👷 use BrowserStack plan endpoint to gate test runs

[thomas-lebeau]

Motivation

While investigating why our CI was blocked from launching new BrowserStack runs, we found that a previous CI job (unit-bs on a now-cancelled pipeline) had left 5 sessions stuck in the running state on BrowserStack even though the job had failed over an hour earlier. The driver client never connected (sessions are internally queued), so neither DELETE /automate/sessions/{id} nor DELETE /automate/builds/{id} can clear them — only BS support can.

waitForAvailability() in scripts/test/bs-wrapper.ts queries /automate/builds.json?status=running, which counts these zombie builds and would block local test runs indefinitely until BS support clears them.

The BrowserStack dashboard, however, correctly showed "Parallels 0/5 running | 0 queued" — because zombie builds don't actually consume parallel slots. That state is exposed via /automate/plan.json (parallel_sessions_running + queued_sessions), which is what BS itself gates new sessions on.

Changes

• Replace BS_BUILD_URL (/automate/builds.json?status=running) with BS_PLAN_URL (/automate/plan.json) in scripts/test/bs-wrapper.ts
• Rename hasRunningBuild → hasParallelSessionsInUse
• Use parallel_sessions_running + queued_sessions > 0 as the availability gate
• Add a comment explaining the zombie-build edge case

Test instructions

1. Set BS_USERNAME / BS_ACCESS_KEY env vars
2. Run yarn typecheck (passes)
3. Manual sanity check:

   curl -s -u "$BS_USERNAME:$BS_ACCESS_KEY" https://api.browserstack.com/automate/plan.json

   should return { parallel_sessions_running, queued_sessions, ... }
4. Run a BS test job (e.g. yarn test:unit:bs) and verify it proceeds when no other parallel sessions are active

Checklist

• Tested locally
• Tested on staging
• Added unit tests for this change.
• Added e2e/integration tests for this change.
• Updated documentation and/or relevant AGENTS.md file

[cit-pr-commenter-54b7da]

Bundles Sizes Evolution

📦 Bundle Name	Base Size	Local Size	𝚫	𝚫%	Status
Rum	179.65 KiB	179.65 KiB	0 B	0.00%	✅
Rum Profiler	6.17 KiB	6.17 KiB	0 B	0.00%	✅
Rum Recorder	27.03 KiB	27.03 KiB	0 B	0.00%	✅
Logs	56.78 KiB	56.78 KiB	0 B	0.00%	✅
Rum Slim	135.50 KiB	135.50 KiB	0 B	0.00%	✅
Worker	23.63 KiB	23.63 KiB	0 B	0.00%	✅

🚀 CPU Performance

Action Name	Base CPU Time (ms)	Local CPU Time (ms)	𝚫%
RUM - add global context	0.0039	0.0069	+76.92%
RUM - add action	0.0131	0.0199	+51.91%
RUM - add error	0.0117	0.0182	+55.56%
RUM - add timing	0.0025	0.004	+60.00%
RUM - start view	0.0119	0.0175	+47.06%
RUM - start/stop session replay recording	0.0006	0.0011	+83.33%
Logs - log message	0.0138	0.0224	+62.32%

🧠 Memory Performance

Action Name	Base Memory Consumption	Local Memory Consumption	𝚫
RUM - add global context	31.78 KiB	32.89 KiB	+1.11 KiB
RUM - add action	56.94 KiB	55.92 KiB	-1.02 KiB
RUM - add timing	32.72 KiB	31.59 KiB	-1.13 KiB
RUM - add error	60.80 KiB	61.76 KiB	+990 B
RUM - start/stop session replay recording	32.10 KiB	31.47 KiB	-651 B
RUM - start view	484.29 KiB	484.03 KiB	-271 B
Logs - log message	95.66 KiB	102.95 KiB	+7.29 KiB

🔗 RealWorld

[datadog-official]

🎉 All green!

❄️ No new flaky tests detected
🧪 All tests passed

🎯 Code Coverage (details)
• Patch Coverage: 100.00%
• Overall Coverage: 77.03% (+0.02%)

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 69e63ad | Docs | Datadog PR Page | Give us feedback!}