[rrd-users] Performance problems using mrtg LogFormat: rrdtool & snmpv3

[Michael Heiming]

Hi!

I'm using mrtg with "LogFormat: rrdtool" and have some performance 
problems on the mrtg server with
solaris targets, running net-snmp (v3).

 From what I see, the problem is in the mrtg.cfg (one per system):

Target[fqdn.something]:`snmpget -t 4 -w 4 fqdn 1.3.someoid 
1.3.someotheroid | awk '{print $NF}`
[...]

There is no additional info needed for snmpv3 MD5 authentication and DES 
encryption, snmptools are compiled to default to v3, the user it runs 
as, has ~/.snmp/snmp.conf with the needed security options.

snmp v2 works without calling snmpget, which looks like my problem, the 
above construction works, but spawns additional procs. The -t/-w is 
needed, or I don't get any info from busy systems. It is running from 
cron, starting a script which calls mrtg with the config serial, 
querying 3 systems at once or the 5 minute interval won't be long enough 
to get all targets queried.

Is there any way to enhance performance, other then using a mrtg server 
with more CPUs? Right now the box is a dual 1.4 GHz P3, with 1 GB RAM 
and a 15K SCSI hw RAID1 dedicated to the .rrd files running Linux.

RAM is fine, I/O isn't a issue, just the load keeps at 10-15, with 
sometimes 0% idle left, it runs with nice 10, to keep the system responsive.

Thx for any info to improve performance

Michael Heiming

--
Unsubscribe mailto:rrd-users-request at list.ee.ethz.ch?subject=unsubscribe
Help        mailto:rrd-users-request at list.ee.ethz.ch?subject=help
Archive     http://www.ee.ethz.ch/~slist/rrd-users
WebAdmin    http://www.ee.ethz.ch/~slist/lsg2.cgi