Postgres fails constantly every 1-2 days for 2-3 min

@tjcberman Due to an Etcd bug pertaining to auth token renewals, some PG clusters were getting thrown into a state where Stolon was no longer able to authenticate. Unfortunately, your PG was one of them and we had to rebuild the VM’s yesterday to get them back in order.

The underlying bug is still being worked out here:

In the meantime, we transitioned our Etcd to leverage stateless auth tokens “JWT”, which allows us to workaround the issue.

Sorry for the trouble!