b82eb1b393
In the spirit of Kolla-Ansible, we generally try to provide workable defaults. The default for Elasticsearch curator schedule was fine except for multinode deploys where it would cause all nodes to run at the same time producing broken runs (race condition in the get-delete cycle). It is easy to improve this situation by embracing poor-man's reimplementation of keystone's fernet key rotation schedule. ES Curator does not need all the complexity of the former so it can be handled very well by shifting by as many hours as the instance's index dictates. It should rarely if ever need more time (most likely still in minutes range rather than hours). Change-Id: I9d6758c8550308d13d936de1a14afbe4124e593b |
||
|---|---|---|
| .. | ||
| main.yml | ||