Oleksii Kliukin 176c6e8b19 Avoid passing the role into the recreatePod. ...

Conceptually, the operator's task is just to change the pod. As it
has no influence over the role the pod will take (either the master
or a replica), it shouldn't wait for the specific role.

This fixes at least one issue, where the pod running in a single-pod
cluster has been waited for forever by the operator expecting it to
have a wrong role (since Patroni callback assiging it the original
replica role has been killed after a quick promote by the next
callback.)

2017-05-12 11:41:32 +02:00

..

cluster

Avoid passing the role into the recreatePod.

2017-05-12 11:41:32 +02:00

controller

Check etcd key availability for the new cluster

2017-05-12 11:41:31 +02:00

spec

Get config from environment variables;

2017-05-12 11:41:29 +02:00

util

go fmt run

2017-05-12 11:41:31 +02:00