The initial release of sqi is now available to download! https://github.com/uberware/sqi/releases
This is alpha software. It is the heart of a render farm manager with almost zero frills or features. It has never been used by anyone in production.
What it can do
- Submit and run distributed render/compute jobs across multiple workers from a browser
- Deploy a working farm with a single binary on each machine; workers find the server automatically on the local network
- Track license pool usage to prevent over-subscription of concurrent software licenses
- Monitor jobs, workers, and task logs in real time via a web UI with live updates
- Submit jobs programmatically via Python API or raw OpenJD-format definitions
- Runs on Windows, Linux, macOS, and in Docker
What’s not there yet
- No product/preset system — job submission requires using raw OpenJD format JSON or YAML
- No S3 or cloud storage support — filesystem paths only
- No DCC submitter plugins (Maya, Houdini, Nuke, etc.)
- No authentication — any client on the network can submit and manage jobs
- No cloud worker support or auto-scaling hooks
Limitations
- SQLite only — not yet suitable for high-availability or high-throughput production deployments
- Path translation is resolved-mode only — no staged, environment, or command-arg modes
- Single scheduler instance — no failover if the server goes down
- Worker discovery is mDNS (local network) — cross-subnet and cloud workers require manual address configuration
- No role-based access control — all users have full farm access
Coming in v0.2
- Product and preset system with a graphical form editor and community preset library
- DCC submitter plugins starting with Maya and Houdini
- S3-compatible storage locations (AWS, Backblaze B2, R2, MinIO)
- Compute location configuration and job affinity routing
- Additional path translation modes for applications without native OpenJD support
While it’s not ready for everyone to start using every day, it’s far enough along to start to play with it. An open source render farm manager can only succeed if people use it, report issues, and request features and changes.
You can see more about the design and implementation plan in the ROADMAP.md document.
