I've just installed 0.6.8, and was wondering about scaleability. I currently monitor 600 interfaces with classic MRTG, and 1800 with RRDTool (in the middle of migration to rrdtool and adding more devices). I acheive this level by parallel launches of mrtg/rrdtool. Each device/router/switch has a seperate config file, and the 5 minute cron launches each one a few seconds apart. viz:
for conffile in *.conf
do
mrtg $conffile
sleep 3
done
By adjusting the sleep period, I can flatten the load so that the last script has launched just before the 5 minutes is up. Because much of the polling is waiting for IO (SNMP) the load is not huge even with a dozen sources being interogated at once, and each script/data source is run almost exactly 5 minutes apart from it's last invocation.
What I'm trying to ask is can Cacti (cmd.php?) when launched from the cron every 5 min, spawn each data source independantly in such a manner to ensure that they are all updated before 5 minutes is up? This ensures that all sources are updated, even if one of them stalls for a longer period than normal. If there is too much to monitor, the load average on the host will show it.
I am in the middle of upgrading from a Sun Enterprise 250 (Dual 400Mhz) to a dual 2.6Ghz Xeon running RH8, and was hoping to increase the number of devices monitored further. It appears Cacti unfortunately might not be able to keep up, which would be a shame.
Thanks.
parallel rrdtool calls
Moderators: Developers, Moderators
Here is a thread discussing forked cmd.php solution: http://www.raxnet.net/board/viewtopic.php?t=349
- bulek
- bulek
Who is online
Users browsing this forum: No registered users and 3 guests