Problem with SNMP queries from cisco devices

[mboeker]

Hi all,

Just started using Cacti, although the environment I'm working with has had it in place for a while. Here's the situation:

We're monitoring about 100 servers and 20 network devices using SNMP, most of the servers are Linux (Centos or Ubuntu) and most of the network devices are Cisco
We're have Cacti host templates for the Linux servers and the cisco devices
- All of the templates have interface traffic (bits/sec), interface discards, interface errors
- The Linux templates have server-specific graph templates like logged in users, disk space, etc
- The cisco templates have cisco-specific graph templates like 5sec cpu usage, 1min cpu usage, etc

The graphs for the servers all seem to work, however the graphs for the cisco devices do not. They're all filled in gray and the stats underneath each graph say that all values are NaN (not a number). Looking at the Cacti log I see that spine is having problems doing snmp queries:

Code: Select all

04/04/2014 08:04:10 PM - SPINE: Poller[0] Host[757] TH[6] DS[72823] WARNING: SNMP timeout detected [60000 ms], ignoring host '192.168.32.1'
04/04/2014 08:04:10 PM - SPINE: Poller[0] Host[757] TH[6] DS[72823] SNMP: v2: 192.168.32.1, dsname: traffic_in, oid: .1.3.6.1.2.1.2.2.1.10.558, value: U
04/04/2014 08:04:10 PM - SPINE: Poller[0] Host[757] TH[6] DS[72823] WARNING: SNMP timeout detected [60000 ms], ignoring host '192.168.32.1'
04/04/2014 08:04:10 PM - SPINE: Poller[0] Host[757] TH[6] DS[72823] SNMP: v2: 192.168.32.1, dsname: traffic_out, oid: .1.3.6.1.2.1.2.2.1.16.558, value: U

However when I do an snmpwalk I get values:

Code: Select all

snmpwalk -v2c -c mycommunity 192.168.32.1 .1.3.6.1.2.1.2.2.1.16.558
IF-MIB::ifOutOctets.558 = Counter32: 409599439

I have double checked the SNMP community string that Cacti is using
I don't know why spine is getting "U" values instead of the actual data returned by snmp responses
It seems that the SNMP timeout that I've set in Cacti (60000ms) isn't working. My manual snmpwalk response for that device was slow, it took 3-4 seconds, but it certainly not the 60000ms specified in Cacti

I've been digging through log files and the forums and can't figure out why Cacti can't get that data...

Any ideas or suggestions? The exact same problem is happening for every Cisco ASA and Catalyst switch, but it's working on every CentOS and Ubuntu server...

Thanks!

[mboeker]

Anyone experiencing similar problems or have any suggestions?

Thanks again in advance for any input!

[bangolio]

Perhaps the port defined in Cacti is not 161?

[mboeker]

Hi Bangolio,

I've triple checked the config, it's definitely the right string and port. I seem to be making some progress though.. I found a cisco device at a different location that is collecting data and I'm comparing the two configs, so far I've noticed these oddities:

- The device that works doesn't have any interface graph templates in the device config, however when I go to Graphs > Location > Switches > Device, I see all of the graphs there
- The device that doesn't work has a bunch of interface graph templates, however when I go to Graphs > Location > Switches > Device
- The cisco device that works has one data source per interface traffic graph
- The cisco device that doesn't work has multiple data sources for many of the graphs
- When I go to the graph tree and select the switch under "Switches" there are no graphs, however when I select the switch under "Hosts" I see the graphs but most of them are gray and have NaN in the values, *except for this*:
- Somehow I did get a little bit of data from the interfaces while messing around with the config of the device that doesn't work, however data has stopped coming in again. Basically when I look at the last 6 hours, I see that for about about 30 minutes I got several data points, but those have stopped again
- In the device config > Data Sources > SNMP Interface Statistics, when I click on "Reload Data Query" I see tons of the following lines in the log:

Code: Select all

04/07/2014 02:04:47 PM - AUTOM8 TRACE: Poller[0] create_dq_graphs Host[757] - checking index: 428
04/07/2014 02:04:47 PM - AUTOM8: Poller[0] create_dq_graphs Host[757] Not Adding Graph - this graph already exists - DS[55815]
04/07/2014 02:04:47 PM - AUTOM8 TRACE: Poller[0] create_dq_graphs Host[757] - checking index: 429
04/07/2014 02:04:47 PM - AUTOM8: Poller[0] create_dq_graphs Host[757] Not Adding Graph - this graph already exists - DS[55814]