Introduction
Configuration
PostgreSQL Connection Pool
High Availability
KubeBlocks integrates the open-source Patroni solution to realize high availability and adopts Noop as the switch policy.
Install kbcli if you want to manage the PostreSQL cluster with kbcli.
Check whether the following role probe parameters exist to verify the role probe is enabled.
kubectl get cd postgresql -o yaml
>
probes:
roleProbe:
failureThreshold: 2
periodSeconds: 1
timeoutSeconds: 1
View the initial status of the PostgreSQL cluster and pods.
kubectl get cluster mycluster -n demo
kubectl -n demo get pod -L kubeblocks.io/role
Currently, mycluster-postgresql-0 is the primary pod and mycluster-postgresql-1 is the secondary pod.
Simulate a primary pod exception.
# Enter the primary pod
kubectl exec -it mycluster-postgresql-0 -n demo -- bash
# Delete the data directory of PostgreSQL to simulate an exception
root@mycluster-postgresql-0:/home/postgres# rm -fr /home/postgres/pgdata/pgroot/data
View logs to observe how the roles of pods switch when an exception occurs.
# View the primary pod logs
kubectl logs mycluster-postgresql-0 -n demo
In the logs, the leader lock is released from the primary pod and an HA switch occurs.
2024-05-17 02:41:23,523 INFO: Lock owner: mycluster-postgresql-0; I am mycluster-postgresql-0
2024-05-17 02:41:23,702 INFO: Leader key released
2024-05-17 02:41:23,904 INFO: released leader key voluntarily as data dir empty and currently leader
2024-05-17 02:41:23,905 INFO: Lock owner: mycluster-postgresql-1; I am mycluster-postgresql-0
2024-05-17 02:41:23,906 INFO: trying to bootstrap from leader 'mycluster-postgresql-1'
# View secondary pod logs
kubectl logs mycluster-postgresql-1 -n demo
In the logs, the original secondary pod has obtained the lock and become the leader.
2024-05-17 02:41:35,806 INFO: no action. I am (mycluster-postgresql-1), the leader with the lock
2024-05-17 02:41:45,804 INFO: no action. I am (mycluster-postgresql-1), the leader with the lock
Connect to the PostgreSQL cluster to view the replication information.
kubectl get secrets -n demo mycluster-conn-credential -o jsonpath='{.data.username}' | base64 -d
>
postgres
kubectl get secrets -n demo mycluster-conn-credential -o jsonpath='{.data.password}' | base64 -d
>
shgkz4z9
kubectl exec -ti -n demo mycluster-postgresql-1 -- bash
root@mycluster-postgresql-0:/home/postgres# psql -U postgres -W
Password: shgkz4z9
postgres=# select * from pg_stat_replication;
From the output, mycluster-postgresql-0 has been assigned as the secondary pod.
View the status of the PostgreSQL cluster and pods again.
kubectl get cluster mycluster -n demo
kubectl -n demo get pod -L kubeblocks.io/role
After the failover, mycluster-postgresql-0 becomes the secondary pod and mycluster-postgresql-1 becomes the primary pod.
View the initial status of the PostgreSQL cluster.
kbcli cluster describe mycluster -n demo
Currently, mycluster-postgresql-1 is the primary pod and mycluster-postgresql-0 is the secondary pod.
Simulate a primary pod exception.
# Enter the primary pod
kubectl exec -it mycluster-postgresql-1 -n demo -- bash
# Delete the data directory of PostgreSQL to simulate an exception
root@mycluster-postgresql-0:/home/postgres# rm -fr /home/postgres/pgdata/pgroot/data
View logs to observe how the roles of pods switch when an exception occurs.
# View the primary pod logs
kubectl logs mycluster-postgresql-1 -n demo
In the logs, the leader lock is released from the primary pod and an HA switch occurs.
2024-09-28 09:54:15,199 INFO: Lock owner: mycluster-postgresql-1; I am mycluster-postgresql-1
2024-09-28 09:54:15,419 INFO: Leader key released
2024-09-28 09:54:15,632 INFO: released leader key voluntarily as data dir empty and currently leader
2024-09-28 09:54:15,634 INFO: Lock owner: mycluster-postgresql-0; I am mycluster-postgresql-1
2024-09-28 09:54:15,635 INFO: trying to bootstrap from leader 'mycluster-postgresql-0'
# View secondary pod logs
kubectl logs mycluster-postgresql-0 -n demo
In the logs, the original secondary pod has obtained the lock and become the leader.
2024-09-28 09:54:17,117 INFO: no action. I am (mycluster-postgresql-0), the leader with the lock
2024-09-28 09:54:17,516 INFO: no action. I am (mycluster-postgresql-0), the leader with the lock
Connect to the PostgreSQL cluster to view the replication information.
kbcli cluster connect mycluster -n demo
postgres=# select * from pg_stat_replication;
From the output, mycluster-postgresql-1 has been assigned as the secondary pod.
Describe the cluster and check the instance role.
kbcli cluster describe mycluster -n demo
After the failover, mycluster-postgresql-1 becomes the secondary pod and mycluster-postgresql-0 becomes the primary pod.