Cacti stop graphing for some minutes and the start again.

[rrestituyo]

The issue happen for some graphs and some devices only and at the same time.

I do not know why. Can someone help?

Please see the photo attached.

Cacti.PNG (88.82 KiB) Viewed 1677 times

[camerabob]

Look into the underlying operating system to see if it is being taxed at that time. Sometimes a virus scan or other background process may be using CPU cycles and that will affect how quickly Cacti can get it's job done as well.

[rrestituyo]

I check for CPU usage and everything but I do not see anything wrong and the problem still there.

[netniV]

That usually means one of a few things. The first is an overloaded resource on the Cacti server, check your CPU/RAM and running processes. I would suggest having 'top' running (though that too could be too resource intensive). The second is the device in question and could be that the device is being overwelmed by requests (whether from the cacti server or just its main function). The third is that the values being returned via SNMP are too high to be included in the RRD file (due to the maximum value setting normally associated to the template). To check the third, use the data source debugging tools in Cacti 1.2.x or the dsdebug plugin in 1.1.x

[rrestituyo]

Hi,

I checked cpu and memory and top, but all looks good. The issue It is not to all the graphs.

[camerabob]

Look at that graph. If you dig deeper, something is occurring on a regular basis. Do you have an offsite backup software configured? Look carefully at the missing graph times, find the pattern and look when the next one is to hit. Also, looking again at the processor graph, I would second netniV. The device's processor usage is through the roof. I would bet a dozen donuts that it is being overwhelmed for that 5(?) minute stretch and simply cannot respond to polls. Also your statement

I checked cpu and memory and top, but all looks good. The issue It is not to all the graphs.

would point to the graphed device being the one causing the gaps.

[rrestituyo]

I followed the pattern and monitored my device when the next blank was suppose to happen but everything was okay there. Memory did not increased and CPU did not decreased as my graphs are showing. There is not backup software.
I'll keep looking.

[camerabob]

Look at the far end device, not the Cacti server.

[rrestituyo]

Hi Guys,

Yesterday I checked the difference between devices with the problem and devices that did not have the problem. The only thing I saw was that the re-index method for the ones with graph issues was in uptime but the other were in index count. I changed to index count and today I do no see the problem. I left some devices in uptime to see if that was the issue and those still have bad graphs but the ones I changed are okay.

I will keep monitoring for a while to see if that was the permanent solution and let you know. Does someone have an idea of why in uptime this happens and not in index count?

Thanks!

[Waterboy2020]

I have the same thing, most day's about 3:00am.
I'm running Cacti on Windows, I checked the event log and found Symantec is running a scan about that time, I think I saw something about Cacti not being able to access the database in some log?....
Probably need to put exceptions in for Symantec to not scan some directories