// the find
koudelka/honeydew
Job Queue for Elixir. Clustered or Local. Straight BEAM. Optional Ecto. 💪🍈
Honeydew is a job queue and worker pool for Elixir with pluggable backends (Mnesia, Ecto, or pure in-memory) and no required external dependency like Redis or RabbitMQ. It's aimed at Elixir/Phoenix teams who want at-least-once background job execution without standing up another service.
The Ecto queue backend is a genuinely distinctive idea — turning a schema into its own queue means the job and the data that triggered it can't drift out of sync, since there's no separate system to fall behind. Job isolation via one-time-use processes plus a documented failure-mode system (abandon/retry/move) gives you real control over crash semantics instead of a black box. Everything (queues, dispatchers, failure/success modes) is pluggable via behaviours, so you're not locked into the defaults if Mnesia doesn't fit your deployment.
Last push was January 2024 and CI is still on Travis, which stopped free service years ago — the badge is almost certainly dead and it's a signal this isn't actively maintained. The default Mnesia queue requires manual schema file management on first deploy (has to run before Mnesia starts), which is an easy footgun for anyone used to zero-config queues. No web UI or dashboard for inspecting queue state — `Honeydew.status/1` and `filter/2` exist but you're working through IEx or building your own tooling. Clustering support exists but the docs push you toward the examples folder rather than a proper guide, so getting global queues right in production takes real trial and error.