Start a reclaim when get_clean_block has to park a waiter, and once at the end of BlockManager::init. Previously reclaim ran only after taking a clean block or finishing a write or reclaim. A device recovered with 0 clean blocks therefore parked every flusher forever and never wrote again. on_reclaim_finish already hands the reclaimed block to a waiting flusher, and reclaim_if_needed still respects clean_block_threshold and reclaim concurrency.
Add test_store_write_after_recovery_without_clean_block. It overwrites every block file with a block that holds data, reopens the store, enqueues an entry, and bounds wait() with a 10 s timeout. This needs tokio's time feature in dev-dependencies.
Validation:
- The new test fails with
Elapsedwithout the fix and passes with it. cargo test -p foyer-storage: 29 lib tests passed.cargo fmtclean.
Fixes #1373.
Activity
3
codecov[bot]18d agoCodecov Report
:white_check_mark: All modified and coverable lines are covered by tests.
| Files with missing lines | Coverage Δ | |
|---|---|---|
| foyer-storage/src/engine/block/engine.rs | 93.75% <100.00%> (+0.22%) | :arrow_up: |
| foyer-storage/src/engine/block/manager.rs | 98.20% <100.00%> (+0.01%) | :arrow_up: |
... and 2 files with indirect coverage changes
:rocket: New features to boost your workflow:
- :snowflake: Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
- :package: JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.
LGTM. Thank you, Anthony.
BTW, would you mind follow https://github.com/foyer-rs/foyer/issues/1373 to pass the DCO test. 🙏 Is not necessary, but recommended.