Skip to content

Roadmap

Built

  • Job queue core — Postgres-backed queue with FOR UPDATE SKIP LOCKED claims, priorities, retry with backoff, dead-lettering, idempotency keys.
  • Executions & leases — per-attempt records with lease expiry; a watchdog reclaims work lost to crashes and returns it to the queue.
  • Cancel semantics — immediate for PENDING, deferred-to-attempt-end for RUNNING.
  • Schedules — 6-field cron + timezone, pause/resume, soft delete; a scheduler tick materializes due schedules into jobs.
  • Auth & audit — JWT with role claims, refresh-token revocation, append-only audit trail (OPERATOR/ADMIN read).
  • Operations UI — React SPA: live dashboard, job list/detail with cancel/retry, schedule management with a create form, role-gated audit views.
  • Cron UX — a schedule builder that turns dropdown repeat patterns into generated cron (restricted to what the scheduler can honor) and live plain-English translations of any expression, both directions visible at once.
  • Admin user management — ADMIN-only create/roles/lock/disable with session revocation on lock, self-modification guard, and full audit; no self-service registration by design.
  • Demo personas — one-click OPERATOR and restricted USER logins so visitors experience the role gating firsthand; private deployments opt out via bootstrap vars plus an account lock (security model).
  • Deployment — Dockerized services, single-droplet compose stack behind Caddy (automatic HTTPS), GitHub Actions CD to control-plane.dev on every merge.
  • Local demo — zero-config compose + seed script (deploy/compose.demo.yml, scripts/seed-demo.sh).

Next

  • Standalone worker — split the executor out of the API JVM into a separately scalable deployable. The claim path was written for this: no schema changes required (see ADR 0001).
  • Job timeouts per type — per-handler execution budgets distinct from lease duration.
  • Metrics dashboard — surface the Prometheus endpoint via Grafana; queue depth, claim latency, handler duration percentiles.
  • Webhooks — notify external systems on terminal job states.
  • Multi-tenancy hardening — per-tenant rate limits and quotas.