Background

We currently have a multi-node config in test/format that closely mirrors the production setup, with a leader and follower running in parallel. This environment allows us to exercise important parts of the WT codebase, including:

• Eviction during ingest
• Checkpoint processing
• Leader/follower correctness validation within the same test

The setup uses predictable replay, which ensures that both leader and follower apply the same sequence of logical operations.

Problem Statement

Predictable replay does not currently support modify ops.

As a result:

• modify operations are not tested in the multi-node configuration.
• Coverage gaps exist between single-node and multi-node test env.

This creates a risk that issues related to modify operations in a replicated-like setup may go undetected.

Expected Outcome

modify operations are fully supported under predictable replay, so that Leader and follower correctness validation includes modify.

Out of scope

• Any WT bugs uncovered as a result of this change will not be addressed as part of this ticket{}