ignore: flake testing

[david-crespo]

Backing out "chore: convert last few routes (#2719)" and add it back bit by bit to see how it behaves in CI.

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Updated (UTC)
console	✅ Ready (Inspect)	Visit Preview	Mar 3, 2025 6:08pm

[david-crespo]

This is helpful because it confirms the flake was introduced with that commit. The flake we see in the first commit here is a regular old one, #2694.

[david-crespo]

Might be interesting that even in the FF and Chrome tests that passed, we are seeing more flake than we did before (I think). This suggests that the PR that made it start failing consistently just took it over some probability threshold rather than being a qualitative difference.

  Slow test file: [firefox] › test/e2e/project-access.e2e.ts (19.5s)
  Consider running tests from slow files in parallel, see https://playwright.dev/docs/test-parallel.
  7 flaky
    [firefox] › test/e2e/breadcrumbs.e2e.ts:35:1 › breadcrumbs ─────────────────────────────────────
    [firefox] › test/e2e/floating-ip-create.e2e.ts:56:1 › can detach and attach a floating IP ──────
    [firefox] › test/e2e/instance-auto-restart.e2e.ts:47:1 › Auto restart policy on running instance 
    [firefox] › test/e2e/instance-create.e2e.ts:538:1 › create instance with additional disks ──────
    [firefox] › test/e2e/ip-pools.e2e.ts:319:1 › no ranges means no utilization bar ────────────────
    [firefox] › test/e2e/login.e2e.ts:19:3 › login › with redirect_uri param redirects to last page 
    [firefox] › test/e2e/lookup-routes.e2e.ts:38:3 › /lookup/instances › looks up instance by ID ───
  1 skipped
  176 passed (11.9m)

  Slow test file: [chrome] › test/e2e/breadcrumbs.e2e.ts (23.6s)
  Slow test file: [chrome] › test/e2e/action-menu.e2e.ts (20.0s)
  Slow test file: [chrome] › test/e2e/scroll-restore.e2e.ts (19.8s)
  Slow test file: [chrome] › test/e2e/floating-ip-update.e2e.ts (19.5s)
  Slow test file: [chrome] › test/e2e/project-access.e2e.ts (19.2s)
  Consider running tests from slow files in parallel, see https://playwright.dev/docs/test-parallel.
  10 flaky
    [chrome] › test/e2e/firewall-rules.e2e.ts:114:1 › firewall rule targets and filters overflow ───
    [chrome] › test/e2e/firewall-rules.e2e.ts:529:1 › can edit rule directly by URL ────────────────
    [chrome] › test/e2e/floating-ip-create.e2e.ts:56:1 › can detach and attach a floating IP ───────
    [chrome] › test/e2e/instance-auto-restart.e2e.ts:47:1 › Auto restart policy on running instance 
    [chrome] › test/e2e/instance-create.e2e.ts:283:1 › add ssh key from instance create form ───────
    [chrome] › test/e2e/instance-create.e2e.ts:538:1 › create instance with additional disks ───────
    [chrome] › test/e2e/instance-networking.e2e.ts:12:1 › Instance networking tab — NIC table ──────
    [chrome] › test/e2e/instance-serial.e2e.ts:12:1 › serial console can connect while starting ────
    [chrome] › test/e2e/lookup-routes.e2e.ts:21:3 › /lookup/i › looks up instance by ID ────────────
    [chrome] › test/e2e/utilization.e2e.ts:108:3 › Silo utilization › works for fleet viewer ───────
  1 skipped
  173 passed (14.2m)

And in the trace, we see similar behavior to more failure-prone commits — the page is just taking slightly longer than the default 5 second timeout to load. Testing a longer timeout in #2729.