Issue 2 • December 23, 2025

Your workflow has been running perfectly for 58 days. Customer orders flowing, data syncing, reports generating. You haven't touched it since deployment.

On day 59, it stops. No error. No warning. Just silence.

You dig into the logs and find it: "Authentication failed." Your OAuth token expired exactly 60 days after you connected it. Every workflow using that connection has been failing for the past 24 hours. You've lost a day of customer orders, sync data, and trust.

This is the Irish Goodbye of your workflow; when your automation disappears without so much as a word. Today, I'm going to show you why OAuth tokens expire silently, how to catch them before they break production, and how to build a health check system that alerts you days in advance.

Fail Pattern: The Sixty-Day-Notice

What Happens:

You connect a service to n8n. Google Sheets, Slack, Salesforce, HubSpot, any OAuth-based API. You authenticate once, test it, deploy it to production. Everything works. Weeks pass. You forget you even built it because it's running perfectly.

Then one morning, every workflow using that connection fails with "401 Unauthorized" or "Authentication failed."

Why It Happens:

OAuth tokens expire. It's a security feature, not a bug. Most services give you a 60 or 90-day token lifespan. When that window closes, the connection dies.

But here's the problem: the connection still appears active in n8n. The credentials panel shows green. No warning. No alert. No countdown timer. Just a silent expiration waiting to break you.

Even worse, some services expire tokens based on inactivity. If you don't use a connection for 30 days, they revoke access. But n8n doesn't know this. It keeps trying to use dead credentials until a workflow execution fails; and by then, you've already lost data.

Which Services Do This:

Almost all OAuth-based services have token expiration:

• Google (Sheets, Drive, Gmail): 7 days for unverified apps, 6 months for verified apps
• Salesforce: 15 minutes (refresh token extends it, but can still expire after 90 days of inactivity)
• Microsoft (Office 365, OneDrive): 90 days
• Slack: Never expires unless manually revoked, but workspace admins can force re-auth
• HubSpot: 6 hours (refresh token extends it indefinitely unless revoked)
• Stripe: Never expires, but can be revoked by user or admin

The problem isn't the expiration. It's that you don't know when it happened until something breaks.

Real-World Disaster:

You built a workflow that syncs Salesforce leads to Google Sheets. It runs daily at 3am. On day 61, the Google OAuth token expires. The workflow fails silently.

Your sales team doesn't check the sheet until Monday morning, three days later. They realize the last three days of leads are missing. They panic. You investigate. You find the authentication error buried in the logs.

You re-authenticate. The workflow starts working again. But those 72 hours of leads? Gone. Salesforce doesn't keep them in a recoverable state because you only sync new leads, not historic data.

Your sales team lost three days of follow-ups. Deals went cold. Competitors got there first. All because an OAuth token expired.

Or worse:

You have a webhook-triggered workflow that processes customer orders. When someone completes checkout, the webhook calls n8n, which writes the order to Google Sheets and sends a confirmation email.

The Google OAuth token expires. The webhook fires, a node tries to write to Sheets, authentication fails, and the entire workflow crashes. The webhook doesn't retry. It delivered the payload successfully as far as it's concerned.

The customer thinks they placed an order. They're waiting for confirmation. It never comes. They email support. Support checks the system. No record of the order. The customer is angry. You look incompetent. All because of an OAuth expiration.

The Pattern:

1. Connect service, authenticate, test, works perfectly

2. Deploy workflow, runs fine for weeks

3. Token expires (day 60, day 90, whenever)

4. Connection dies silently (n8n still shows it as "active")

5. Workflow fails, you don't know until someone complains

6. Data lost, no recovery possible

The Irish Goodbye Metaphor

In social settings, an "Irish Goodbye" is when someone leaves a party without telling anyone. No farewell, no explanation, just gone. One moment they're there, the next they've gone home.

Your OAuth tokens do the same thing.

They work perfectly for weeks. Then they expire, often without warning. No notification. No email. No countdown. They just disappear, and your workflows break in silence.

The worst part? n8n can't know they're dead until it tries to use them.

When you view credentials in n8n, it doesn't test them. It just shows you what's stored. The credential could have expired 30 days ago, and n8n would still display it as "connected." You only discover it's dead when a workflow execution fails.

This isn't n8n's fault. OAuth tokens are opaque strings; there's no way to know if they're valid without testing them against the actual service. And testing every credential every time you open the credentials panel would be slow and expensive (API rate limits).

So the solution isn't to fix n8n. The solution is to proactively monitor your connections and get alerted before they break production.

Health Check Workflow (The Solution)

Here's how you catch these failures before they hurt you:

Build a Connection Health Check workflow that:

1. Runs daily (or hourly for critical systems)

2. Tests every important connection

3. Alerts you if any connection is broken

4. Logs results so you can track degradation over time

How It Works:

Instead of waiting for a production workflow to fail, you create a dedicated workflow that tries to use each connection and reports the results. If the connection works, great. If it fails, you get a Slack alert or email before any customer-facing workflow breaks.

Example: Google Sheets Health Check

// n8n workflow:
// 1. Schedule Trigger (daily at 8am)
// 2. HTTP Request node for each connection to test
// 3. IF node to check for failures
// 4. Slack node to alert on failure

// HTTP Request Node - Test Google Sheets
// Method: GET
// URL: https://sheets.googleapis.com/v4/spreadsheets/YOUR_SHEET_ID
// Authentication: Use your Google Sheets credentials

// If this succeeds, connection is healthy
// If this fails with 401, OAuth token expired

In n8n, this looks like:

Node 1: Schedule Trigger

• Runs: Every day at 8am

Node 2: Google Sheets (Read Row)

• Spreadsheet: Your test sheet (create a dummy sheet just for health checks)
• Sheet: Sheet1
• Range: A1
• This attempts to read one cell. If the connection is dead, this fails.

Node 3: IF Node

• Condition: Check if previous node succeeded
• If failed, go to Slack alert
• If succeeded, go to next connection test

Node 4: Slack Alert

• Message:

Connection Health Check Failed

Connection: Google Sheets (Production Account)
Error: { { $json.error.message } }
Time: { { $now } }

Action Required: Re-authenticate Google Sheets credentials in n8n.

Why This Works:

You catch the authentication failure during a scheduled health check, not during a critical customer workflow. You have time to fix it before production breaks.