Error Handling & Recovery Patterns

Documentation

This skill has comprehensive documentation:

•Production Examples - Real-world code examples from the codebase
•API Reference - Complete API documentation with official links
•Implementation Patterns - Best practices and anti-patterns

Core Principles

Use ErrorBoundary components to isolate errors and prevent app crashes. Integrate OpenTelemetry spans to track errors in distributed tracing and provide user-friendly fallback UI with recovery actions.

Why: Proper error handling prevents entire app crashes, improves user experience with actionable error messages, and enables debugging production issues through distributed tracing.

When to Use This Skill

Use these patterns when:

•Wrapping feature components to isolate errors
•Integrating error tracking with OpenTelemetry
•Creating fallback UI for error states
•Handling GraphQL query/mutation errors
•Implementing retry logic for transient failures
•Testing error boundaries and error handling paths
•Tracking errors in production with distributed tracing

ErrorBoundary Component

Component-Level Boundaries

Wrap features with ErrorBoundary to isolate errors and prevent app-wide crashes:

typescript

import { ErrorBoundary } from '@libs/error-boundary';

export const RecipeListScreen = () => {
  return (
    <ErrorBoundary
      scope={{
        moduleName: 'RecipeList',
        attributes: {
          squad: 'cookbook',
          feature: 'recipe-list',
        },
      }}
      fallback={(error, componentStack, resetError) => (
        <ErrorFallback error={error} onRetry={resetError} />
      )}
    >
      <RecipeListContent />
    </ErrorBoundary>
  );
};

Key patterns:

•scope.moduleName: Identifies where error occurred (required)
•scope.attributes: Additional context for debugging (optional)
•fallback: Custom error UI with retry functionality
•Automatic OpenTelemetry span creation with recordException()

Why: Component-level boundaries prevent one failing component from crashing the entire app. Errors are isolated, logged to OpenTelemetry, and users see helpful recovery UI.

Production Example: git-resources/shared-mobile-modules/src/modules/comms/screens/inbox/InboxScreen.tsx:1

Screen-Level Boundaries

Wrap entire app or navigation stacks with top-level ErrorBoundary:

typescript

import { ErrorBoundary } from '@libs/error-boundary';

export const ScreenEntryProvider: React.FC<PropsWithChildren> = ({ children }) => {
  return (
    <QueryClientProvider client={queryClient}>
      <ApolloProvider client={apolloClient}>
        <SafeAreaProvider>
          <AppWithTranslation>
            <ZestProvider>
              <ErrorBoundary scope={{ moduleName: 'App' }}>
                {children}
              </ErrorBoundary>
            </ZestProvider>
          </AppWithTranslation>
        </SafeAreaProvider>
      </ApolloProvider>
    </QueryClientProvider>
  );
};

Why: Top-level boundaries catch errors that escape component boundaries, ensuring the app never shows a blank white screen.

Production Example: git-resources/shared-mobile-modules/src/entry-providers/providers.tsx:140

Error Scope Configuration

Always provide scope information for debugging:

typescript

<ErrorBoundary
  scope={{
    moduleName: 'Checkout',
    attributes: {
      screenName: 'PaymentScreen',
      feature: 'payment-processing',
      userId: user.id,
      cartId: cart.id,
      squad: 'conversions',
    },
  }}
  onError={(error, componentStack) => {
    // Optional: Custom error handling
    logError(error, { moduleName: 'Checkout', componentStack });
  }}
>
  <PaymentForm />
</ErrorBoundary>

Key patterns:

•moduleName: Required identifier (e.g., 'Checkout', 'RecipeList')
•attributes: OpenTelemetry attributes for filtering/grouping errors
•onError: Optional callback for custom error handling

Why: Rich scope data helps identify error sources in production logs and enables filtering errors by team, feature, or screen.

Production Example: git-resources/shared-mobile-modules/src/libs/error-boundary/ErrorBoundary.tsx:76

OpenTelemetry Integration

Span Recording with Exceptions

Always use try-catch-finally pattern with spans:

typescript

import { useTracer } from '@libs/observability';
import { SpanStatusCode } from '@opentelemetry/api';

export const useCheckoutFlow = () => {
  const { startSpan } = useTracer();

  const processPayment = async (paymentData: PaymentData) => {
    const span = startSpan('checkout.processPayment', {
      attributes: {
        'payment.method': paymentData.method,
        'cart.total': paymentData.total,
        'payment.currency': paymentData.currency,
      },
    });

    try {
      const result = await paymentService.process(paymentData);
      span.setStatus({ code: SpanStatusCode.OK });
      span.setAttributes({
        'payment.transaction_id': result.transactionId,
        'payment.status': 'success',
      });
      return result;
    } catch (error) {
      span.setStatus({
        code: SpanStatusCode.ERROR,
        message: (error as Error).message,
      });
      span.recordException(error as Error);
      span.setAttributes({
        'payment.status': 'failed',
        'error.type': (error as Error).name,
      });
      throw error; // Re-throw for caller to handle
    } finally {
      span.end(); // ALWAYS end span
    }
  };

  return { processPayment };
};

Key patterns:

•startSpan() with descriptive name (checkout.processPayment)
•setStatus() with OK or ERROR
•recordException() for detailed error tracking
•span.end() in finally block (CRITICAL)
•Attributes for request context and results

Why: Distributed tracing enables debugging complex async flows in production. Always ending spans in finally blocks prevents memory leaks.

Production Example: git-resources/shared-mobile-modules/src/libs/networking-client/client/useFetch.ts:89

Automatic GraphQL Tracing

GraphQL operations automatically create spans via tracingLink:

typescript

// Automatic instrumentation in Apollo Client setup
import { tracingLink } from '@libs/graphql/links/tracing';

const apolloClient = new ApolloClient({
  link: ApolloLink.from([
    tracingLink, // Automatically creates spans for all GraphQL operations
    httpLink,
  ]),
  cache: inMemoryCache,
});

How tracingLink works:

•Creates span for each GraphQL operation: QUERY GetRecipes, MUTATION CreateOrder
•Adds operation type, name, variables (sanitized)
•Records GraphQL errors with recordException()
•Sets span attributes: response status, error count, error messages
•Automatically ends span on completion

Why: Consistent telemetry for all GraphQL operations without manual instrumentation. Errors are automatically tracked in OpenTelemetry.

Production Example: git-resources/shared-mobile-modules/src/libs/graphql/links/tracing.ts:115

Fallback UI Patterns

User-Friendly Error Messages

Provide clear, actionable error messages based on error type:

typescript

export const ErrorFallback = ({ error, onRetry }: ErrorFallbackProps) => {
  const styles = useZestStyles(stylesConfig);

  // Network errors
  if (error.networkError || error.message.includes('Network request failed')) {
    return (
      <View style={styles.container}>
        <Icon icon="WifiOff24" />
        <Text type="headline-md">No Internet Connection</Text>
        <Text type="body-md">Please check your connection and try again.</Text>
        <Button onPress={onRetry}>Retry</Button>
      </View>
    );
  }

  // Authentication errors
  if (error.message.includes('unauthorized') || error.message.includes('401')) {
    return (
      <View style={styles.container}>
        <Icon icon="LockOutline24" />
        <Text type="headline-md">Session Expired</Text>
        <Text type="body-md">Your session has expired. Please sign in again.</Text>
        <Button onPress={handleReauth}>Sign In Again</Button>
      </View>
    );
  }

  // Generic errors
  return (
    <View style={styles.container}>
      <Icon icon="ErrorOutline24" />
      <Text type="headline-md">Something Went Wrong</Text>
      <Text type="body-md">We encountered an unexpected error.</Text>
      <Button onPress={onRetry}>Try Again</Button>
    </View>
  );
};

Key patterns:

•Detect error type from error properties or message
•Show appropriate icon and message for each error type
•Always provide action (Retry, Sign In Again)
•Never show technical stack traces to users

Why: User-friendly messages improve UX and guide users toward recovery actions. Technical errors confuse non-technical users.

Development vs Production Fallbacks

Show different UI in development vs production:

typescript

export const DefaultErrorFallback = ({
  error,
  errorInfo,
  resetError,
}: DefaultErrorFallbackProps) => {
  if (__DEV__) {
    return (
      <DevErrorFallback
        error={error}
        componentStack={errorInfo?.componentStack ?? ''}
        resetError={resetError}
      />
    );
  }

  return (
    <ProductionErrorFallback scope="default-error" resetError={resetError} />
  );
};

Development fallback:

•Shows full error message and stack trace
•Displays component stack for debugging
•Helpful for developers during development

Production fallback:

•Shows generic user-friendly message
•Hides technical details
•Provides retry action

Why: Developers need detailed error info for debugging. Users need simple, actionable messages without technical jargon.

Production Example: git-resources/shared-mobile-modules/src/libs/error-boundary/fallback-ui/DefaultErrorFallback.tsx:1

GraphQL Error Handling

Error Policy Configuration

Use errorPolicy to control how GraphQL errors are handled:

typescript

export const useGetProductDetails = (productId: string) => {
  const { data, loading, error } = useQuery(GetProductDetailsDocument, {
    variables: { productId },
    errorPolicy: 'all', // Return partial data + errors
  });

  const hasErrors = error?.graphQLErrors.length > 0;
  const hasNetworkError = error?.networkError != null;

  return {
    product: data?.product || null,
    loading: loading && !data,
    hasErrors,
    hasNetworkError,
  };
};

ErrorPolicy options:

•none (default): Throw on any GraphQL error, no data returned
•ignore: Ignore errors entirely, return data only
•all: Return partial data + errors (RECOMMENDED)

Why: errorPolicy: 'all' enables graceful degradation. Show partial data when available instead of blank error screen.

Production Example: git-resources/shared-mobile-modules/src/data-access/graphql/product-details/queries.ts:26

Error Type Discrimination

Handle different GraphQL error types appropriately:

typescript

export const handleGraphQLError = (error: ApolloError) => {
  // Authentication errors (401)
  const authError = error.graphQLErrors.find(
    (err) => err.extensions?.code === 'UNAUTHENTICATED'
  );
  if (authError) {
    return {
      type: 'auth' as const,
      message: 'Please sign in again',
      action: 'reauth',
    };
  }

  // Validation errors (400)
  const validationError = error.graphQLErrors.find(
    (err) => err.extensions?.code === 'BAD_USER_INPUT'
  );
  if (validationError) {
    return {
      type: 'validation' as const,
      message: validationError.message,
      fields: validationError.extensions?.fields as Record<string, string>,
    };
  }

  // Network errors (no response)
  if (error.networkError) {
    return {
      type: 'network' as const,
      message: 'Connection failed. Please check your internet.',
      action: 'retry',
    };
  }

  // Generic errors
  return {
    type: 'unknown' as const,
    message: 'Something went wrong. Please try again.',
    action: 'retry',
  };
};

Key error types:

•UNAUTHENTICATED: Session expired, need re-auth
•BAD_USER_INPUT: Validation errors with field details
•networkError: Connection failures
•Generic: Unknown server errors

Why: Specific error handling enables appropriate user feedback and recovery actions.

Retry Strategies

TanStack Query Retry Configuration

Configure automatic retry for transient errors:

typescript

export const useLoadRecipes = () => {
  const { data, error, refetch, isRefetching } = useQuery({
    queryKey: ['recipes'],
    queryFn: fetchRecipes,
    retry: 3, // Retry 3 times
    retryDelay: (attemptIndex) => Math.min(1000 * 2 ** attemptIndex, 30000), // Exponential backoff
  });

  return {
    recipes: data ?? [],
    error,
    refetch,
    isRefetching,
  };
};

Retry patterns:

•retry: 3: Retry up to 3 times
•retryDelay: Exponential backoff (1s, 2s, 4s, max 30s)
•Automatic for temporary network failures
•No retry for 4xx errors (client errors)

Why: Automatic retry handles temporary network issues without user intervention. Exponential backoff prevents overwhelming servers.

Manual Retry Hook

Provide manual retry for user-initiated actions:

typescript

export const useManualRetry = (onRetry: () => Promise<void>) => {
  const [isRetrying, setIsRetrying] = useState(false);
  const [retryCount, setRetryCount] = useState(0);

  const retry = useCallback(async () => {
    setIsRetrying(true);
    setRetryCount((prev) => prev + 1);

    try {
      await onRetry();
    } catch (error) {
      console.error('Retry failed:', error);
    } finally {
      setIsRetrying(false);
    }
  }, [onRetry]);

  return { retry, isRetrying, retryCount };
};

// Usage
const ErrorScreen = ({ error, refetch }) => {
  const { retry, isRetrying } = useManualRetry(refetch);

  return (
    <View>
      <Text>Error: {error.message}</Text>
      <Button onPress={retry} loading={isRetrying}>
        Retry
      </Button>
    </View>
  );
};

Why: Users can manually retry after fixing issues (e.g., enabling wifi). Visual loading state shows retry in progress.

Error Boundary Lifecycle

Lifecycle Callbacks

Use lifecycle callbacks for custom error handling:

typescript

<ErrorBoundary
  scope={{ moduleName: 'Checkout' }}
  onMount={() => {
    // Component mounted successfully
    console.log('Checkout error boundary mounted');
  }}
  beforeCapture={(scope, error, componentStack) => {
    // Before error is captured
    console.log('About to capture error in:', scope.moduleName);
  }}
  onError={(error, componentStack) => {
    // After error is captured
    logErrorToAnalytics(error, { moduleName: 'Checkout' });
  }}
  onReset={(error, componentStack) => {
    // After error is reset (before app restart)
    console.log('Resetting error boundary');
  }}
  onUnmount={(error, componentStack) => {
    // Component unmounting
    if (error) {
      console.log('Unmounting with active error');
    }
  }}
>
  <CheckoutFlow />
</ErrorBoundary>

Lifecycle order:

•onMount() - Component mounted
•Error occurs in children
•beforeCapture() - Before capturing error
•Error state set
•onError() - After error captured
•User clicks retry
•onReset() - Before app restart
•RNRestart.restart() - App restarts

Why: Lifecycle callbacks enable analytics tracking, custom logging, and cleanup before app restart.

Production Example: git-resources/shared-mobile-modules/src/libs/error-boundary/ErrorBoundary.tsx:60

Reset Error Strategy

ErrorBoundary provides resetError that restarts the app:

typescript

const resetError = () => {
  // Reset error state
  this.setState({
    error: null,
    componentStack: null,
  });

  // Call onReset callback
  if (onReset) {
    onReset(error, componentStack);
  }

  // Reload JS Bundle
  RNRestart.restart();
};

Why: RNRestart.restart() ensures a clean slate after errors. State is cleared and app reloads.

Production Example: git-resources/shared-mobile-modules/src/libs/error-boundary/ErrorBoundary.tsx:130

Testing Error Handling

Test Error Boundaries

Verify ErrorBoundary catches errors and shows fallback:

typescript

import { render, screen } from '@testing-library/react-native';
import { useEffect } from 'react';

const ThrowError = ({ shouldThrow = true }) => {
  useEffect(() => {
    if (shouldThrow) {
      throw new Error('Test error');
    }
  }, [shouldThrow]);

  return null;
};

describe('ErrorBoundary', () => {
  it('renders fallback when error occurs', () => {
    render(
      <ErrorBoundary
        scope={{ moduleName: 'Test' }}
        fallback={(error) => <Text>Error: {error.message}</Text>}
      >
        <ThrowError />
      </ErrorBoundary>
    );

    expect(screen.getByText('Error: Test error')).toBeTruthy();
  });

  it('calls onError callback', () => {
    const onError = jest.fn();

    render(
      <ErrorBoundary scope={{ moduleName: 'Test' }} onError={onError}>
        <ThrowError />
      </ErrorBoundary>
    );

    expect(onError).toHaveBeenCalledWith(
      expect.objectContaining({ message: 'Test error' }),
      expect.any(String) // componentStack
    );
  });
});

Key patterns:

•Use useEffect to throw errors (componentDidCatch catches lifecycle errors)
•Verify fallback UI is rendered
•Verify onError callback is called with error and componentStack

Why: Testing ensures error boundaries work correctly and capture all error details.

Production Example: git-resources/shared-mobile-modules/src/libs/error-boundary/ErrorBoundary.test.tsx:1

Test OpenTelemetry Spans

Verify spans are created with correct attributes:

typescript

import { mockTracerProvider } from 'jest-utils';

describe('ErrorBoundary OTEL Tracing', () => {
  const mockSpanExporter = mockTracerProvider();

  beforeEach(() => {
    mockSpanExporter.reset();
  });

  it('creates span when error occurs', () => {
    render(
      <ErrorBoundary
        scope={{
          moduleName: 'TestModule',
          attributes: {
            feature: 'checkout',
            userId: '123',
          },
        }}
      >
        <ThrowError />
      </ErrorBoundary>
    );

    const spans = mockSpanExporter.getFinishedSpans();
    expect(spans.length).toBe(1);

    const span = spans[0];
    expect(span.name).toBe('ErrorBoundary TestModule');
    expect(span.attributes).toEqual(
      expect.objectContaining({
        'module.name': 'TestModule',
        feature: 'checkout',
        userId: '123',
        componentStack: expect.any(String),
      })
    );
  });
});

Why: Testing OTEL integration ensures errors are properly tracked in distributed tracing.

Production Example: git-resources/shared-mobile-modules/src/libs/error-boundary/ErrorBoundary.test.tsx:92

Test Retry Logic

Verify retry behavior with mock failures:

typescript

test('retries failed queries with exponential backoff', async () => {
  const mockFetch = jest
    .fn()
    .mockRejectedValueOnce(new Error('Network error'))
    .mockRejectedValueOnce(new Error('Network error'))
    .mockResolvedValueOnce({ data: mockRecipes });

  const { result } = renderHook(() =>
    useQuery({
      queryKey: ['recipes'],
      queryFn: mockFetch,
      retry: 3,
      retryDelay: (attemptIndex) => Math.min(1000 * 2 ** attemptIndex, 30000),
    })
  );

  await waitFor(() => expect(result.current.isSuccess).toBe(true));
  expect(mockFetch).toHaveBeenCalledTimes(3); // 2 failures + 1 success
  expect(result.current.data).toEqual({ data: mockRecipes });
});

Why: Testing retry logic prevents infinite loops and validates backoff strategy.

Common Mistakes to Avoid

❌ Don't forget to end spans:

typescript

// ❌ Missing span.end() - memory leak
const span = tracer.startSpan('operation');
await doSomething();
// Forgot span.end()

// ✅ Always end span in finally block
const span = tracer.startSpan('operation');
try {
  await doSomething();
} catch (error) {
  span.recordException(error);
  throw error;
} finally {
  span.end(); // CRITICAL
}

❌ Don't show technical errors to users:

typescript

// ❌ Technical stack trace confuses users
<Text>{error.stack}</Text>

// ✅ User-friendly message
<Text>Something went wrong. Please try again.</Text>

❌ Don't swallow errors without logging:

typescript

// ❌ Silent failure - no logging
try {
  await fetchData();
} catch (error) {
  // Silent failure
}

// ✅ Log errors with context
try {
  await fetchData();
} catch (error) {
  logError(error, { moduleName: 'DataFetcher', operation: 'fetchData' });
  throw error; // Re-throw if caller needs to handle
}

❌ Don't use errorPolicy: 'ignore' blindly:

typescript

// ❌ Ignoring all errors silently
useQuery(GetDataDocument, {
  errorPolicy: 'ignore', // Errors disappear
});

// ✅ Use 'all' to get partial data + errors
useQuery(GetDataDocument, {
  errorPolicy: 'all', // Partial data + errors
});

✅ Do provide scope context for errors:

typescript

// ✅ Rich context for debugging
<ErrorBoundary
  scope={{
    moduleName: 'Checkout',
    attributes: {
      feature: 'payment',
      squad: 'conversions',
      userId: user.id,
    },
  }}
>
  <CheckoutFlow />
</ErrorBoundary>

✅ Do use component-level boundaries:

typescript

// ✅ Isolate errors per feature
<ErrorBoundary scope={{ moduleName: 'RecipeList' }}>
  <RecipeList />
</ErrorBoundary>

<ErrorBoundary scope={{ moduleName: 'Cart' }}>
  <Cart />
</ErrorBoundary>

✅ Do provide retry actions:

typescript

// ✅ User can retry after fixing issue
<ErrorFallback
  error={error}
  onRetry={() => {
    refetch();
  }}
/>

Performance Considerations

Avoid Creating Too Many Spans

Don't create spans for every small operation:

typescript

// ❌ Too many spans - performance overhead
items.forEach((item) => {
  const span = tracer.startSpan(`process-${item.id}`);
  processItem(item);
  span.end();
});

// ✅ Single span for batch operation
const span = tracer.startSpan('process-items', {
  attributes: {
    'items.count': items.length,
  },
});
try {
  items.forEach(processItem);
  span.setStatus({ code: SpanStatusCode.OK });
} finally {
  span.end();
}

Why: Too many spans impact performance and create noise in traces. Group related operations.

Quick Reference

ErrorBoundary Pattern:

typescript

<ErrorBoundary
  scope={{
    moduleName: 'FeatureName',
    attributes: { squad: 'team', feature: 'name' },
  }}
  fallback={(error, componentStack, resetError) => (
    <ErrorFallback error={error} onRetry={resetError} />
  )}
  onError={(error, componentStack) => {
    logError(error, { moduleName: 'FeatureName' });
  }}
>
  <FeatureComponent />
</ErrorBoundary>

OpenTelemetry Span Pattern:

typescript

const span = startSpan('operation.name', { attributes: { /* context */ } });
try {
  const result = await operation();
  span.setStatus({ code: SpanStatusCode.OK });
  return result;
} catch (error) {
  span.recordException(error);
  span.setStatus({ code: SpanStatusCode.ERROR, message: error.message });
  throw error;
} finally {
  span.end(); // ALWAYS
}

GraphQL Error Handling:

typescript

useQuery(Document, {
  errorPolicy: 'all', // Partial data + errors
});

Retry Configuration:

typescript

useQuery({
  queryKey: ['data'],
  queryFn: fetchData,
  retry: 3,
  retryDelay: (attemptIndex) => Math.min(1000 * 2 ** attemptIndex, 30000),
});

Key Libraries:

•React Native 0.75.4
•@opentelemetry/api 2.0.1
•@apollo/client 3.13.6
•@tanstack/react-query 5.59.16
•react-native-restart (for ErrorBoundary reset)

For production examples, see references/examples.md.