusion driver is now updated to version 3.04.07rh v2. This
applies several bug fixes, including: (BZ#475455)"

(I can't view BZ#475455 as it needs a username and password and I can't
be arsed registering)

An update of CentOS 5.2 to 5.4 on the box is impractical for several
reasons, so as a workaround I am going to try swapping the LSI Logic
U320 SCSI card with an Adaptec 29160 so that the LSI driver(s) are not
loaded and thus will not tickle this bug.

(='.'=) Bunny says Windows 7 is Vi$ta reloaded.