Reduce false Firefox NS_BINDING_ABORTED health-check failures on cold starts

[Copilot]

Firefox was reporting NS_BINDING_ABORTED for GET /health and GET /ml/docs because client-side AbortController timeouts were too aggressive during backend/ML cold start windows. This updates health probe timing to reduce premature cancellation while preserving the existing wake/retry flow.

• Health-check timeout alignment (Login flow)

  • Increased /health probe timeout from 5s → 15s for both initial and retry checks.
  • Increased /ml/docs probe timeout from 8s → 15s for both initial and retry checks.

• ML probe timeout alignment (Predict flow)

  • Increased /ml/docs probe timeout from 8s → 15s for both initial and retry checks.

• Behavioral impact

  • Fewer client-aborted probes during Render cold starts.
  • Lower incidence of false-negative “server unavailable” states caused by short frontend abort windows.

// Before
const timeoutId = setTimeout(() => controller.abort(), 8000);

// After
const timeoutId = setTimeout(() => controller.abort(), 15000);

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
expense-keeper	Ready	Preview, Comment	Jun 9, 2026 5:52pm

[Copilot]

Pull request overview

This PR adjusts frontend AbortController timeout windows used by /health and /ml/docs probes to reduce false “server unavailable” states (notably Firefox NS_BINDING_ABORTED) during backend/ML cold starts, while keeping the existing wake + retry behavior.

Changes:

• Increased ML probe timeout (/ml/docs) to 15s in both Predict and Login flows (initial + retry).
• Increased backend health probe timeout (/health) to 15s in the Login flow (initial + retry).

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
frontend/src/pages/Predict.jsx	Extends /ml/docs abort timeout from 8s to 15s for initial + retry ML readiness checks.
frontend/src/pages/Login.jsx	Extends /ml/docs and /health abort timeouts to 15s for initial + retry readiness checks during login/signup.

Comments suppressed due to low confidence (5)

frontend/src/pages/Predict.jsx:96

• Same timer-cleanup issue in the retry probe: if the retry fetch throws before reaching clearTimeout(retryTimeoutId), the timeout remains scheduled. Use try/finally so the timeout is always cleared.

						const retryController = new AbortController();
						const retryTimeoutId = setTimeout(() => retryController.abort(), 15000); // 15s timeout

						const retryResponse = await fetch('/ml/docs', {
							method: 'GET',
							signal: retryController.signal
						});

frontend/src/pages/Login.jsx:146

• clearTimeout(timeoutId) is only executed if fetch('/ml/docs', ...) resolves; if fetch throws (abort/network), the timeout stays scheduled. Wrap the fetch in a try/finally so the timer is always cleared.

				const controller = new AbortController();
				const timeoutId = setTimeout(() => controller.abort(), 15000); // 15s timeout for ML server

				const response = await fetch('/ml/docs', {
					method: 'GET',
					signal: controller.signal
				});

frontend/src/pages/Login.jsx:177

• Same timer-cleanup issue in the ML retry probe: if the retry fetch throws before clearTimeout(retryTimeoutId), the timeout remains pending. Use try/finally so the timeout is always cleared.

						const retryController = new AbortController();
						const retryTimeoutId = setTimeout(() => retryController.abort(), 15000); // 15s timeout

						const retryResponse = await fetch('/ml/docs', {
							method: 'GET',
							signal: retryController.signal
						});

frontend/src/pages/Login.jsx:220

• clearTimeout(timeoutId) is only reached when the backend health fetch resolves successfully; if it throws (abort/network error), the timeout remains scheduled. Wrap the fetch in a try/finally so the timer is always cleared.

				const controller = new AbortController();
				const timeoutId = setTimeout(() => controller.abort(), 15000); // 15s timeout

				const response = await fetch(`/health`, {
					method: 'GET',
					signal: controller.signal
				});

frontend/src/pages/Login.jsx:251

• Same timer-cleanup issue in the backend retry probe: if the retry fetch throws before clearTimeout(retryTimeoutId), the timeout remains scheduled. Use try/finally so the timer is always cleared.

						const retryController = new AbortController();
						const retryTimeoutId = setTimeout(() => retryController.abort(), 15000); // 15s timeout

						const retryResponse = await fetch(`/health`, {
							method: 'GET',
							signal: retryController.signal
						});

---

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

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 6 comments.

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated no new comments.

[SumeetDUTTA]

done