On 6/4/18 1:14 PM, AMG Zollner Robert wrote: > Yes, I was enslaving while the interface was up. > > Just tested some of the builds that where not working earlier and they > are working if I keep the interface down when enslaving as you suggested. > > Is this the expected behavior?
Not expected from my perspective. The VRF device cycles interfaces when they are enslaved or unenslaved to clean up route and neighbor tables. This is a day 1 property of VRF. I guessed that was the problem based on the commit you bisected the problem to. If nothing else, it gives you a workaround until it is fixed.
