Almost every web app ends up needing to do some work outside the request. Send the sign-up email. Resize the photo. Remind someone about an appointment. Check the website overnight. The standard answer is a job queue backed by Redis or RabbitMQ, which means another service to install, secure, monitor and keep alive, often bigger than the app it serves.
Ukue, a micro queue, puts the whole thing in one SQLite file. Queues, jobs, retries and dead letters all live in that file. You can embed Ukue in your app, or run it as one small binary that any language can talk to. Copy the file and you have a backup.
The Pitch in Four Numbers
- 1 file holds every queue, job and retry.
- 0 jobs lost out of 8,000 while worker processes were killed 250 times.
- 900 jobs a second run by one worker, with every step written safely to disk.
- Apache 2.0, free, with the test logs on GitHub.
How It Works
Ukue does what a job queue should: retries with waits, delayed jobs that run later, and dead letters for jobs that keep failing so they don’t block everything else. The difference is where it keeps it all, in one SQLite file, flushed to disk on every change. The speed figure was measured on a two-core cloud machine with that flushing on, which is the honest way to measure a queue.
There are two ways in. Embed it in a Go program and enqueue a job inside the same SQLite transaction as the rest of your data, so the job only exists if the change was saved. Or run the binary and talk to it over HTTP from Python, Ruby, PHP or anything else. A terminal interface shows what’s going on, with commands like ukue stats and ukue work.
The use cases on the site are deliberately ordinary: a Flask app sending sign-up emails, a Go shop with SQLite, a Raspberry Pi logging sensor data over a patchy connection, GPU transcription workers, appointment reminders, nightly website checks and photo thumbnails.
Who It’s For
- Small teams and solo developers who want background jobs without running Redis or RabbitMQ.
- Apps already on SQLite, which can enqueue jobs in the same transaction as their data.
- Edge and device projects, like a Raspberry Pi, where one more service is one too many.
- Polyglot teams whose workers are written in different languages.
Limits
It’s an early release. It’s built for one machine and one file, so it isn’t the answer for a fleet of servers sharing a queue across the network. Very high throughput across many machines is what the big brokers are for. For the large number of apps that don’t need that, it’s a lot less to look after.
The Four-Question Screening
Every project featured here gets the same four questions, the way you’d screen a startup idea.
1. Where is the money in it?
Ukue is free under Apache 2.0, with no paid tier. Job queues are infrastructure everyone needs and few want to run, which is why managed queue services do well. The paid paths are a hosted Ukue with a dashboard, alerting and backups for teams that want zero operations, a commercial support plan for companies that embed it in their own products, and a multi-machine edition for when an app outgrows one server. The “SQLite for jobs” idea has a big audience of small apps.
2. What real problem does it solve?
Running a whole message broker just to send emails in the background. For most small apps that’s more moving parts than the app itself, and each part is something that can break at night. Ukue gives them durable background jobs with nothing extra to run.
3. The technology in plain words
Think of a to-do list in a notebook that never gets lost. Whoever’s free picks the next task, ticks it off when done, retries it later if it didn’t work, and moves it to a “problems” page if it keeps failing. If they collapse halfway through, the task is still in the notebook for the next person. Ukue is that notebook, as one file.
4. Moat and differentiation
Redis- and RabbitMQ-backed queues are the default, and they’re excellent at scale. Ukue’s edge is having nothing to run: one file, one binary, durable on disk, with a crash test that lost nothing across 250 killed workers. For the many apps that live on one server, that’s the better trade. The moat that can grow is reputation for reliability and the ease of being the first queue a developer reaches for.
Why It Matters
Most apps never reach the scale that justifies a message broker, but they still need background jobs that don’t get lost. A queue that’s just a file makes the reliable choice the simple one too.
Follow the Project
Read what a job queue is, start with the quick start, get it from the download page, or see the code in the Ukue repository on GitHub. Ukue is listed with the other open-source projects on SoftDevStudio.com, and more featured projects are in the Projects section.