Candidate fix seems to work fine and addresses the first regression, however I still see problems in the second (upgrade) test. Even when I try to upgrade from Sodium SR1, and even when I run the "only" test, controller fails to boot up:
https://logs.opendaylight.org/releng/vex-yul-odl-jenkins-1/controller-csit-1node-akka1-all-sodium/478/odl_1/odl1_karaf.log.gz

Any idea why it is failing?

Is it possible to grab the Carbon snapshot somewhere so I can try it out locally?

as reported here I saw an OOM/GC crash when doing a simple
netconf test to mount 10 devices. Each device has a relatively large yang schema. The test was done using the revert patch
(87277) but now re-reading the comments here I realize it probably was not intended to fix the 2nd issue described (OOM).

tcere I tried the CSIT test with several distributions, in theory the test should be using the previous ODL release (Sodium SR1) but I figured it is broken and uses old Carbon instead. Anyway, this is the ink if you want to download ODL distros: https://nexus.opendaylight.org/content/repositories/opendaylight.release/org/opendaylight/integration/karaf/

That seems probably unrelated to this problem so lets:

1) Wait for next RC CSIT results to confirm there is no issues with the revert.
2) Dig more into your issue, like is it present in Sodium SR1 or master?

for the OOM/GC crash I am seeing with my one-off netconf mount test, I also see it in SR1. so, although likely a very
serious bug, it's not a regression. I have filed a new jira for this issue and we can discuss there if we would
want to block SR2 on it or let it go since it's not exactly a regression. Not exactly, because it was not there in the
original/first Sodium release.