SNMP error in PCM

[Anders_35]

My SNMP is quite rusty. Can someone decipher this one for me, please?
I don't see what's wrong.
I get hundreds of these in PCM traffic manager, on various switches and ports.

SNMP error on segment 10.101.60.238_port_49. State = 12, PDU type = 162, SNMP error = 2, error index = 2

Any hints appreciated.
Cheers,
Anders :)

[Mathias Sundman]

Me too wonders about a similar error message:

SNMP error on segment 10.0.20.10_port_13. State = 78, PDU type = 162, SNMP error = 2, error index = 1.

I only get this on my newly installed 3400 switch. No other gives this error (2524, 4000, 2626, 2824).

I'm polling using SNMPv2 via an unrestricted SNMP community. The port it complains about is part of a 3 cable trunk to another switch. The switch itself doesn't log any errors, it's just PCM reporting this SNMP error.

Even if no one knows what causing this, does anybody know where I can find info about what the various SNMP error codes means?

[Les Ligetfalvy · Accepted solution]

If memory serves me right, trying to get ease or sflow stats for trunk members or mesh ports may throw that error. Instead of the individual ports, try targetting the virtual port.

[Mathias Sundman]

I figured that out right after posting, the problem now is just that when I go to "Traffic Devices", there is no row for port 13, as it's a member of a trunk, so why is PCM trying to poll this interface?

No, other ports were selected, so now I've selected Trk1 for polling instead. Hopefully this will cause it to stop polling port 13...

[Anders_35]

Les,

Thanks! That sounds about right. I won't be able to change things to test that right now, but it sounds veeeery much like you are on to something.

Cheers,
Anders :)

[Les Ligetfalvy]

Mathias,
You will have to deselect port 13. The PCM GUI is not an exact science... very much dependent on the version of JRE and how you navigate through the list. Also, results may vary by PCM version and whether you are at the server or a remote client session.

If all else fails, try standing on one foot and reboot the PCM server. YMMV

[Mathias Sundman]

The damn software doesn't seem to understand what I'm telling it ;-) I select Trk1 for Stats, and nothing more and hits OK. Everythings looks correct, I can even restart the PCM application (the client GUI) and it lists Trk1 for Stats as it should, then halv an hour later when I check again, it has removed Trk1 again is complaining about SMTP error for port 13!

Is there some config textfile or such that I can edit manually to remove port 13?

It seems silly having to remove port 13 from the trunk just so it gets visible in PCM again so I can remove it from "Sampler".

Thanks anyway.

[Hector Manzo]

Anders and Mathias,

I'd like to ask a few questions to see if these issues have already been fixed and if not then we would need dedicate engineering time.

What version of PCM are you using?
What model ProCurve device are you collecting sampled data from?
Is device 10.101.60.238 configured for trunking or meshing. If so then is logical port 49 the logical port for a mesh or trunk?

Thanks,

[Mathias Sundman]

Unfortunally, I ended up removing my trunk, going back to only using a single normal tagged port. There were a few things I found odd about the trunk / PCM.

First these SNMP error messages.

The 3400 doesn't send CDP packges (As I've now been informed by a HP tech is normal, as it's using LLDP instead), which means my 2524 doesn't see the 3400 as a CDP neighbour.

My 3400 sees the 2524 as a CDP neighbour on port 13, not Trk1. I should it should show up on Trk1 as is does on another 2524 I have trunked to a 4100.

In PCM+ it says when holding the mouse pointer over the link between the 3400 and 2514 on the topology map that the 3400 is connected via port 13 (100Mbps) and the 2524 via Trk1 (300Mbps). Shouldn't the 3400 report 300Mbps via Trk1 as well??

PCM+ version: B.02.005
Polled device: 3400 (M8.69)
Trunk to dev: 2524 (F4.08)
Port 13 on the 3400 is a member of a trunk (Trk1)

I've tried both using a normal trunk and a LACP trunk. Made no diffrence.

[Anders_35]

Les: It seems you were spot on. I've excluded the ports from traffic monitor today, and added the mesh and trunks instead. Have not seen errors in the log since. I will leave it until tomorrow, to make sure.
Funnily enough, in the "traffic devices" window, it lists errors for some of those deselected ports, in the "Last arrival column", as if it is still polling them for something.

Hector: I think it has been resolved. Our original version was a 1.6 evaluation that expired, and only the PCM, not PCM+ functionality was available. Thus, I could not manually configure traffic monitor. (Was TM even supposed to be running?)
I installed 2.0+ today, and it looks good so far.

As for which ports can be selected/deselected: In my PCM+, trunks are shown, and the individual ports not shown in the list. However for mesh ports, both the individual ports and the mesh port is shown.