Comment 2 for bug 1804812

had a look here as homework from yesterday's ci escalation call (marios|ruck ci sprint 23).
This is definitely not failing consistently, for example current run on the master gate check is green for both scenario000-multinode-oooq-container-updates [1] and upgrades [2] so this supports Thomas assessment in comment #1 about a race condition.

Digging a little further I more suspect this patch [3] as responsible and in particular am wondering if the hacky sleep at [4] was helping us to avoid the race (though we might be able to come up with a better solution...)

[1] http://logs.openstack.org/45/560445/203/check/tripleo-ci-centos-7-scenario000-multinode-oooq-container-updates/fbe4a42/
[2] http://logs.openstack.org/45/560445/203/check/tripleo-ci-centos-7-scenario000-multinode-oooq-container-upgrades/f05d39d/
[3] https://review.openstack.org/#/q/I39b5a1a154b10683ac0de85afd6bbadc3491192a
[4] https://review.openstack.org/#/c/563000/2/tripleoclient/workflows/package_update.py@92