Error handling patterns that actually work in production

Most n8n tutorials skip error handling. That's fine for demos. But when a workflow runs 10,000 times a day, you need it to fail gracefully.

Here's what we've learned from 300+ production workflows.

The problem with default behavior

By default, when a node fails in n8n, the whole workflow stops. That's often not what you want.

Imagine you're processing 500 leads. Lead #47 has a malformed email. Should the other 453 leads get stuck? No.

Pattern 1: Continue on fail

The simplest fix. On any node, enable "Continue On Fail" in settings. The workflow keeps running even if that node errors.

// The failed item gets an error property you can check
if ($input.item.json.error) {
  // Handle the error case
  return { success: false, reason: $input.item.json.error.message };
}

When to use: Processing batches where individual failures shouldn't stop everything.

Pattern 2: Try/catch with IF node

For more control, route errors explicitly:

1. Enable "Continue On Fail" on the risky node
2. Add an IF node after it
3. Check for $json.error to route failures separately

This lets you:

• Log failures to a database
• Send alerts for certain error types
• Retry with different parameters

Pattern 3: Error workflow

n8n has a built-in error workflow feature. When any workflow fails, it triggers a separate workflow.

We use this for:

• Slack notifications to the team
• Logging to our monitoring system
• Auto-creating tickets for critical failures

Set it in Workflow Settings → Error Workflow.

Pattern 4: Graceful degradation

Sometimes the right response to an error is doing something simpler instead of failing.

Example: Your AI summarization fails because the API is overloaded. Instead of failing the whole workflow, fall back to extracting the first 500 characters.

// In a Code node after the AI node
const aiResult = $input.item.json;

if (aiResult.error) {
  // Fallback: just use the first 500 chars
  return {
    summary: $('Previous Node').item.json.content.substring(0, 500) + '...',
    method: 'fallback'
  };
}

return { summary: aiResult.text, method: 'ai' };

What we use on every project

1. Error workflow for monitoring
2. Continue on fail for batch processing
3. Explicit try/catch for critical paths
4. Fallbacks for non-essential features

The goal isn't to prevent errors. It's to handle them without waking anyone up at 3am.