8600 ERS communication with HP GBE2

[lpatella]

Anyone have any experience establishing communication between Passport 8600 ERS and HP proliant Class-C GBE2?

Having a few issues:
1. looping
2. When no loop: no communication.

Thanks

 

[DaddyOfThree]

I have about 26 of the HP GbE2s connected (SMLT) to two ERS 8600 switches.

You really didn't give any details whatsoever with respect to the configuration. You state that you believe you have a loop, and with the loop your unable to communicate. Any loop will eventually drive the 869xSF to 100% utilization (along with the switch fabric) and cause a complete network failure (no legitimate packets can make it across the network).

Let me take a guess and assume the following;
1. you have two HP GbE2 switches (A side and B side)
2. you have connected a link from GbE2-A to the ERS 8600
3. you have connected a link from GbE2-B to the ERS 8600

Your probably overlooking the fact that there is a trunk interface between GbE2-A and GbE2-B on ports 17 and 18.

Unless your going to be using the GbE2 as a Layer 3 switch you'll need to either use Spanning Tree to block one of the links or you should ONLY cable either ether the A switch or the B switch to the ERS 8600. If you cable both GbE2 switches you'll put a loop in the network (assuming there's no STP to save you). You could disable ports 17 and 18 on both GbE2s to disable the trunk between the GbE2-A and GbE2-B switch.

I currently only cable/uplink the GbE2-A switch to my ERS 8600 core. The GbE2-B switch is reachable via the trunk between GbE2-A and GbE2-B.

You can find additional information here;

http://michaelfmcnamara.blogspot.com/2007/11/hp-gbe2-switch-blade.html

Cheers!

 

[anthonyanderberg]

Based on my experience it sounds like DaddyOfThree is probably on the right track.

I'll just mention that HP also sells pass-thru modules that present the blade's ethernet interfaces as individual ports on the back of the chassis, and those ports can be patched into external switches however you like - some people mount switches in their racks, others have 'core' switches in adjacent racks. Of course having that many ports presents problems of its own, but physical complexity can be easier to manage than virtual complexity... its a design choice.