All issues

Hot-key vs distributed-key: moon's loss is a per-shard command-path deficit, not a sharding one

#799

· TinDang97 opened

on Sep 2, 2026

SPOP is ~1000x slower than Redis: it clones the entire set to pick one member

#798

· TinDang97 opened

on Sep 1, 2026

Profile: moon's per-op CPU excess is refcounting, parsing and the write gate — not the hash table (~39% on both engines)

#797

· TinDang97 opened

on Sep 1, 2026

List encoding: moon has no quicklist — a 600-element list falls back to linkedlist where Redis uses quicklist

#796

· TinDang97 opened

on Sep 1, 2026

ZADD CH miscounts sub-epsilon score changes: absolute f64::EPSILON compare where Redis compares exactly

#792

· TinDang97 opened

on Sep 1, 2026

perf_v0112_insert_or_update_single_probe fails on x86_64 under both runtimes (passes on aarch64)

#789

· TinDang97 opened

on Sep 1, 2026

used_memory under-reports container types up to 5.75x on Linux — --maxmemory cannot bind (quantifies #475)

#788

· TinDang97 opened

on Sep 1, 2026

ZADD and SADD never reach their listpack encodings — 20.5x / 2.01x memory vs Redis

#787

· TinDang97 opened

on Sep 1, 2026

docs/production-guide.md documents a cross-shard fast path that does not exist; the flag it recommends prevents startup

#776

· TinDang97 opened

on Aug 30, 2026

INFO total_commands_processed never advances on Linux (dead on the shipping runtime; macOS counts correctly)

#775

· TinDang97 opened

on Aug 28, 2026

perf(metrics): ungated unpadded global atomics on the per-command path; monoio still uses per-command recorders while tokio uses the batched ones

#774

· TinDang97 opened

on Aug 28, 2026

perf(shard): CoalescedReadBatch is dead code - 87.5% of cross-shard reads take the un-batched hop it was designed to fix

#773

· TinDang97 opened

on Aug 28, 2026