cancel
Showing results for 
Search instead for 
Did you mean: 

x670, snmp oid for ALL packets counter on interface

x670, snmp oid for ALL packets counter on interface

amindomao
New Contributor

Hello!

I’m trying to graph packets/s in cacti using standard counters - IF-MIB::ifInUcastPkts.1034, IF-MIB::ifOutUcastPkts.1034. But this is not correct way, at least for rx.

In “dark” screenshots you can see pkts/s from both sides of the same link. Values are almost the same. But in “cacti” screenshot Out value is nearly the same but In is almost twice lower then real.

What is the correct SNMP OID to get the same counters as in x670 cli?

 

3f28807f6ce74afaa44de4d511a0a980_90e59b4f-129e-42bd-88de-c4de78e6c8b1.png
3f28807f6ce74afaa44de4d511a0a980_2cf14fd6-c4ea-4cde-bc98-64c3a52eb4e4.png

 

3f28807f6ce74afaa44de4d511a0a980_9a195365-364f-4178-8d30-44dc7498ef31.png

 

6 REPLIES 6

amindomao
New Contributor

Excellent idea, thanks. I’ll try.

jeronimo
Contributor III

I still think you should take it step by step if you want to find out what is going on.

Since the source system shows the absolute packet count (two lines about the average one that you circled in red) you should compare those values to those retrieved from the switch using CLI and/or SNMP.

Only then should you add Cacti to the equation.

amindomao
New Contributor

Ok. Now I’ve tried to suppress an impact of “averaging” and ran vnstat for 6 minutes. “tx” on the vnstat screenshot not even close to corresponding “rx” on the cacti screenshot.

b538ac1c8efd4a1884f70a2de2b7aaab_a8cc0fae-61fd-4397-89d2-95fc2fec3108.png
b538ac1c8efd4a1884f70a2de2b7aaab_9c388bf8-c09f-40e6-a96e-9ac88666776b.png

 

jeronimo
Contributor III

If everything in Cacti is configured correctly (i.e. there are no CDEFs mangling your data and it looks up the correct OIDs), then you might simply be seeing an artifact of averaging.

When you're running the CLI “show utilization” command in CLI, the refresh is done every 5 seconds I believe. So the average it shows there is for the latest 5 seconds.

Cacti however polls every 5 minutes IIRC. So what you see in the Cacti plots is a +/- 5 minute average.

Try polling the SNMP values manually once every 5 minutes and take the average manually. Do the same for CLI output, in case it allows to retrieve absolute values somehow, certainly it does. Compare that to Cacti and to the CLI output.

GTM-P2G8KFN