Cisco Network Help

[Undertaker225]

Hello everyone, I was wondering if there were any networking people that could help me solve an issue on an enterprise level, this isn't a critical issue but still it would be great to solve:

 

I have a whole bunch of 2950 Cisco switches unfortunately. They run on IOS v12.1 and they seem to occasionally have trouble connecting to our network management server via SNMP but are still able to be contacted through ICMP. What could be causing this other than the software, because it's only a handful of these switches with this same IOS version that are behaving like this.

 

Any help is appreciated.

[The_Denks]

Did you try turning it off and on again?

[Undertaker225]

2 minutes ago, UberGamerKing said:

Did you try turning it off and on again?

I really hope that is an IT crowd Reference because that is great. However, does not solve my problem.

[Lurick]

12.1? That's some old software right there. It's very well possible that it's a bug with SNMP, either with the tool itself or with that software version. Are you able to post a sanitized version of the SNMP configs?

[Undertaker225]

42 minutes ago, Lurick said:

12.1? That's some old software right there. It's very well possible that it's a bug with SNMP, either with the tool itself or with that software version. Are you able to post a sanitized version of the SNMP configs?

I'm afraid that I can't post any versions of the SNMP configs, sanitized or not, part of the job I'm afraid, but I'll look into getting some... details when I get a chance. Do you happen to know any possible solutions based on theory? We are looking at almost 250 switches here.

[Lurick]

5 minutes ago, Undertaker225 said:

I'm afraid that I can't post any versions of the SNMP configs, sanitized or not, part of the job I'm afraid, but I'll look into getting some... details when I get a chance. Do you happen to know any possible solutions based on theory? We are looking at almost 250 switches here.

That's fine, just searching for SNMP connectivity bugs brings up a lot of stuff, lol. So the more details the better

Have you tried pointing one switch to a second or different SNMP tool and seeing? Not sure if you have access to any extra switches running that code in a lab or something, but that could rule out one side of the equation.

[Undertaker225]

4 minutes ago, Lurick said:

That's fine, just searching for SNMP connectivity bugs brings up a lot of stuff, lol. So the more details the better

Have you tried pointing one switch to a second or different SNMP tool and seeing? Not sure if you have access to any extra switches running that code in a lab or something, but that could rule out one side of the equation.

Hooking up some old 2950's (as opposed to new ones?) and I am going to see what I can find out! I'll repost when I have some more data.

[Lurick]

1 minute ago, Undertaker225 said:

Hooking up some old 2950's (as opposed to new ones?) and I am going to see what I can find out! I'll repost when I have some more data.

Yah, if you have any 2950 running the same image as what you have in production and seeing if you can reproduce the issue would be awesome. Not always easy but if we can see if it happens using a different program then we know it's a switch issue. If not then it might be a timing issue on the program side where it's polling too fast.

[Undertaker225]

1 minute ago, Lurick said:

Yah, if you have any 2950 running the same image as what you have in production and seeing if you can reproduce the issue would be awesome. Not always easy but if we can see if it happens using a different program then we know it's a switch issue. If not then it might be a timing issue on the program side where it's polling too fast.

We also have an old version of Network management. We are using Cisco Prime v2.2.

[Lurick]

Just now, Undertaker225 said:

We also have an old version of Network management. We are using Cisco Prime v2.2.

Ah, interesting. So it could be a Prime issue, that helps. Let me see what I can dig up on that as well

[Lurick]

I did forget to ask earlier but is the error an Inventory Collection error?

 

If you're able to, check and see if you have the latest device packages installed:

https://supportforums.cisco.com/discussion/12388746/cisco-prime-21-collection-failure-2950-series-switch

 

 

[Undertaker225]

7 minutes ago, Lurick said:

I did forget to ask earlier but is the error an Inventory Collection error?

 

If you're able to, check and see if you have the latest device packages installed:

https://supportforums.cisco.com/discussion/12388746/cisco-prime-21-collection-failure-2950-series-switch

 

 

Lurick,

 

I am able to see the switches in my network management software. They respond to ICMP, or ping, but they are not, for some reason, being able to contact the PI in a sporadic manor. There is no rhyme or pattern to these errors which makes it all the more frustrating. All I get are email notifications letting me know, but no connectivity loss.

[Lurick]

1 minute ago, Undertaker225 said:

Lurick,

 

I am able to see the switches in my network management software. They respond to ICMP, or ping, but they are not, for some reason, being able to contact the PI in a sporadic manor. There is no rhyme or pattern to these errors which makes it all the more frustrating. All I get are email notifications letting me know, but no connectivity loss.

Hmmm, alright. So for some reason they stop responding to SNMP polling at random times, any log messages on the boxes themselves by chance?

[Undertaker225]

1 hour ago, Lurick said:

Hmmm, alright. So for some reason they stop responding to SNMP polling at random times, any log messages on the boxes themselves by chance?

I am receiving no log messaged from the switches.

[Undertaker225]

I am getting "Too Big errors" from the 2950's in question but not the newer switches. Could this be the case, and if so, do you know what the solution might be?

[Undertaker225]

So, from what I can tell, the amount of data is handled by the application that sends the probe, which is to my knowledge PI. I also noticed that SNMP logging is disabled on this test switch, you wouldn't happen to know that command off the top of your head would you?

[Lurick]

2 hours ago, Undertaker225 said:

So, from what I can tell, the amount of data is handled by the application that sends the probe, which is to my knowledge PI. I also noticed that SNMP logging is disabled on this test switch, you wouldn't happen to know that command off the top of your head would you?

Just so I make sure are you referring to the debug logging?

debug snmp [sessions | packets]

and then

term mon

 

To turn off term mon just do 'no term mon' instead.

[Undertaker225]

56 minutes ago, Lurick said:

Just so I make sure are you referring to the debug logging?

debug snmp [sessions | packets]

and then

term mon

 

To turn off term mon just do 'no term mon' instead.

under the show snmp command you will see something called "snmp logging" which is indicated as disabled.

[Lurick]

3 minutes ago, Undertaker225 said:

under the show snmp command you will see something called "snmp logging" which is indicated as disabled.

Ah! That's for sending syslog messages to a server. You can enable it with 'logging history [1-7]'

[Undertaker225]

Much appreciated, I'm afraid I am out for the day, but I will let you know if I come up with anything else. You have been a tremendous help!