Early warning sytem for postgres' transaction ID wraparound

[priiiiit]

Based on this blog post from AWS, it implements a warning system to avoid transaction ID wraparound.

If this alarm is triggered, you should check your AUTOVACUUM progress. More info about this in AWS blog "Understanding autovacuum in Amazon RDS for PostgreSQL environments"

New resources:

• aws_cloudwatch_metric_alarm.maximum_used_transaction_ids_too_high

New inputs:

• var.engine - by default it is set to an empty string. If that is configured to be either postgres or aurora-postgresql (same values that are used for aws_db_instance), then this new alarm is created.
• var.maximum_used_transaction_ids_too_high_threshold - set to 1,000,000,000 (1 billion), as recommended by AWS blog post mentioned above.

New outputs:

• alarm_maximum_used_transaction_ids_too_high

All resources, inputs and outputs are following the same naming and style convention as existing code. Only real difference is the conditional creation in case it is a postgres engine.

Tested and already in daily use in prod. Backwards compatible, does not affect existing setups.

[priiiiit]

@lorenzoaiello, have you abandoned this module? If you don't have time to maintain, then I am happy to take over.

[lorenzoaiello]

looks good, thanks @ppihus