Hi all,
I couldn't find a good solution anywhere so here I come to see if someone faced that kind of need/problem.
So we are migrating an existing IOS LAN who run RSTP to a new NEXUS LAN running MST and fabricpath beetwen sites.
During migration (wich will last for arround one year) we need :
- L2 extension of VLAN from the "RSTP IOS LAN" to the "MST/Fabricpath NEXUS LAN",
- and accross 2 links for redundancy and eventually load-sharing.
Destination NEXUS LAN is made of 2 sites, with two 5548 in vpc on each site, and each all four 5548 interconnected in fabricpath, and one STP domain on each site.
We tried to connect the RSTP IOS LAN via two links : one link towards each site STP domain.
(MST 2x5548)==={fabricpath}===(MST 2x5548)
\Bound(PVST) /Bound(PVST) *PVST_Inc
\ /
\ /
[ **** IOS RSTP LAN **** ]
Doing that, trunk ports on the Nexus 5548 mount in Bound(PVST), but on one site the Nexus trunk port fall in Bound(PVST) *PVST_Inc.
This almost work but gives 3 problems :
- we cant do loadsharing of vlans on the 2 links
- in case of failure of the first link, the second link recover from *PVST_Inc and spanning tree in 40 to 50 seconds (quite long)
- and in this case of failure the topology of the LAN changes but some mac tables on Nexus side are note flushed so we loose some traffic for 300sec (mac timeout) (very long)
I think we have understand how everything work behind (MST0, each 5548 is root send its own BPDU, etc..),
and we begin to think there is no good solution on this way.
But with this need (connecting NEXUS MST to IOS RSTP with redundancy) what solution did you or would you implement ?
(and fabricpath is mandatory because we will have to connect a 3rd site with 2x5548)
Thanks all.
