getsysinfo.sh (current version: 9.19, 24.09.2013 )
Latest revision of this document is also available on HP Support Center.
getsysinfo.sh is a data collection shell (ksh) script for HP-UX (≥ 10.20) used by Hewlett Packard
hardware support for Business Critical Server (BCS).
It collects diagnostic data for hardware troubleshooting from a running HP 9000 or Integrity
Server.
The script is optimized for small output and reliability even on systems with bad hardware. It not
only collects various log files, but also runs some common commands to obtain additional
troubleshooting information.
The result is a gzip`ed tar archive that can be viewed with WinZIP, tar/gzip or other tools. The
default output filename (can be changed with the -n option) is:
/tmp/sysinfo_hostname_timestamp.tgz
Some of the raw binary log files in the sysinfo.tgz archive need special viewers and can only be
analyzed by HP support.
New since Version 9.0 is the built in data collection from Onboard Administrators (option -oa) via
telnet. In addition to this getsysinfo.sh can now call the additional data collectors
hpmpl.285.jar - java tool to collect data from iLO Management Processors (MP) : option -mp
sd2collect.pl - perl script for data collection from Superdome 2 OA: option -sd2
sginfo -shell script to obtain Service Guard related information.
Note that these additional tools have to be present in the local directory unless you have one of the
embedded getsysinfo versions (getsysinfo.sd2.sh, getsysinfo.mp.sh, getsysinfo.all.sh) which have
one or all tools embedded as a shar archive.
Syntax (run as root !)
ksh ./getsysinfo.sh [options]
default output will be written to /tmp/sysinfo_hostname_timestamp.tgz
Options (no option is needed unless HP support asks to use a specific option)
-a : get all information except -diag (-h -f -san -xp -p -net -d)
-s : secure, use only non hanging (script hangs) commands
-h : get all files from /var/tombstones directory instead only those
: younger than 180 days.Collect SFM history and SAL records
-f : do not truncate some of the logfiles to 1024 kB
-f "regexp" : filter syslog files with regexp (Example: "inetd|ftpd")
-sh : make shell archive /tmp/sysinfo_hostname_timestamp.shar (ascii)
-u : uuencoded tgz file => /tmp/sysinfo_hostname_timestamp.uu (ascii)
-n FN: explicitly set output filename to FN
-d : bdf / diskinfo from all disks (scsimgr get_info on 11.31)
-san : capture fcddiag (or td-,fclp-,fcocdiag)
-net : capture netfmt /var/adm/nettl.LOG000
-p : include information from print_manifest
-diag: capture additional data for diagnostics SW troubleshooting
-diag -diag: capture more diag data - Huge Output may be the result !!!
-sas : force running of sasmgr
-sw : capture additional information needed for software support
-c [all] : run crashinfo on last or all crashdumps (<180 days)
-xp : get information about XP diskarrays with xpinfo (may take long)
-oa : capture show all etc. from onboard administrator (OA) via telnet
