PCM+ vs. Trunks

[André Beck]

Hi,

I'm having problems with PNM+ (or PCM+ or whatever it's acronymed today) identifying Trunks (static ones, no LACP or FEC) as such. In my case there are (besides a bunch of other devices):

- A 4108gl and a 4104gl
- Two 3400cl-24G

Both of the 3400cl have a Trunk (consisting of two 1000baseT links) to the 4108gl, one 3400cl has an additional Trunk (two 1000baseSX) to the 4104gl. What I'm observing in the Network Map:

- One of the 3400cl-4108gl trunks is not showing up at all. The fact that this is an RSTP blocked link is probably not a coincidence. I still expect it to show up, though as a dotted line as usual.
- The other 3400cl-4108gl trunk shows up in the correct color for a trunk, but the balloon help description is incorrect. It states the trunk goes "from Port: Trk3[C2,D2] (2000Mbps)" on the 4108gl "to Port: 19 (1000Mbps)" on the 3400cl.
- The 3400cl-4104gl trunk shows the same misbehavior, it is colored as a trunk but port assignment on the 3400cl side is wrong.

I would consider the latter two as a display glitch if it wouldn't break the VLAN discovery and network mapping. It appears that on every such trunk, the VLAN discovery only sees a connection for the VID 1 (DEFAULT_VLAN). In all other VLANs (there are seven more tagged ones on all the ISLs including the trunks), the connections are broken, the 4108gl (which is connected by nothing but the trunks in question) even is left as an unmapped device.

The 4100gl as well as the 3400cl are at the latest firmware and freshly rebooted. They were also deleted from PNM+ and newly discovered just to be sure.

PNM+ is 1.60 with patch A.

TIA,
Andre.

[Gijs Hoek]

Hello André,
Similar here: trunk not visible between two 5308xl systems. Resolved after manual rediscovering in PCM (so: not through nnm) and so forcing the ip address to default vlan. Regards, Gi

[Hector Manzo]

Andre,

I've been doing some testing with similar equipment on PCM 2.0. I am not done with my testing but so far PCM has correctly discovered the trunk, added the correct color to the link between the 3400 and 4100 plus correctly represented the blocked port with the dashed lines.

However when you mouse over the trunk link between the 3400 and 4100 the speed is not correctly represented. This is a known issue and the PCM lab is working on a fix for 2.0.

-Hector

[André Beck]

Hector,

> I've been doing some testing with similar
> equipment on PCM 2.0. I am not done with my
> testing but so far PCM has correctly
> discovered the trunk, added the correct color
> to the link between the 3400 and 4100 plus
> correctly represented the blocked port with
> the dashed lines.

Huh, PCM2.0 isn't yet available I thought? Anyway, good to know it will be fixed by then. I'm also having numerous other problems with 1.60 (also read here from others, especially trafficd going weird after some runtime), so an upgrade is planned.

> However when you mouse over the trunk link > between the 3400 and 4100 the speed is not > correctly represented. This is a known
> issue and the PCM lab is working on a fix
> for 2.0.

This is actually an SNMP issue, MIB2 cannot represent the interface speed with just 32 bits. I'm unsure though how it comes out at 14xx Mbps as it does in 1.60, maybe an additional signedness issue. I ignored this essentially as it indeed is just a display glitch and hearing they are going to work around it is great.

Thanks,
Andre.