Nagios Support Forum

Hi,

I use check_snmp_int.pl to graph the IN/OUT traffic of our network devices. The strange thing is that i get different behaviour depending on if i specify the SNMP community in the command or as an argument:
when i specify the Community then i get 2 performance graphs, one for IN and one for OUT.
$USER1$/check_snmp_int.pl -H $HOSTADDRESS$ -C public -2 -r -n $ARG1$ -fY -kBM -w $ARG2$ -c $ARG3$ when the community is an argument both IN and OUT are on the same graph.
$USER1$/check_snmp_int.pl -H $HOSTADDRESS$ -C $ARG1$ -2 -r -n $ARG2$ -fY -kBM -w $ARG3$ -c $ARG4$ So my first question is why such a behavior ? Then is there a way to specify the community in the command and having all on the same graph. This will allow us to easily modify the community for all the equipment in the futur.

Aurelien.

That is interesting. What is the exact text in the "Monitor service with this command..." under

Service Detail > (Your Service) > Configure > Re-Configure This Host ?

I think what the case may be is they are both doing the same thing, but NagiosXI is using two different methods to show them to you. As for the multiple items per graph, it doesn't look like thats possible with check_snmp_int as it requires an absolute host name.

Hi nscott,

The text appears like this:

For a service where the community is send as an argument:

• check_snmp_int!COMUNITY_NAME!"INTERFACE NAME"!50,50!70,70!!!!

For a service where the community is defined in the command:

• check_snmp_int_cisco!"INTERFACE NAME"!50,50!70,70!!!!!

Well, that explains why you're getting two different graphs, as far as why it defines the two differently I'm not really sure, but will be a good trick to know if stacking graphs is desirable.

Sorry but my english (or my french thinking ) make me not sure to understand properly what you are saying.

The two definitions are identical as:
check_snmp_int!COMUNITY_NAME!"INTERFACE NAME"!50,50!70,70!!!! where check_snmp_int=$USER1$/check_snmp_int.pl -H $HOSTADDRESS$ -C $ARG1$ -2 -r -n $ARG2$ -fY -kBM -w $ARG3$ -c $ARG4$ (This one gives stacked graphs)
and
check_snmp_int_cisco!"INTERFACE NAME"!50,50!70,70!!!!! where check_snmp_int_cisco=$USER1$/check_snmp_int.pl -H $HOSTADDRESS$ -C public -2 -r -n $ARG1$ -fY -kBM -w $ARG2$ -c $ARG3$ (This one gives two graphs)

So finally check_snmp_int.pl is executed with the same arguments no ?

The _cisco part is appeneded to the check, which I"m thinking cause nagios to setup the graph to read differently.

Hi Nicholas,

As the plugin is executed in the same way, shouldn't the behavior be the same ?

Aurélien.

Hi,

A new problem came up. If the check_snmp_int was working since i have installed the XI Server, now it seems to be unstable. I use it to monitore an Extreme Networks switch on 63 interfaces. And since few days, most of thoses services get stuck in UNKNOW status with two different messgaes:
No answer from host X.X.X.X
or
ERROR: Description table : No response from remote host 'X.X.X.X'.
If i try to execute the command from the shell it works normally. If i schedule the next check of the services, it also works once and get stucks again.

Aurelien.

Is this immediately after you updated to 1.4?

No reponse from remote host, that makes me think iptables is getting in the way or SELinux is getting in the way, can you check to see if SELinux got re-enabled? Is there a firewall between you and that switch that was recently changed?

And that is something interesting about the command line working, and then Nagios getting stuck after one. When you type stuff into the command line are you wrapping it in quotes?

Hi Nicholas,

The problem happened under 2011R1.3 and remains under 2011R1.4.

IPTables and SELinux are not enabled on the XI server and there is no firewall between the XI server and the switch.
To reassure you on this, as i said in my previous post, if the service get stuck in UNKNOW status, when I schedule the next check of it, it works for a while before getting UNKNOW status again, and from the shell SNMP request from the server to th switch works perfectly.
I also monitore environmental services on the switch by snmp, using check_snmp_environment.pl which works great.

Moreover, to valid that there wasn't any problem of the snmp process on the switch, i duplicate it into XI with few interfaces and no problem seems to be on its side: Aurelien.