Sorry for late summary, but still evaluating.
Answers from many. Thanks to all.
Big Brother with LARRD and RRDTOOL
mrtg/rrdtool
collect, monitor
Don't load sar. Too expensive and higher on CPU load.
BMC Patrol is incredibly expensive.
I use Compaq product, ECP and pawz
TeamQuest
Hope I got all listed.
Found collect+perl+ksh+rrdtool to expensive on overhead. I wanted to log
data every 2 seconds. Also collect got stuck once or twice grabbing all CPU
time. {BAD/BAD, some bug}.
OK, now for all those interested. Wrote a small C program that logs values
to a rrd database every 2 seconds. Hardly any extra load/overhead generated
by this prog. Make use of Tru64 "table" function". Gets data directly from
kernel.
Anybody with up to date MAN on this function ???
NO SNMP OVERHEAD.
Mail me for a copy.
Perhaps not need to measure every 2 seconds. Actually nice with some stuff
where you can see where a database checkpoint.
Also you can get any value via SNMP, bet surly extra overhead due to SNMP.
Still have to measured extra overhead.
-----Original Message-----
From: Hannes Visagie [mailto:visagi_h_at_mtn.co.ug]
Sent: Thursday, 29 March, 2001 1:09 PM
To: ALPHA-System-Managers (E-Mail)
Subject: Best tool to monitor system and produce weekly load statistic
gra phs
Hi all
What do you guys recommend. I need to produce stats for system load,
disk/cpu/swap/net, etc on a daily/weekly/monthly graph.
This will be for a 2 mem cluster Tru5.1. All graphs are per member and
perhaps also for cluster.
sar ? {Will load system 5 tools}
BMC Patrol ?
Performance Manager ?
Thanks
Hanes
Received on Tue Apr 17 2001 - 08:45:11 NZST