[Docs]: "run-server" is undocumented

[aczekajski]

Page(s)

https://playwright.dev/docs/docker#remote-connection - this is the only place in the entire docs that ever mentions the "run-server" capability so it could be here or rather the server should have its own page.

Description

There is zero information on what one should expect from the PW server. I'm trying to create an infrastructure where PW servers could be reused between multiple users/CI workers but besides experimenting, there is no way to know how it will behave.

Does it have any startup options?
What are and aren't intended use cases of it?
What happens if more then one client tries to connect to a single server and execute tests on it? Especially with concurrency in mind, the server can only spawn a limited number of workers before it runs out of resources, but is there a way to control that?
Does the server just blindly accept instructions from clients or there is some queuing mechanism?
What browsers are preinstalled in the docker image and how to run "branded" versions of the browsers with PW server?

These are only my most pressing questions now and I could probably have way more. My point is that it should have more explanations about it in general ^^' It's very useful for taking consistent screenshots between multiple platforms and sounds like it could be useful in CI/CD scenarios if it was better documented.

[pavelfeldman]

This method is used to connect to the server: https://playwright.dev/docs/api/class-browsertype#browser-type-connect
It points to the method that is used to run the server: https://playwright.dev/docs/api/class-browsertype#browser-type-launch-server

[aczekajski]

@pavelfeldman Does that mean that the run-server is "dumb" and will simply do whatever the connecting clients tell it to do? So if 10 clients connect to the same server at the same time and try running a test, the server will try spawning 10 workers and probably just run out of resources? Meaning that the server was never really meant to work in a scenario where the server is always up and can have multiple clients connecting to it to run tests?

I don't really see how the two methods you pointed to answer my questions about the "run-server".

[aczekajski]

Looking a bit into code myself, I can see that the "run-server" indeed has some options but again, they are undocumented:

playwright/packages/playwright-core/src/cli/program.ts

Line 295 in c73129a

.command('run-server')