Back on Mar 30th I mentioned a recurring error on two
identical fibrechannel-connected raids connected to two
very similarly configured Origin 350 machines. They both
show errors like this:

unix: ALERT: /hw/module/001c02/Ibrick/xtalk/14/pci/2a/scsi_ctlr/0/target/112/lun/2: No SCSI status, key=825
unix: ALERT: /hw/module/001c02/Ibrick/xtalk/14/pci/2a/scsi_ctlr/0/target/112/lun/2: IO Terminated, key=825

These pairs of errors recur and eventually I get something like this:

unix: ALERT: /hw/module/001c02/Ibrick/xtalk/14/pci/2a/scsi_ctlr/0/target/112/lun/2: IO Terminated, key=825

The RAID itself indicates no errors at all.
A similar fibrechannel attached raid on an Origin 3900 system shows no errors.
The fact that this occurs on two sets of identical equipment suggests that it is not a
hardware failure but more likely a configuration and/or compatability problem. Efforts
to generate this problem with command line instructions such as "cp" or "tar" failed.
A perl program that does a "copy" does regularly generate the problem.
Running this program on a remote host and accessing the RAID via NFS never generates
the error. Running this program on the local RAID and using the perl optional third
parameter (buffer size) succeeds when this parameter is reduced to about 512k.

We have tried altering the connection speed and see failures at all three
possible speeds (4gb/s, 2gb/s, and 1gb/s). Any suggestions are appreciated.