Different Readings from fping and Advanced Ping

[NWA Cacti Admin]

Greetings,

We installed Cacti a couple months back and are steadily adding tracking capabilities. We currently have both fping and "Advanced Ping" set up (ee the attached graphic which shows fping (left side of picture) and Advanced Ping (right side of picture) results). Based on this a few questions:

Technical Environment:
Cacti version 0.8.6h
Windows Server 2003
IIS


Thanks in advance for any help.

-mark

[NWA Cacti Admin]

Sorry, here are the questions:

First, why are the results different between fping and Advanced Ping. I could understand subtle differences based on pings occurring fractions of a second apart but the Advanced Ping's results are consistency and almost always significantly higher. For example the approximate longest response time for Google via fping is just under 300 ms where Advanced Ping shows over 800 ms.

Also, while the peaks and valleys between fping and Advanced Ping are generally similar there are differences. Look for example at the Google fping response. After the Mon 12:00 time marker the responses remain about the same magnitude with fping but drop dramatically with Advanced Ping.

[NWA Cacti Admin]

Second, there are gaps in the graphs for Advanced Ping see especially Application Server ping. Do gaps indicate 100 percent packet loss. If so, I thought packet loss was shown in red color. Why would Advanced Ping show 100 percent packet loss while fping continued to produce results.

[NWA Cacti Admin]

Third, all graphs show a small spike in what would be late Sunday night. This would indicate something happening on our end that slowed all network packets. However the timing of this spike varies between the graphs. Assuming that each time the poller runs, every 5 minutes, each RRD is updated, shouldn't the end occur at the same time among all graphs.

Sorry for the elongated post, I kept getting "SQL Error : 1271 Illegal mix of collations for operation ' IN ' " errors. mdr

[BSOD2600]

1) You didn't upload any attachments

2) I don't use either ping tool, but my first guess would be the ping method they're using (i.e. udp vs icmp) which could give different results. Another thing could be the timeout values they use.

[NWA Cacti Admin]

Sorry here is the attachment.

[TheWitness]

Please upgrade to Cacti 0.8.6i. I had some issues with out of order packet returns messing up the numbers. This was corrected in that version. If you have several hosts (over 10k data sources), you will need to install a patch. Otherwise it's proven stable.

Then, if you want to see RED during 100% packet loss, in this version anyway, you must add a host with no community string and use snmp for pinging. Sorry, it's not the way I wanted it either. But you are correct in your assumption.

TheWitness