I had noticed lots of FSCSI_ERR10 ‘Configuration mismatch’ messages with identifier 1FB4FD21. The errpt entry had the PID as 12D0E4 hex in the sense data:
SENSE DATA
0000 0000 0000 00D3 0000 0016 0200 0000 0000 0000 0000 0000 0000 0000 0000 0000
0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 00C8 0800 0000 0000
0000 0000 0300 0000 0000 0003 0000 0000 0012 D0E4 0000 0000 00C8 2500 03FD 0000
Which translated to 1233124 decimal. I setup an ODM errnotify entry to dump the process table when the driver error occurred:
server–> odmget -q “en_name = mousetrap” errnotify
errnotify:
en_pid = 0
en_name = “mousetrap”
en_persistenceflg = 0
en_label = “FSCSI_ERR10″
en_crcid = 0
en_class = “”
en_type = “”
en_alertflg = “”
en_resource = “”
en_rtype = “”
en_rclass = “”
en_symptom = “”
en_err64 = “”
en_dup = “”
en_method = “/usr/bin/ps -efk >> /tmp/ps.out”
Wasn’t long before more errors showed up. The process turned out to be xpinfo an HP supplied command to list LUNs on an HP XParray storage system.
The misconfiguration was because the command used a legacy ioctl call that left the N-PORT ID blank, which isn’t supported when Dynamic Tracking (dyntrk) is enabled. The version of xpinfo was the latest:
AIX XPinfo, v 21.00.00 2010-02-21
I later realised this problem didn’t occur on another system which the same version of AIX. I need to check if RMLIB is installed correctly on both systems.
P.S. Have now checked RMLIB and it was version 5.1 on the problem system and 5.8 on the working system