I see users ask about this now and then, seems like I forget about this forum post though.

This would be very helpful in implementing blue-green deploys

We have a backend and frontend service that need to be deployed in lockstep, I get anxious about one of the two builds failing and leaving us in a broken state.

Being able to orchestrate rollouts for those dependent services together would be much appreciated!

We have two separate services, with one service depending on the other - at startup, the one service subscribes to the other, but of course that only works if the other service is actually available. Currently I use a bash script that pings to check if it's available as part of the deploy script, but that means I have to extend the health check timeout since it gets stuck in the "deploying" status for much longer.

It would be awesome to mark a service as depending on the healthcheck of another service to pass, so I can have an appropriately short health check timeout, and I don't need to modify the startup script to have to wait. It would also mean I can get a more descriptive error if the "child" service ends up not being able to start because the "parent" service crashed/didn't start.

+1 would be very helpful for zero-downtime deployment of dependent services.

Ideally this could be at a service group or environment level to enable full blue/green deploys.

This is rolled out BTW https://railway.com/changelog/2024-11-22-google-auth#control-service-startup-order

IIUC, this is only for the first deployment (when no previous deployment is running)?

I.e if i push to a branch and it kicks off builds for serviceA and serviceB where A depends on B, there is nothing stopping this new deployment of serviceA from starting before the new deployment of serviceB (which could cause downtime if theres a breaking change, and what if serviceB actually failed to deploy, then extended downtime).

Don't think this is complete unless I am missing something?