Add shutdown watchdog for stalled Electron workers

[aleksandr-voitenko]

Adds a coordinated, bounded shutdown flow across Electron’s main process and worker renderer. The main process waits up to 10 seconds for acknowledgement, 30 seconds for worker teardown, and 10 seconds for Electron to exit after completion, force-exiting when a phase stalls.

Worker shutdown now persists local scene and session state before destructive OBS teardown, continues remaining steps after individual failures, bounds analytics flushing to three seconds, and reports whether shutdown completed cleanly. File flushing now supports concurrent shutdown callers and writes that race an in-flight flush.

If a second instance starts during shutdown, its arguments are preserved and one relaunch is scheduled; the active watchdog continues to bound the original process’s exit. Scene-collection cloud synchronization is deferred to startup so network stalls cannot block shutdown.

[bundlemon]

BundleMon

Files added (4)

Status	Path	Size	Limits
✅	renderer.(hash).js	+10.36MB	-
✅	vendors~renderer.(hash).js	+4.67MB	-
✅	updater.js	+115.29KB	-
✅	guest-api.js	+40.23KB	-

Total files change +15.18MB

Final result: ✅

View report in BundleMon website ➡️

---

Current branch size history

[Copilot]

Pull request overview

Adds a coordinated, time-bounded shutdown watchdog across Electron main + worker, and restructures worker shutdown into ordered phases that preserve local state before OBS teardown while keeping best-effort work bounded.

Changes:

• Introduces a main-process shutdown coordinator with acknowledgement / completion / exit watchdog timers, plus controlled relaunch scheduling during shutdown.
• Adds a worker shutdown runner with ordered persistence → teardown → best-effort phases, step-level failure reporting, and bounded analytics flushing.
• Improves shutdown durability/robustness via scene-collection persistence split, stricter crash-reporter state writes, and a flushAll implementation that supports concurrent callers and racing writes.

Reviewed changes

Copilot reviewed 13 out of 13 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
test/regular/worker-shutdown.ts	Adds unit-style coverage for ordered worker shutdown phases, failures, and time-bounded best-effort steps.
test/regular/file-manager-shutdown.ts	Adds tests for concurrent/racing flushAll() behavior and retry exhaustion.
test/regular/app-shutdown.ts	Adds tests validating shutdown watchdog timing, idempotency, and relaunch scheduling semantics.
main.js	Integrates shutdown coordinator into main-process shutdown flow, adds force-exit behavior, and preserves argv for relaunch during shutdown.
electron-builder/base.config.js	Ensures the new shutdown coordinator module is packaged into builds.
app/util/worker-shutdown.ts	New generic shutdown-step runner with reporting, per-step timeouts, and clean-exit marking.
app/util/shutdown-coordinator.js	New main-process watchdog state machine coordinating shutdown acknowledgement/completion/exit phases.
app/services/usage-statistics.ts	Adds optional AbortSignal support so analytics flushing can be bounded during shutdown.
app/services/streaming/streaming.ts	Accumulates shutdown teardown failures and throws a consolidated error after attempting all destroys.
app/services/scene-collections/scene-collections.ts	Splits persistence for shutdown from teardown, and removes cloud sync from shutdown path.
app/services/file-manager.ts	Reworks flushAll() to coordinate concurrent callers and handle racing rewrites/new files during shutdown.
app/services/crash-reporter.ts	Makes shutdown state persistence failures explicit by returning success and throwing on failure.
app/services/app/app.ts	Refactors worker shutdown into immediate ingress-stop + phased shutdown plan executed via the new shutdown runner.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[summeroff]

is 0 intentional?
it is not exactly clean exit, but not a full failure too.

[summeroff]

as we now send additional information with this message.

          electron.ipcRenderer.send('shutdownComplete', {
            clean: report.clean,
            failedSteps: report.failures.map(failure => failure.name),
          });

probably be useful to log it?

ipcMain.on('shutdownComplete', (event, payload) => {
  const { clean, failedSteps } = payload;