Skip to content

fix(chromium): skip on-device model performance measurement - #44014

Merged
Yury Semikhatsky (yury-s) merged 2 commits into
mainfrom
triage/43992/42684574106-1
Nov 9, 2026
Merged

Yury Semikhatsky (yury-s) merged 2 commits into
mainfrom
triage/43992/42684574106-1

Conversation

@microsoft-playwright-automation

Copy link
Copy Markdown
Contributor

Chromium schedules on-device model performance measurement even when startup metrics are disabled. Use --optimization-guide-performance-class=8, matching Chrome for Testing's GPU-blocked class, to avoid this background work.

The new test needs a real page and a delayed startup measurement; the existing launcher tests check argument validation, debugging ports, or the absence of automatic pages. It fails on Edge 157 without the fix and passes with it. Launch checks passed on Edge, Chromium, Firefox and WebKit; build and flint passed. The reported long CPU spike did not occur on this runner.

Fixes #43992

Triaged on Linux by the Playwright bot - agent run

@github-actions

This comment has been minimized.

@github-actions

github-actions Bot commented Nov 9, 2026

Copy link
Copy Markdown
Contributor

🟢 The reported failure is a pre-existing flake

Hi, I'm the Playwright bot. I checked the PR diff and refreshed CI history. The single failure in the latest MCP report is unlikely to be caused by this PR.

Details

The PR adds --optimization-guide-performance-class=8, a launcher test, and a test registry entry. The flag applies to this test's Chrome launch, but the same failure already occurs on earlier commits without this change. No WebMCP tool collection or navigation code changes here.

Pre-existing flake

[chrome] › mcp/webmcp-dynamic.spec.ts:189 › a tool registered in an iframe can be called after the page is reloaded — on Ubuntu, the second tool call returns Tool "webmcp_add" not found instead of "text": "5".

The identical assertion failed on the same Chrome/Ubuntu bot:

The refreshed database records 17 failed and 720 passed runs for this exact test on the same bot outside #44014, from October 12 through November 9. This confirms a pre-existing intermittent failure, rather than a new failure introduced here.

CI failures checked by the Playwright bot - agent run

Triaged by the Playwright bot - agent run

@github-actions

This comment has been minimized.

@yury-s
Yury Semikhatsky (yury-s) marked this pull request as ready for review November 9, 2026 22:26
@github-actions

github-actions Bot commented Nov 9, 2026

Copy link
Copy Markdown
Contributor

Hi, I'm the Playwright bot. I started to address the review comments on Linux: agent run

@microsoft-playwright-automation

Copy link
Copy Markdown
Contributor Author

Hi, I'm the Playwright bot. I removed the benchmark test and its registry entry as requested; the Chromium launch switch is unchanged.

Build and flint passed. Launch tests across Chromium, Firefox and WebKit: 60 passed, 1 skipped. I temporarily restored the removed test to check Edge 156: it failed without the switch and passed with it, then I removed it again. The reported long CPU spike was not verified in this revision.

Revised on Linux by the Playwright bot - agent run

@github-actions

github-actions Bot commented Nov 9, 2026

Copy link
Copy Markdown
Contributor

Test results for "MCP"

5229 passed, 637 skipped


Merge workflow run.

@github-actions

github-actions Bot commented Nov 9, 2026

Copy link
Copy Markdown
Contributor

Test results for "tests 1"

23618 passed, 705 skipped


Merge workflow run.

@yury-s
Yury Semikhatsky (yury-s) merged commit b43d534 into main Nov 9, 2026
31 checks passed
@yury-s
Yury Semikhatsky (yury-s) deleted the triage/43992/42684574106-1 branch November 9, 2026 23:20
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[Bug]: msedge on Linux starts OnDeviceModelService ~3 min after launch, using ~150% CPU per browser for ~90 s

1 participant