This PR introduces changes that should reduce memory consumption (which could result in a leak) on both graphql-engine and Postgres for a mutation heavy workload.

Description

With mutations we end up with several 'single-use' prepared SQL statements, i.e, statements which cannot be reused for multiple mutations. The underlying reason for ending up with 'single-use' prepared statements is we embed values inside the statement instead of parameterizing them to reduce the complexity around handling relationships in returning clause. The problem with single-use prepared statements is that we end up leaking memory both on graphql-engine (we store a mapping of "SQL prepared statement" to "postgres's prepared statement id" in each PGConn) and on Postgres (no limits on number of prepared statements per connection).

Changelog

• CHANGELOG.md is updated with user-facing content relevant to this PR. If no changelog is required, then add the no-changelog-required label.

Affected components

• Server

Related Issues

#2012
#3388
#4077

Solution and Design

Prepared statements are not used for mutations any more. We might potentially end up with slightly slower but the benefit of predictable memory usage outweighs the potential degradation.

Steps to test and verify

@tirumaraiselvan has been running some benchmarks with a mutation heavy workload. This PR will be updated with the findings from those benchmarks.

Does this PR add a new Metadata feature?

• No

GraphQL

• No new GraphQL schema is generated

Breaking changes

• No Breaking changes