[Bug]: Renderer stops reporting client activity after an interrupted report, so usage limits never refresh again #15734

Before submitting

• I searched existing issues and did not find a duplicate.
• I included enough detail to reproduce or investigate the problem.

Area

apps/web

Steps to reproduce

1. Run the desktop app on macOS with the default balanced background activity profile.
2. Close the lid so the Mac sleeps and dark-wakes for a while, so the renderer's environment connection is torn down while a server.reportClientActivity request is in flight.
3. Open the lid and keep using the app.

Deterministic repro as a unit test: in backgroundActivityReporterLayer, make the first registry.run(...) for the activity report end with Effect.failCause(Cause.interrupt()), then advance the clock 60 s. No further reports are sent. The same test with Effect.fail(...) keeps reporting.

Expected behavior

The renderer keeps sending reportClientActivity every 25 s, so provider-status leases stay active and provider health and usage limits keep refreshing every 5 minutes.

Actual behavior

The renderer never reports client activity again until the window is reloaded. With no leases, BackgroundPolicy.shouldRunScopeWork({ type: "provider-status" }) is always false, so the provider refresh loop skips every interval. Usage limits froze for 4+ hours while the server kept running. All ~/.t3/caches/*.json usageLimits.checkedAt stopped at the same instant, and interacting with the window did not help. A manual server.refreshProviders call still worked immediately.

Cause: in apps/web/src/lib/backgroundActivityReporter.ts, each per-environment report is wrapped in Effect.ignore. That swallows failures but not interruption. An interrupted request ends report, which ends the Stream.runForEach(() => report) fiber. After that, the 25 s schedule and the interaction listeners keep offering to the sliding queue, but nothing consumes it.

Impact

Major degradation or frequent failure

Version or commit

T3 Code (Alpha) 0.0.45 desktop (official release). The same code is on main at 4ee6bfd.

Environment

macOS (Apple silicon), desktop app, default balanced background profile, several saved environments.

Logs or stack traces

From ~/.t3/userdata/logs/server.trace.ndjson* (local time):

• syncProvider ran every ~5 min, the last time at 22:31:49, then never again.
• web.backgroundActivity.report ran every ~25 s until 22:38:28. That last report took 15.1 s and ended with status.interrupted: true (⚠︎ Interrupted) during a lid-closed dark wake. There are zero web.backgroundActivity.report / server.reportClientActivity spans after it, through 4+ hours of active use.
• Other 5-minute server work (SourceControlProviderRegistry.detectProviderContext) continued normally the whole time.

Workaround

Reload the window (Cmd+R), or call server.refreshProviders manually.

I have a two-line fix with tests ready: Effect.ignore → Effect.ignoreCause for the per-environment report, plus a reporter-layer test covering both a failed and an interrupted report. The interrupted case fails without the fix. Happy to open a PR if this direction looks right.