fix(ci): don't fail e2e job when dev server already exited

[aterga]

Problem

The e2e-playwright matrix' "Stop dev server" cleanup step runs:

- name: Stop dev server
  if: ${{ always() }}
  run: kill ${{ steps.dev-server-start.outputs.dev_server_pid }}

under bash -e. The dev server is started as a backgrounded npm run dev | tee … & pipeline, and the captured $! PID is frequently already reaped by the time this cleanup step runs. kill then exits non-zero (kill: (NNNN) - No such process), and because the step runs under bash -e, the whole job fails — after every test has already passed.

This produces spurious red e2e shards (observed repeatedly on desktop, 2_6 and mobile, 2_6) on PRs that don't touch any test or app code, e.g. a docs/scripts-only change. Example log tail:

  31 passed (5.4m)
Run kill 3405
kill: (3405) - No such process
##[error]Process completed with exit code 1.

Fix

Make the cleanup non-fatal:

run: kill ${{ steps.dev-server-start.outputs.dev_server_pid }} 2>/dev/null || true

Kill semantics are unchanged — anything still alive is reaped by the runner's orphan-process cleanup (visible as Terminate orphan process: … at job end). The step simply no longer turns a green run red. This is the only kill-based cleanup step in the workflows (verified), so one edit covers all e2e shards.

Testing

• yaml.safe_load parses the workflow.
• The change is confined to the cleanup step; it cannot affect test outcomes.

🤖 Generated with Claude Code

https://claude.ai/code/session_01CGvgPmEVcgyUP2mAnwnDcj

---

Generated by Claude Code