Looking for feedback: I built an open-source Terraform/OpenTofu HTTP backend because global state locking felt too coarse
Hi guys, first post here!
I’m the author of KiloLock. I built it after running into the same large-state problem many infra teams eventually hit: the problem is not only state file size, but coordination around one shared state graph.
The stable path is intentionally boring:
- vanilla Terraform/OpenTofu HTTP backend compatibility
- PostgreSQL-backed state storage
- Docker Compose self-hosting
- no custom Terraform fork required
The experimental path is what motivated the project:
- queryable state graph
- resource-level history
- repair workflows
- foundations for narrower reservations / resource-aware locking
- future parallel-safe operations through kl
I’m not claiming this should replace your current backend if S3/GCS/HCP works fine. I’m looking for technical feedback from people who have dealt with large shared states, long plans, state lock contention, or awkward state splitting.
GitHub: https://github.com/kilolockio/kilolock
Documentation: https://kilolock.dev/documentation/