d646c75790
Add the ability to replace an overcloud node that's running the etcd service by updating the etcd cluster's membership to delete the old node (the one being replaced) with the new replacement node. etcd maintains its cluster membership using each node's name and it's cluster peer address on the internal_api network. This data typically changes when replacing an overcloud node. The cluster membership is updated using a shell script that's triggered to run after etcd is deployed on the new node. A new EtcdInitialClusterState parameter is introduced. It defaults to "new," and only needs to be set to "existing" when the cluster membership changes. NOTE(stable/victoria): The backport dealt with conflicts due to these changes that were introduced in wallaby: - https://review.opendev.org/c/openstack/tripleo-heat-templates/+/771832 - https://review.opendev.org/c/openstack/tripleo-heat-templates/+/820111 Change-Id: I8daa5500e7ee722eef64808306dc2aa5ca027a15 (cherry picked from commit |
||
---|---|---|
.. | ||
etcd-container-puppet.yaml |