Restoring health of a Managed Services for PostgreSQL cluster after its disks become Read Only

Issue descriptionIssue description

The storage in your Managed Services for PostgreSQL cluster has become read-only.
New data is not written to your cluster's database tables.

Troubleshooting and reproducing issuesTroubleshooting and reproducing issues

Go to the Monitoring tab in the cluster management interface, open the Disk capacity on primary chart, and look up the ratio between the free and used bytes values.
This way, you can evaluate the percentage ratio of the cluster storage utilization.

When the storage is more than 97% full, the host automatically switches to read-only mode.
For all databases in this case, DEFAULT_TRANSACTION_READ_ONLY is set to TRUE using the ALTER DATABASE query.

INSERT, DELETE, or UPDATE queries result in an error in this mode.

SolutionSolution

You can avoid this by setting an alert on switching over to read-only mode. You can find the steps to set up such an alert here.

When you receive the alert, you can use two methods to disable this mode in your cluster:

• By expanding the storage size: in this case, the service will automatically disable the read-only mode.
• Manually, by following this guide.

Currently, there is no automatic method to expand cluster storage.