Durable automation made simple

Implementing deploy-sentry workflow with few lines of code

To reduce operational risk, a new code version can be deployed to production gradually. For example, deploy the version to a portion of the servers and monitor them. If all works well, deploy to additional servers; otherwise, stop deployment or rollback.

This workflow is long-running by definition and may be quite challenging to implement. The challenge is not necessarily in the code but in the system supporting it. You need to take care of APIs and secrets, deployment, monitoring, and ensure reliability in case of temporal network or infrastructure failures.

Since it is an asynchronous, long-running task that orchestrates APIs, you would probably need to use queues and database to store the state and build an infrastructure to support it.

In the demo we presented in Termporal’s Meetup in San Fransico on March 2024, we showed how easy and elegant it is to build such a workflow on top of autokitteh with just a few lines of code. Check out the video.