Remove or recover a cluster
This is an internal article. Do not use it in production!
Contact the Canonical Data Platform team if you are interested in this topic.
This guide will cover how to manage clusters using an example PostgreSQL deployment with two servers: one in Rome and one in Lisbon.
Prerequisites
juju v.3.4.2+- A cross-regional async replication setup
- Refer to the page How to set up clusters
Summary
Switchover
If the primary cluster fails or is removed, it is necessary to appoint a new cluster as primary.
To switchover and promote lisbon to primary, one would run the command:
juju run -m lisbon db2/leader promote-cluster force-promotion=true
Detach a cluster
Clusters in an async replica set can be detached. The detached cluster can then be either removed or reused.
Assuming lisbon is the current primary, one would run the following command to detach rome:
juju remove-relation -m lisbon async-primary db2:async-replica
The command above will move the rome cluster into a detached state (blocked) keeping all the data in place.
Reuse a detached cluster
The following command creates a new cluster in the replica set from the detached rome cluster, keeping its existing data in use:
juju run -m rome db1/leader promote-cluster
Remove a detached cluster
The following command removes the detached rome cluster and destroys its stored data with the optional --destroy-storage flag:
juju remove-application -m rome db1 --destroy-storage
Recover a cluster
If the integration between clusters was removed and one side went into a blocked state, integrate both clusters again and call the promote-cluster action to restore async replication - similar to the “Reuse a detached cluster” step above.
If the cluster group lost a member entirely (e.g. rome is suddenly no longer available to the cluster group originally consisting of rome and lisbon), deploy a new postgresql application and set up async replication. The data will be copied automatically after the promote-cluster action is called, and the new cluster will join the cluster group.