Happy Friday! This week, we rolled out a long awaited improvement to the platform, this one is going to be a fun read.

We’ve reduced the time it takes for deployments to become live from 3 minutes to just 10-20 seconds!

This should help restore our “Zero Downtime Default” state of application deploys

 configuration change to disabled CPU-intensive stats collection on our tens of thousands of proxy routes. However, figuring this out and making the change in a zero-downtime way was a multi-week effort.

Since we knew the effort would be large, we took the time to also:

 by adding multiple Envoy proxies behind a TCP load balancer,

 by automatically removing unhealthy Envoy instances,

 by testing new configs with canary Envoy instances, and

Our public proxy is one of the most important parts of our infrastructure and we’re proud to have it back in a dependable state.

Everything else on this changelog pales in comparison to the above but we feel it’s key to note.

The team thinks deeply about how we give our user’s easy access to guides and relevant information about how to use the platform. We have now added relevant setup guides in a few starters within the starter grid, helpful for involved starters that require multiple steps to setup.

we added a starter for... you guessed it, Remix! We maybe shouldn’t have spoiled it with the heading text.

Redeploy Remix via our new template links system: 

https://res.cloudinary.com/railway/image/upload/v1647652767/docs/Proxy-OG_ljy96h.png

Proxy++,  Starter Resources, New Remix Starter

Changelog #0073