b p e r f |
bperf is a shell script that calls gathers performance data for Brocade Silkworm or HP B-Series switches, then stores it in a RRD database. Performance data is then plotted and stored on a web server.
Consider this as beta software. It's been running fine for me for a while, but
it may not be the best solution for you. Documentation is limited. If you already run cacti
or cricket in your shop, you definetely won't want to use this! If you have any trouble
installing this, drop me an e-mail.
In 2006, I was looking for a way to plot graphs with performance data for my
HP B-Series switches. The only official solution was to purchase an expensive
piece of software that did more than I needed. Since I was looking for a reason to learn
RRD Tool, I wrote my own in a
few hours after reading the RRD Tool documentation. I also wrote this tool since I found
cacti
to be an overkill solution which would take too much time to implement.
bperf has been developped and tested on HP-UX 11.23 only.
bperf is released under a BSD license.
Collection is done as following:

Ports are pooled each 30 seconds. Polling must be done frequently as the counter is 32bit and it quickly gets wrapped; it cannot wrap twice or else the data is lost.
1. Grab a working copy of rrdtool. On HP-UX, you'll probably need to build it by yourself and it is beyond the scope of this webpage to explain this. The building instructions given by RRD Tool's author are easy to follow.
2. Install snmpget. You can get it easily from the Net-SNMP package on the HP-UX Internet Express DVD.
3. Download bperf here: bperf.tar.gz. It will untar in ./opt/bperf.
4. Open the file /opt/bperf/etc/bperf.cfg and configure it for your site.
5. Add a startup script in /sbin/init.d, along with a hook in /sbin/rc3.d, that starts bperf as
following:
nohup nice -19 /opt/bperf/bin/bperf.sh /opt/bperf/etc/bperf.cfg collect >/dev/null 2> /var/adm/syslog/bperf.errorlog &
Note: For added security, the collectors can be run as another user than root.
6. Setup a few cron jobs like these to graph your data:
00,05,10,15,20,25,30,35,40,45,50,55 * * * * nice -19 /opt/bperf/bin/bperf.sh /opt/bperf/etc/bperf.cfg graph daily-small >/dev/null 05 * * * * nice -19 /opt/bperf/bin/bperf.sh /opt/bperf/etc/bperf.cfg graph daily >/dev/null 1 * * * * nice -19 /opt/bperf/bin/bperf.sh /opt/bperf/etc/bperf.cfg graph weekly >/dev/null 2 21 * * * nice -19 /opt/bperf/bin/bperf.sh /opt/bperf/etc/bperf.cfg graph monthly >/dev/null 3 21 * * 1 nice -19 /opt/bperf/bin/bperf.sh /opt/bperf/etc/bperf.cfg graph yearly >/dev/null 05 00 * * * nice -19 /opt/bperf/bin/bperf.sh /opt/bperf/etc/bperf.cfg archive >/dev/null |
