Deployment Lifecycle - Replov Docs
Skip to main content

Deployment Lifecycle

From a dashboard deployment request to a running Docker service

A deployment uses the repository and branch saved on the service. It starts when you click Deploy in the Replov dashboard, or automatically on each push to that branch when Auto Deploy is on.

Deployment stages

1

Queued

Replov creates a deployment record and waits for a worker.

2

Clone

The worker clones the configured repository and branch.

3

Prepare

Service settings, environment variables, and the applicable Docker build path are prepared.

4

Build

Docker builds an immutable image while logs stream to the deployment record.

5

Replace service

The Docker Swarm service is created or updated to the new image and resource limits.

6

Verify

Replov checks container and service state, then records success or failure.

Current release strategy
Replov currently updates or recreates the Docker service rather than running a blue-green pair. A deployment, restart, or plan change can briefly interrupt traffic, so schedule production changes appropriately.

Failures and recovery

  • Inspect the deployment log to identify clone, dependency, build, or startup failures.
  • Correct the repository or service configuration before redeploying.
  • Use an available previous deployment image for rollback only after confirming it exists and is compatible with current environment variables and data.

Application readiness

Bind the application to 0.0.0.0, listen on the configured service port, and expose a lightweight health endpoint when the framework supports one. Keep database migrations backward-compatible with the version being replaced.

Enable Auto Deploy from the service header to deploy on signed GitHub pushes. Replov matches the repository and the service branch, so development and production services deploy independently.