bug: Orphan dev servers — the usual cause of a slow machine

[wtostivint]

Is there an existing issue for this?

• I have searched the existing issues

Relevant guide(s)

Dev Problem : not link to guide

Current Behavior

Ctrl-C on pnpm dev kills the pnpm wrapper but frequently leaves the rspress.js dev child
running. It keeps holding its port, so the next pnpm dev silently starts on 3001, then 3002,
and so on. Each instance holds 250MB–1.4GB, so a few sessions add up to several GB with no
visible window to close.

If the dev server announces a port other than 3000, an orphan is already there — clean up
rather than working on the new port.

Expected Behavior

No response

Steps To Reproduce

pnpm dev
Ctrl-C on pnpm dev

Describe alternatives you've considered

Note the process name: the command line is rspress.js dev, so the intuitive
pkill -f "rspress dev" matches nothing. Use:

pgrep -fl "rspress.js dev"         # list what is still running
pkill -9 -f "rspress.js dev"       # kill all of them
lsof -nP -iTCP:3000 -sTCP:LISTEN   # check a specific port

Which operating systems have you used?

None

Which web browser have you used?

None

Anything else?

No response