Aruba & ProVision-based
cancel
Showing results for 
Search instead for 
Did you mean: 

procurve 2824 vs aruba/procurve 2920

 
Highlighted
Tom Lyczko
Super Advisor

procurve 2824 vs aruba/procurve 2920

I'm trying to migrate from a Procurve 2824 to an Aruba/Procurve 2920.

The Procurve when I do show lacp shows all the ports as passive or active or disabled in the lacp column.

I am trying to make the two switches as alike as possible so I can figure out why when the SAN cables and vSphere host pNIC cables are moved from the 2824 to the 2920, the vSphere hosts lose contact with the SAN storage.

2824 is this:

3 Passive 3 Up No Success
4 Passive 4 Up No Success
5 Passive 5 Up No Success
6 Passive 6 Up No Success

2920 showed nothing until I explicitly told the switch interfaces to be lacp passive, now it is this:

3 Passive 3 Down No Success
4 Passive 4 Down No Success
5 Passive 5 Down No Success
6 Passive 6 Down No Success

Nothing is now connected to the 2920 until I can get people not to fear things will break again.

The other two 2920s that replaced a different 2824 and a 2910 did not need this explicit lacp passive configuration, I need to understand the differences between the switches so I can effectively swap in the 2920 for the 2824.

The 2824 has a SAN VLAN 10 for the SAN ports and the vSphere storage vSwitch ports and so does the 2920.

Thank you, Tom

 

 

1 REPLY 1
Highlighted
parnassus
Honored Contributor

Re: procurve 2824 vs aruba/procurve 2920

Don't know if this is OT or not here but, focusing on the VMware vSphere side only and not on the SAN side, does your VMware vSphere host use NIC Teaming on its four pNICs connected to the actual HP ProCurve 2824 switch?

If so the NIC Teaming Load Balancing algorithm was probably left set to its default valute "Route based on the originating virtual port ID" and generally this is a good default setting *IF* a vSphere Standard Switch (VSS) is used *and* you don't want to touch the switch configuration. I said so beacuse this scenario doesn't require any specific setting switch side: switch connected ports should be left in their normal state (no LAG: no Static Trunking or - absolutely - no LACP=Dynamic Trunking).

There is also another scenario: in case your vSphere host is still using VSS (instead of the VSD vSphere Distributed Switch) the NIC Teaming Load Balancing algorithm could be set to "Route based on IP Hash" *but*, in this specific case, switch connected ports to vSphere host must be part of a LAG with Static Trunking (as know as "trunk" or "non protcol", no LACP=Dynamic Trunking at all).

VMware vSphere 5.1/5.5 versions don't support LACP (Dynamic Trunking) *when* VSS is used. In order to use LACP vSphere *should* use a VSD instead of a VSS.

So basically if I were you I would start to check your actual working configuration both on vSphere side (VSS/VSD and NIC Teaming settings) and on HP ProCurve 2824 side (maybe here with a show lacp or similar command just to start?).

If your vSphere uses vSphere Distributed Switch (VSD) the NIC Teaming Load Balancing algorithm will be able to support LACP and, switch side, you should set a LAG with LACP setting.