Database backup

How to back up and restory Comentario database in a Kubernetes cluster

It’s usually a good idea to make periodic backups of Comentario database. Below are procedures that apply when you run Comentario in a Kubernetes cluster.


Making a backup

When using a local, file-based SQLite storage, you specify the database file location yourself. In this case you can simply make a copy of that file to another medium.

However, Comentario should not be running at the moment of making the copy. This downtime requirement makes the local storage option pretty inconvenient.

Restoring a backup

The same applies to restoring the backup:

  • Make sure Comentario is shut down.
  • Copy the file from the backup over to the required location.


Making a backup

To get a full database dump from a PostgreSQL database running in the cluster, issue the following command (assuming your PostgreSQL instance is named comentario-postgres and the NAMESPACE variable is set to your namespace name):

kubectl exec -t -n $NAMESPACE \
    $(kubectl get -n $NAMESPACE pods -l -o name) \
    -- pg_dump -U postgres -d comentario > /path/to/comentario.sql

Restoring a backup

To restore the database from a previously downloaded dump file (made above), you can use these commands (also assuming your PostgreSQL instance is named comentario-postgres).

We cannot send it via the pipe directly (I’m not sure why), so we copy it over first and clean up afterwards.

PG_POD=$(kubectl get -n $NAMESPACE pods -l -o 'jsonpath={}')
kubectl cp -n $NAMESPACE /path/to/comentario.sql $PG_POD:/tmp/c.sql
kubectl exec -t -n $NAMESPACE $PG_POD -- psql -U postgres -d comentario -f /tmp/c.sql
kubectl exec -t -n $NAMESPACE $PG_POD -- rm /tmp/c.sql