Hi AWS Support,
During a Production RDS maintenance (minor version upgrade from PostgreSQL 9.4.20 to 9.4.21), the instance has become stuck in "Incompatible-parameters" status.
The affected RDS resource ID is db-DZAT6VNZZ3JSXFDU6BQXRQOCJQ
The errors in the event log are:
-This DB instance was placed into incompatible-parameters status because the huge pages configuration for the instance would exceed 90% of the available instance memory. To fix this issue, modify the associated parameter group to lower the shared memory usage, and then reboot the instance. For more information, please refer to the RDS Postgres documentation
-Postgres could not be started due to incompatible parameters. Shared memory segment exceeded available memory or swap space. Please modify shared memory usage parameters to lower values and reboot the instance.
The maintenance only included minor version upgrade. Huge pages parameter is already set to OFF and we did not alter any parameters for this maintenance therefore we are unsure why the upgrade failed. To try and resolve the issue, we tried reducing max_connections, restarting the instance, but none of them worked.
This is a production instance and we had to urgently restore it using a different name.
To try replicate the issue, we created a test instance from the production snapshot with the same settings (instance class, parameters, etc), and tried the same upgrade again (9.4.20 -> 9.4.21) and this time it was successful.
Can you please check the cause of the upgrade failure? If upgrade failed, why is it not rolling back automatically? Also can you please allow the instance to be deleted so that it stops incurring further cost?
Regards,
James