Changes to Tiltfile Config do not stop previously started services

[andymartin-sch]

I'm not sure if this should be considered a Tilt bug or not, but we use set_enabled_resources from https://docs.tilt.dev/tiltfile_config.html to selectively enable services from our helm charts. The problem is that when we try to stop services that were previously enabled, Tilt doesn't actually stop those services (pods stay Running, etc).

This sort of intuitively makes sense, since changes to the Tiltfile Config trigger a Tiltfile re-execution and on that secondary execution Tilt has no way of really knowing that it should stop previously-launched pods/services, but it seems to go against the "user story" of https://docs.tilt.dev/tiltfile_config.html where people can select arbitrary sets of services to run without stopping everything and restarting.

In Slack, someone else mentioned:

I've had similar issues with tilt_options.json . My guess is this really feeds back into the need for lifecycle hooks, but I think removing a service should (by default) destroy related resources. There is something to be said about unexpected side effects of doing such a thing, though

Potentially related: #1699

This is not a critical/major issue, but it does come up as a developer QoL thing and it makes us more likely to come up with our own custom workflow down the road if more developers find this annoying.

[nicks]

also related to #3137. I've always wanted a garbage collector so that when you remove resources from a tiltfile during a run, Tilt deletes any resources it deployed during that session.

[nicks]

I'm pretty sure this will get fixed in #4700 in the next release

[andymartin-sch]

Yep thanks! This seems to be fixed in 0.21.1 - nice :)

[nicks]

🎉 🎉 🎉