Hi,
I am working on setting up MC-LAG on two EX4600 towards EX4300 as "client".
But due to existing topology, the layout is as follows:
--trunk-to-other-EX-device-RSTP-->[EX4600]<-->[EX4600]<--trunk-to-other-EX-device-RSTP--
The EX4600 is connected with 1 RSTP trunk, and 1 ICCP/ICL interface.
The EX4300 is connected (will be) to both EX4600.
The links "other-EX" is a part of the existing RSTP-domain.
To maintain the availability with the loop and traffic, we need, for some time, run both RSTP and MC-LAG.
We were recommended (based from Junipers homepage about MC-LAG) to remove all mcae-VLANs from the RSTP trunk and add them to the non-RSTP-ICL/ICCP link.
At this point, the mcae was not up.
We noticed that some hosts, connected directly to the EX4600 couldnt ping/connect to devices in other parts of the network. I guess ARP-issues.
What I was thinking as a possible cause within this scenario is maybe, maybe.... we should have "all VLANs" allowed on the RSTP-trunk (except ICCP) and also have the same VLANs allowed on the ICL-trunk, to get the traffic to flow between the EX4600 as before, since i think that is a major part of the issue we've seen.
Any inputs or ideas on this scenario is welcome.
I have not seen any good documentation from Juniper on this kind of scenario, but are still searching :)
//Rob
------------------------------
Rob
------------------------------