February 4, 2026
File not found? It’s the feature
Show HN: Bunqueue – Job queue for Bun using SQLite instead of Redis
Dev ditches Redis for a tiny file — cheers, fear, and “SQLite is web‑scale” energy
TLDR: Bunqueue is a new job tool that skips extra servers by using SQLite, a simple built‑in database, instead of Redis. The crowd is split between fans who love the no‑infrastructure setup and skeptics who question Bun‑only lock‑in, big speed claims, and how it handles growth and reliability
Bunqueue crashed the party with a spicy promise: forget extra servers — your background jobs can live in one little file. The creator, kernelvoid, says they built it because they were “tired of spinning up Redis,” and claims SQLite (a lightweight built‑in database) can push 100k+ jobs per second. Cue the split: Team Minimalism is clapping for one‑command setup and a familiar API, while Team Serious Systems is clutching their pagers about scaling and durability.
In the comments, a veteran from Apple, tombert, drops a mic: iTunes once ran job queues on Oracle and it “worked fine.” Translation: databases-as-queues aren’t new, just unfashionable. Fans call Bunqueue the “no-ops dream” and love that it’s BullMQ‑compatible (so it looks like popular tools), while skeptics side‑eye the “Bun only” requirement and grill the eye‑popping speed claims with classic benchmark snark: “500k/sec? On whose laptop?”
The memes flow: the eternal “SQLite is web‑scale” joke resurfaces, along with quips like “my queue fits in a tweet.” Critics warn about multi‑process file locking and container weirdness; supporters fire back that single‑server apps don’t need a warehouse forklift to move a tricycle. Verdict from the vibe check: this is a culture war between people who hate extra infrastructure and people who design for disaster, and both sides came ready with receipts
Key Points
- •Bunqueue is a Bun-only job queue library with a BullMQ-compatible API.
- •It uses SQLite for persistent storage, eliminating external dependencies like Redis or MongoDB.
- •Two modes are available: Embedded (in-process) and Server (standalone TCP/HTTP).
- •Performance claims include 100K+ to 500K+ jobs per second, with SQLite in WAL mode.
- •Features include priorities, delays, retries, cron scheduling, dead letter queues, real-time events, and Prometheus metrics.