Friday, December 21, 2018

Possible loop or just terrible design.

I'm in the initial design of this WAN and am seeing some unexpected behavior. 3 sites which all have fortigate firewalls. All in the same as and setup with ADVPN over iBGP. Now each site also has an eBGP over mpls. I then have the mpls neighbor weighted higher than the ADVPN peer. I also have the allow-same-as-in set on the eBGP peer since the provider couldn't do the as-override like initially requested. Also set to 1 instead of the default 3.

When testing, if I shut the MPLS port everything fails correctly but after a few hours to a day I start to run into problems. It's like the MPLS maybe flaps for a second causing the route to failover and the other sites seem to bounce too and become very unstable.

The routes seem to stop propagating over eBGP and the only way to get it back stable is to turn down the iBGP and reset eBGP peer and wait.



No comments:

Post a Comment