Issues

FIFO max_data_files_size drops far more user data than its budget requires: the trimming loop subtracts a flat per-file average instead of each file's real size

#15252

· hanke580 opened

on Sep 22, 2026

kTriggerCompaction is never re-registered, so time-based compaction set via SetOptions never fires on an idle DB

#15251

· hanke580 opened

on Sep 22, 2026

Stats history deltas use an unguarded uint64 subtraction, so Statistics::Reset() records counts near 2^64 (also persisted to disk)

#15250

· hanke580 opened

on Sep 22, 2026

`WAL_size_limit_MB` is enforced as a file count derived from the largest archived WAL, so the archive is pruned far below the limit — and wiped entirely at the boundary

#15249

· hanke580 opened

on Sep 22, 2026

One `SeekToFirst()` permanently disables `max_sequential_skip_in_iterations` for the rest of the iterator's life

#15248

· hanke580 opened

on Sep 22, 2026

MultiGet makes no progress when the first key's value is in a blob file and exceeds `value_size_soft_limit`

#15247

· hanke580 opened

on Sep 22, 2026

Why does the coroutine read executor default to one thread?

#15246

· ronag opened

on Sep 21, 2026

Add coroutine support for integrated BlobDB payload reads

#15245

· ronag opened

on Sep 21, 2026

Support native coroutine reads for iterators

#15244

· ronag opened

on Sep 21, 2026

Unused param compile error in utils/compression

#15243

· mkjt2 opened

on Sep 21, 2026

Coroutine read API: Let the caller keep its own executor

#15242

· ronag opened

on Sep 21, 2026

Feature request: officially supported Rust bindings and integration

#15217

· maskjelly opened

on Sep 16, 2026

Navigation Menu

All issues

FIFO max_data_files_size drops far more user data than its budget requires: the trimming loop subtracts a flat per-file average instead of each file's real size

kTriggerCompaction is never re-registered, so time-based compaction set via SetOptions never fires on an idle DB

Stats history deltas use an unguarded uint64 subtraction, so Statistics::Reset() records counts near 2^64 (also persisted to disk)

`WAL_size_limit_MB` is enforced as a file count derived from the largest archived WAL, so the archive is pruned far below the limit — and wiped entirely at the boundary

One `SeekToFirst()` permanently disables `max_sequential_skip_in_iterations` for the rest of the iterator's life

MultiGet makes no progress when the first key's value is in a blob file and exceeds `value_size_soft_limit`

Why does the coroutine read executor default to one thread?

Add coroutine support for integrated BlobDB payload reads

Support native coroutine reads for iterators

Unused param compile error in utils/compression

Coroutine read API: Let the caller keep its own executor

Feature request: officially supported Rust bindings and integration

Search results