01-17-2012 05:42 PM
I have a stack of four FCX 648s. The switches are hooked together in a ring topology using the CX4 interconnect. Now since it is a loop would you need to run a STP protocol like your Fast Port on the switch? I wonder as the active controller manages the whole stack and it uses a single MAC address so it would seem that you don't need to run a STP protocol.
So if I say no spanning on the switch am I in trouble? These switches are edge devices connecting ro compute nodes in a cluster.
Also it seems that setting hittless failover and setting the mac address for the stack to what the active controller is necessary for failback and also to stay away from a spanning tree issue. Is that correct?
Do you have a "best practice manual on these switches and how to manage them in a stack?
01-18-2012 12:20 AM
Please check the following doc here in the communities:
Hitless FCX stacking best practise.
If you use the dedicated stacking ports on a FCX-6xxS then a special stacking protcol is managing the stack and the "loop". It is not the common STP which you configure on the FCX.
Hope this helps,
01-18-2012 05:39 AM
Would you not use RSTP as this is faster, if it supports it. If they are in a ring / loop using the correct part you may not need to do anything, On the HP units I have fitted using their casscade module it sets it up for you, then if you happen to add another interconnect in to your stack it will change to blocking,