[kibanamachine]

Fixes #300968

Summary

• Five EventFlyout cases that open the footer Actions menu drove the entire cost of the file and ran past Jest's 5s budget under CI load — only offers Mark inactive... at 5.358s, and opens a new chat from the Actions menu at 5.233s in the latest failure. Nothing failed an assertion.
• They opened the menu with await userEvent.click, whose pointer sequence re-checks computed pointer-events styles across the whole flyout tree on every click. This swaps all nine clicks for fireEvent.click: togglePopover is a plain setIsOpen state update (menu_button.tsx:134-137) and RTL wraps fireEvent in act, so the item is queryable on the next line — the tests need the menu open, not a full pointer sequence.
• Mark inactive was the only entry in actionMenuPanels without a data-test-subj, so five assertions reached for role + accessible name — RTL's most expensive query. It now carries one, making those assertions plain attribute lookups like its siblings'.
• No timeout bump: the patch makes the test cheaper rather than waiting longer.

Runtime vs. 5s budget	Passed	Avg	Max
Before fix	25/25	655ms	707ms
After fix	25/25	142ms	278ms

Relevant history

• [SigEvents] Redesign the Nightshift significant event flyout #299456: redesigned the significant event flyout, consolidating the former footer buttons into the EuiPopover + EuiContextMenu Actions menu and adding these five cases together with it — before it no case had to open a popover at all.

Context

• Follows the latest investigator analysis, which supersedes its own earlier one: the first analysis pinned the cost on the *ByRole(..., { name }) menu assertions, but opens a new chat from the Actions menu already queries by test id with a single render and still blew the budget, so the dominant cost is the shared await userEvent.click path. Local profiling confirms this — see Verification.
• The failing case does not reproduce on an idle runner (707ms against a 5000ms budget); CI timings are ~7.6x local for this case, which is what puts it over the edge.
• Failures landed on kibana-on-merge for 9.6, twice in three days (build 114241, build 114692); the second build timed out on two of the five cases. The shard was not degraded (2 failures out of 15,857 cases). The same unfixed test is on main, so the fix lands here.
• Ruled out a product-side race: every hook the flyout consumes is mocked to settled data, so no longer wait could resolve anything.

Verification

Verified locally

• ✅ Passed: node scripts/eslint x-pack/solutions/observability/plugins/nightshift/public/components/entities/significant_event/event_flyout.test.tsx x-pack/solutions/observability/plugins/nightshift/public/components/entities/significant_event/event_flyout.tsx

• ✅ Passed: node scripts/jest x-pack/.../significant_event/event_flyout.test.tsx — 25 runs before the fix and 25 after, 25/25 passing in both loops (the flake needs CI load and does not reproduce on an idle runner), with the five Actions-menu cases dropping from 2384ms avg / 2537ms max combined to 611ms / 767ms — 3.9x cheaper:

  Case	Before (avg/max)	After (avg/max)
  only offers Mark inactive... (reported)	655ms / 707ms	142ms / 278ms
  opens a new chat from the Actions menu	476ms / 505ms	147ms / 167ms
  marks an active event inactive...	503ms / 534ms	136ms / 175ms
  runs an investigation from the footer...	375ms / 503ms	104ms / 119ms
  offers a new chat instead...	373ms / 389ms	80ms / 89ms

  Afterwards no Actions-menu case is among the most expensive in the file — they sit at the level of the plain-render cases (~56–143ms).

• ✅ Passed: no not wrapped in act warnings appear after the userEvent → fireEvent swap, confirming the menu item is in the DOM synchronously and no gating findByTestId is needed.

Not verified locally

• Behavior under CI's parallel load, which is what pushed these cases over the budget — neither loop ran under contention.
• Type checking was left to CI.

Important

The flaky test runner may verify the PR, but the owning team is responsible the next steps:

• Review and merge it (update the branch or add missing labels if needed)
• Make changes (check out the branch locally, or ask @copilot if Copilot is enabled on your account)
• Close it if it is not worth shipping (share feedback in #kibana-qa)

Generated by Flaky Test Fixer for #300968 · claude · opus · 394 AIC · ⌖ 58.5 AIC · ⊞ 15.4K · ◷

[infra-vault-gh-plugin-prod]

🤖 Jobs for this PR can be triggered through checkboxes. 🚧

ℹ️ To trigger the CI, please tick the checkbox below 👇

• Click to trigger kibana-pull-request for this PR!
• Click to trigger kibana-deploy-project-from-pr for this PR!
• Click to trigger kibana-deploy-cloud-from-pr for this PR!
• Click to trigger kibana-entity-store-performance-from-pr for this PR!
• Click to trigger kibana-storybooks-from-pr for this PR!
• Click to trigger kibana-static-checks for this PR!

[kibanamachine]

⏭️ Flaky-fix verification skipped

This is a Jest-only change, so the flaky test runner can add no signal and the required CI Jest run is the whole verdict. Applied release_note:skip, backport:version and v9.6.0.

Why the flaky test runner wasn't used

The /flaky runner accepts only FTR and Scout configs. This PR touches event_flyout.test.tsx (Jest) plus a one-line data-test-subj addition in event_flyout.tsx, and the plugin's only Scout spec (test/scout/ui/tests/nightshift.spec.ts) never opens the event flyout's Actions menu — so no runner-supported config exercises the change. Repeated runs would also add nothing here: the fix removes cost rather than mitigating a race.

Screened the patch against the fix guidelines before skipping. togglePopover is a plain setIsOpen state update and resolvePanels spreads ...item, so the menu items are in the DOM synchronously after fireEvent.click and the new data-test-subj reaches the rendered item. The assertions still drive mockOpenChat / mockSignificantEventsFetch, nothing is skipped or loosened to pass, and no timeout was raised — this cuts the work rather than hiding the failure.

How release-note and backport labels were chosen

Applied release_note:skip because the only product change adds a data-test-subj attribute, which changes no user-facing behavior.

• v9.6.0 → has both changed files, with the five Actions-menu cases and the Mark inactive item byte-identical to main's pre-fix state, so the patch applies unadapted.
• v9.5.6 → the nightshift plugin does not exist on 9.5.
• v9.4.9 → the nightshift plugin does not exist on 9.4.
• v8.19.24 → the nightshift plugin does not exist on 8.19.

Warning

Firewall blocked 1 domain

The following domain was blocked by the firewall during workflow execution:

• raw.githubusercontent.com

To allow these domains, add them to the network.allowed list in your workflow frontmatter:

network:
  allowed:
    - defaults
    - "raw.githubusercontent.com"

See Network Configuration for more information.

Generated by Flaky Fix Verifier for #301793 · claude · opus · 345.8 AIC · ⌖ 58 AIC · ⊞ 16.9K · ◷

[infra-vault-gh-plugin-prod]

Pinging @elastic/nightshift (Team:nightshift)