LACP Cisco-Foundry

Hi, we have a Problem with LACP between a Cisco Catalyst 6509 an a Foundry FastIron Super-X.

The LACP is build of 2 Links. If wee unplug the first (primary) port of the LACP-Channel we will not see any traffic disruption. Also if we replug the port, no traffic disruption will happen. That's like expected.

If we unplug the second port, the traffic will not be disrupted in the first moment. But after roundabout 90 seconds the traffic will stop an will come back again after other 35 seconds. If we replug the link, the traffic will stop again for 35 seconds. Any Idea?

We can see that the LACP-Channel is build on both sides. So this is not a native Spanning Tree event, as it ca be expected by the 35 seconds downtime. Furthermore we are using Rapid Spanning Tree.

!!!!!!!Config Cisco: ! interface Port-channel57 switchport switchport trunk encapsulation dot1q switchport mode trunk switchport nonegotiate ! interface TenGigabitEthernet6/1 switchport switchport trunk encapsulation dot1q switchport mode trunk switchport nonegotiate channel-group 57 mode active ! interface TenGigabitEthernet6/1 switchport switchport trunk encapsulation dot1q switchport mode trunk switchport nonegotiate channel-group 57 mode active

!!!!!!!Config Foundry: ! interface ethernet 6/1 link-aggregate configure key 10057 link-aggregate active ! interface ethernet 6/2 link-aggregate configure key 10057 link-aggregate active !

Reply to
helmers
Loading thread data ...

Interesting. 90 seconds is 3 times Slow Hello timer (30sec). It looks like switches don't react on port going down and rely on LACP hello to detect the failure. IIRC as minimum on Cisco side STP event is quite expected - when you have only one port left in aggregate it will revert to regular port and go through normal STP transition.

debug lacp, debug spanning-tree and equivalents on Foundry side should help to figure out what is going on.

Regards, Andrey.

Reply to
Andrey Tarasov

This was mentioned here a while back. I don't believe it!

If true would make the use of dual channel etherchannel for redundant link less than useful. I have used such links and never observed this behaviour. It would be astonishing to me if it had been present and in all of the cases I had worked on had never been observed. There seems no need for this behaviour and has obvious disadvantages, why would cisco put it in?

Unfortunately I cannot do test at present. Maybe behaviour is platfrom dependent which would explain the differing observations.

Ah!, maybe LACP is different from PAgP? I have not used LACP much.

Reply to
bod43

You don't have to believe. It's quite possible to touch and see it.

More important question is was PAgP or LACP used on those links? "mode on" doesn't use either one.

Regards, Andrey.

Reply to
Andrey Tarasov

Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.