Closed viniarck closed 1 year ago
Initially assigning it to @Alopalao to assess and work on it, since I've hit this issue when reviewing https://github.com/kytos-ng/mef_eline/pull/371#pullrequestreview-1646864908, but cc'ing also @Ktmi and @italovalcy for you guys to collaborate on it since it has some overlap with handling link up and consistency check responsibilities that are being assessed/refined in other issues https://github.com/kytos-ng/mef_eline/issues/218#issuecomment-1718153743. Thanks, guys.
I'll go ahead and close this, since it's been discussed that mef_eline should keep reacting to link_up|link_down, due to https://github.com/kytos-ng/mef_eline/pull/383 now it's keeping track of the activation state so this problem has been minimized. If this discussion comes back a new requirement will be mapped accordingly.
I'm using this topology (ring INT, with an EVPL between s1-eth1 and s3-eth1):
It might be related to https://github.com/kytos-ng/mef_eline/issues/218, but since the scenario wasn't resulting in a traceback and it became easier to reproduce I went ahead and reported this one too.
How to reproduce:
mn --controller=remote,ip=127.0.0.1,port=6653 --switch=ovsk,protocols=OpenFlow13 --custom custom_topos/ring_int.py --topo mytopo
kytosd
(stop and start it again), and check the console how switches will reconnect, in most cases it's reproducing this issueWe'll need to reassess when it's safe to start handling a link up, and whether or not if consistency check itself will be responsible for the final activation and or reploy.