Fragmented graphs in cacti

[mahade047]

I have been using Cacti to monitor my network for almost a year. But for last 6 days some of the devices are producing broken graphs. Not all the devices are doing this. I have posted a screenshot of a switch, you can see that the graph was ok untill last 5/6 days. The devices both router and switch that are doing this are producing fragmented graphs for all the ports.

graph frag post.JPG (67.27 KiB) Viewed 1225 times

I cant understand why this is happening! please help. thanks

[cigamit]

Check your Cacti log. First thing to check is to make sure you aren't going over your polling time.

[mahade047]

where can I see the poller time? in settings>poller ??
and if that was the problem shouldn't all the device produce broken graphs!!! Like I said only a few devices(not all) in the network are making broken graphs!!!

[cigamit]

System Utilities >> View Cacti Log File
There is usually all sorts of good information about what is going wrong in there.

If you system is taking too long to poll, then its possible that only a few (the last few remaining devices) are not being polled correctly. There could be any number of different things going on, so until you look at the log and report back, there is virtually no way to tell what is going on.

[mahade047]

thanks .. I looked into the log file.. but what should I look for . there are mainly 2 kinds of warnings for different devices, like

WARNING: Result from SNMP not valid. Partial Result: U
WARNING: SNMP Get Timeout for Host:'x.x.x.x', and OID:'XXXXXXX'

[cigamit]

That last error pretty much tells the story. The Cacti host is timing out when contacting the other host. So either network issues or the remote host is having issues responding in an appropriate time (based upon your SNMP timeout settings). You can ofcourse try to up the SNMP timeout for that host.

[mahade047]

thanks.. I have set the SNMP timeout to 2 second [by default it was 500 ms] I will let you know what happens.. and whats the first warning?? Some of the devices which ID the warning shows also making wrong interface names, not showing interface information... like

Switch HW S9312 - Traffic - |query_ifName|

I have tried "re-apply suggested names" but it didnt work!

[mahade047]

I have tried changing the SNMP uptime for some devices to 3/5 seconds. But the graphs are still broken. Any idea whats happening !!
thanks