jiengup opened a new pull request, #4202: URL: https://github.com/apache/iggy/pull/4202
## Which issue does this PR address? Closes #4142 ## Rationale Group commit used the padded full prepare size even when segment storage wrote only a small reference record to the WAL. This prevented large segment-backed messages from coalescing despite their small WAL footprint. ## What changed? Append batching now tracks retained capacity, actual padded WAL extent, and segment-body work separately. The WAL extent controls group-commit coalescing, while an independent segment-body limit preserves the safeguard requested in the #4092 review discussion. The previously ignored eight-message test now completes in one barrier group. A nine-message regression test verifies that batches still split when segment-body work exceeds its limit. ## Local Execution - Passed `cargo fmt --all` - Passed `cargo sort --no-format --workspace` - Passed `cargo clippy -p journal -p partitions -p simulator --all-features --all-targets -- -D warnings` - Passed `cargo test -p journal -p partitions -p simulator` - Passed `./scripts/ci/taplo.sh --check` - Pre-commit hooks ran and passed ## AI Usage 1. Tool: OpenAI Codex 2. Scope: Repository analysis, implementation assistance, test updates, and review of the final diff. 3. Verification: Ran the relevant journal, partitions, and simulator tests, clippy, formatting, Taplo, and pre-commit checks. 4. I reviewed and can explain every line of the change. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
