Comment 16 for bug 1684272

Hmm, last known good is no longer good. Better, but not good.

2cbd4593261ba90ed61d2d293e69a930ef0b59b7 ControllerServiceChain 189.0

Looking at a pre-regression job, that's the commit we had in delorean at the time: openstack-heat-common-8.0.0-0.20170405074820.2cbd459.el7.centos.noarch

So something outside of heat has also impacted this regression. Based on Zane's findings in https://bugs.launchpad.net/tripleo/+bug/1666072/comments/31 maybe oslo.messaging? I don't see anything particularly suspicious from around that time though. :-/

I don't see a significant amount of swapping on the undercloud during the regressed jobs. Our resident memory usage seems to top out around 6 GB on 8 GB VMs. There is some swap being used, but it's minimal -- ~100 MB.