Browser mode passes all tests but then does not exit cleanly: ✘ [ERROR] The build was canceled

[kgetz-arista]

I'm trying to use browser mode (with Playwright) and I've been able to migrate a small package, but Vitest does not exit cleanly, and I don't understand how I'm supposed to debug this. Here's the output when I run the tests:

 RUN  v4.0.16 C:/Users/kgetz/Work/event-viewer/geiger-lib/geiger-devices-tables

9:38:25 PM [vite] (client) Re-optimizing dependencies because lockfile has changed
 ✓  chromium  src/components/TableChangesList/Message/__test__/index.spec.tsx (3 tests) 46ms
 ✓  chromium  src/components/TableChangesList/ListItem/__test__/index.spec.tsx (2 tests) 66ms
 ✓  chromium  src/components/PartitionTable/__test__/utils.spec.ts (1 test) 14ms
 ✓  chromium  src/utils/__test__/routingTableUtils.spec.ts (3 tests) 6ms
 ✓  chromium  src/components/HardwareCapacityTable/__tests__/utils.spec.ts (2 tests) 2ms
 ✓  chromium  src/components/HardwareCapacityTable/__tests__/SharedFeaturesCell.spec.tsx (2 tests) 40ms
 ✓  chromium  src/components/RoutingTable/__test__/Prefix.spec.tsx (2 tests) 32ms
 ✓  chromium  src/components/Nexthops/__test__/index.spec.tsx (2 tests) 50ms

 Test Files  8 passed (8)
      Tests  17 passed (17)
   Start at  21:38:25
   Duration  28.87s (transform 0ms, setup 4.88s, import 23.36s, tests 255ms, environment 0ms)

✘ [ERROR] The build was canceled

close timed out after 10000ms
Tests closed successfully but something prevents the main process from exiting
You can try to identify the cause by enabling "hanging-process" reporter. See https://vitest.dev/config/#reporters

I'm seeing this issue both locally on my laptop (Windows 11) and in our CI system, which is using the official Playwright Docker image (mcr.microsoft.com/playwright:v1.56.1-noble). If I enable the hanging-process reporter, I see this output:

21:56:25   Test Files  8 passed (8)
21:56:25        Tests  17 passed (17)
21:56:25     Start at  05:56:13
21:56:25     Duration  11.85s (transform 0ms, setup 6.29s, import 14.97s, tests 217ms, environment 0ms)
21:56:25  
21:56:37  There are 810 handle(s) keeping the process running
21:56:37  
21:56:37  # FILEHANDLE
21:56:37  (unknown stack trace)

(repeated hundreds of times)

21:56:37  # FSEVENTWRAP
21:56:37  file:///home/jenkins/workspace/Geiger/geiger-review/common/temp/node_modules/.pnpm/vite@7.3.1_@types+node@24.10.4_less@4.5.1_lightningcss@1.30.2_terser@5.36.0_tsx@4.21.0_yaml@2.8.1/node_modules/vite/dist/node/chunks/config.js:12942
21:56:37  file:///home/jenkins/workspace/Geiger/geiger-review/common/temp/node_modules/.pnpm/vite@7.3.1_@types+node@24.10.4_less@4.5.1_lightningcss@1.30.2_terser@5.36.0_tsx@4.21.0_yaml@2.8.1/node_modules/vite/dist/node/chunks/config.js:12984
21:56:37  file:///home/jenkins/workspace/Geiger/geiger-review/common/temp/node_modules/.pnpm/vite@7.3.1_@types+node@24.10.4_less@4.5.1_lightningcss@1.30.2_terser@5.36.0_tsx@4.21.0_yaml@2.8.1/node_modules/vite/dist/node/chunks/config.js:13098
21:56:37  file:///home/jenkins/workspace/Geiger/geiger-review/common/temp/node_modules/.pnpm/vite@7.3.1_@types+node@24.10.4_less@4.5.1_lightningcss@1.30.2_terser@5.36.0_tsx@4.21.0_yaml@2.8.1/node_modules/vite/dist/node/chunks/config.js:13142
21:56:37  file:///home/jenkins/workspace/Geiger/geiger-review/common/temp/node_modules/.pnpm/vite@7.3.1_@types+node@24.10.4_less@4.5.1_lightningcss@1.30.2_terser@5.36.0_tsx@4.21.0_yaml@2.8.1/node_modules/vite/dist/node/chunks/config.js:13316
21:56:37  file:///home/jenkins/workspace/Geiger/geiger-review/common/temp/node_modules/.pnpm/vite@7.3.1_@types+node@24.10.4_less@4.5.1_lightningcss@1.30.2_terser@5.36.0_tsx@4.21.0_yaml@2.8.1/node_modules/vite/dist/node/chunks/config.js:13958
21:56:37  
21:56:37  # FILEHANDLE
21:56:37  (unknown stack trace)

(repeated dozens of times)

21:56:37  close timed out after 10000ms
21:56:37  Tests closed successfully but something prevents the main process from exiting

Does anyone know what this means? It appears that Vite itself is the one keeping the file handles open. What's the next step here?