0cec56d80b
This patch fixes two large problems with our teardown logic: (a) the tags applied to the teardown plays combined with the way we were calling ansible in quickstart.sh meant that we never executed the teardown tasks with the default `OPT_TEARDOWN=nodes` behavior. (b) We were calling all of the teardown tasks at the same point, which meant that the environment and node teardown tasks were being called with `ansible_user=root`. This meant that the node teardown in particular was a no-op, because the libvirt environment of the stack user was invisible to the tasks running as root. With this change in place, I can successfully run quickstart.sh several times in a row without error (previously this would fail because the teardown tasks wouldn't execute, causing conflicts) Change-Id: Ic94241c1d87b2dae335c6cb6627a4fdb5419482e Closes-bug: 1610677 Closes-bug: 1610537 |
||
---|---|---|
.. | ||
build-images-and-quickstart.yml | ||
build-images.yml | ||
dlrn-gate.yml | ||
gate-roles.yml | ||
libvirt-setup.yml | ||
libvirt-teardown.yml | ||
noop.yml | ||
provision.yml | ||
quickstart.yml | ||
teardown-environment.yml | ||
teardown-nodes.yml | ||
teardown-provision.yml | ||
teardown.yml | ||
tripleo-roles.yml |