2.0 Prerelease Quickstart
A first-time path for Laravel, Python, and operator users evaluating Durable Workflow 2.0 prerelease artifacts.
A first-time path for Laravel, Python, and operator users evaluating Durable Workflow 2.0 prerelease artifacts.
Give AI agents a stable v2 workflow for discovering, changing, running, and diagnosing Durable Workflow applications.
Treat MCP tools, CLI JSON, server diagnostics, Waterline exports, and SDK fixtures as one machine-operable Durable Workflow v2 contract.
Install and use the Durable Workflow CLI to operate the standalone server.
Freeze the Durable Workflow Cloud control-plane contract, runtime-target boundary, worker connectivity model, and support-led topology edge.
Diagnose activity failures, non-retryable exceptions, run timeouts, cancellation cleanup, and recovery actions.
Waterline is a separate UI
The operator-facing contract for Durable Workflow v2 health, queue state, rebuilds, exports, archives, and deployment topology.
Run a self-serve rolling upgrade across Durable Workflow server API nodes, workers, and the scheduler without stopping live traffic.
Deploy and configure the Durable Workflow standalone server.
Reference for Durable Workflow server environment variables, defaults, legacy aliases, and runtime infrastructure settings.
Use sticky workflow-task routing as a supported v2 replay optimization with cold replay as the correctness fallback.
Understand how Durable Workflow turns ready durable tasks into worker assignments, from long-poll wakeups to dedicated matching-role deployments.
Tune worker slots, server-side queue budgets, and query-task backpressure for the standalone server.
Canonical compatibility and release-authority contract for Durable Workflow public surfaces.
Roll out, canary, drain, and roll back worker builds without stranding executions using build-id rollout state on the standalone server.
Keep worker build rollouts safe by pinning in-flight work to compatible executors and making rollout gaps explicit.
Pin a sequence of durable activity attempts to one worker-session lease for GPU, filesystem, or process-local state affinity.