Strange multiple host graphing issue.

[cloud819]

I'm running 0.8.6j with the SNMP-php ping patch on a Cent 4.4 machine. On installation, I compiled net-snmp and RRDtool as in Lee Carters instructions. I used the CentOS php, snmp-php, mysql and apache rpms.

At the start everything seems to fine. I added servers and graphs populated as the poller ran. After few hours I noticed all of the graphs looked very much alike, a bit too much. After a day it was obvious something wasn't working right. All of the graphs were populating with information from from the cacti host (the first host to be polled). They all contianed the same information. The load averages are all the same, number of user logged on are all the same, etc. They aren't exactly the same but the differences are just enough to show that they are obviously stats from the same machine taken milliseconds apart. It is like the poller is incrementing to the next host, but still making snmp calls to the first one.

My logs look good, even with high logging turned on. According to the logs every host is being checked an returns a value. The servers all show beleavable ping times in the Management, Device menu. I gave the forum a good search and haven't found anyone with the same issue. I'm not sure what else to do.

Lee Carter compiled PHP in his install so I'm off to try that, but I can't imagine that is it.

Any ideas?

[gandalf]

I'm quite sure, that you always applied the "local linux machine" Host Template. Please change to "ucd/net-snmp template"
Reinhard