HDDS-9793. [Recon][UI][Test] Add tests for Containers page

[sravani-revuri]

What changes were proposed in this pull request?

This task adds test for the container page on recon UI

What is the link to the Apache JIRA

https://issues.apache.org/jira/browse/HDDS-9793

How was this patch tested?

Ran tests.

 ✓ src/__tests__/containers/Containers.test.tsx (12 tests) 1814ms
   ✓ Containers Component (12)
     ✓ renders component correctly 217ms
     ✓ Loads data on mount 66ms
     ✓ loads missing containers on mount and updates highlights 62ms
     ✓ Renders table with correct number of rows 60ms
     ✓ Displays no data message if the missing containers API returns an empty array 65ms
     ✓ lazy-loads data when switching to under-replicated tab 205ms
     ✓ lazy-loads data when switching to over-replicated tab 134ms
     ✓ lazy-loads quasi closed containers when switching tabs 122ms
     ✓ Handles search input change  375ms
     ✓ Displays a message when no results match the search term  366ms
     ✓ Handles API errors gracefully by showing error message 51ms
     ✓ Handles API errors when loading under-replicated tab 89ms

 Test Files  1 passed (1)
      Tests  12 passed (12)
   Start at  11:26:25
   Duration  5.18s (transform 129ms, setup 136ms, collect 2.19s, tests 1.81s, environment 537ms, prepare 124ms)

[chihsuan]

Thanks for working on this! @sravani-revuri The structure looks good and matches the other Recon tests nicely.

I tried breaking the behavior each test names, and a few still pass. See inline for details.

[chihsuan]

nit: Is this special case needed? The mock server already handles this endpoint, and the tests still pass for me without it.

[sravani-revuri]

The server already handles /api/v1/clusterState and the test assertions do pass, but vitest run reports unhandled CanceledErrors on that URL without this check . Containers fetches clusterState twice on mount (initialFetch: true + clusterState.refetch() from autoReload), so the second request aborts the first and that rejection isn’t caught. I checked out how pipelines is doing it and it uses initialFetch: false instead hence it doesn't face this problem.

I've currently added a comment to justify this, happy to drop / modify in a follow-up accordingly.

[chihsuan]

Thanks for digging into this! @sravani-revuri The double fetch also happens on the page itself, so I've opened HDDS-16877 as a follow-up, and #11484 fixes it. Since this PR will merge first, we can drop the special case in #11484.

[smengcl]

The 12 tests pass locally, but the suite also passes with these deliberate regressions:

• An empty API response incorrectly produces a row: the empty-state assertion passes against the initial “No Data” placeholder.
• The Quasi Closed highlight shows N/A: the whole-card match for 2 also matches 20 or 25.
• Tab data loads eagerly on mount or reloads on every visit: the lazy-loading tests only check that rows eventually appear.

Could we strengthen these assertions before merging? The existing inline threads already cover the highlight and lazy-loading gaps.

[smengcl]

Thanks for the update. lgtm

[chihsuan]

Thanks @sravani-revuri for the patch, @smengcl for the review.