Thanks for reaching out Roman, and I appreciate the detailed context and code snippet.

What you’re describing sounds like a stale or dropped MongoDB connection that isn’t being gracefully handled by your app or by Mongoose’s internal connection pooling. This happens when:

• The MongoDB connection is interrupted (e.g., network blip, TLS renegotiation, idle timeout on the DB side).

• Mongoose doesn’t automatically reconnect or retry on certain classes of disconnections.

• The app doesn’t listen for connection errors and recover gracefully.

I would first try to do the following:

1. Add connection error listeners:

   Add listeners for 'disconnected', 'reconnectFailed', or general 'error' events on the Mongoose connection so your app can log or respond (e.g., attempt a reconnection).

(I generated this for you as an example on what it would look like)

mongoose.connection.on('error', (err) => {
  console.error('Mongo connection error:', err);
});

mongoose.connection.on('disconnected', () => {
  console.warn('MongoDB disconnected! Attempting to reconnect...');
  // Optional: Trigger reconnection logic or alerting here
});

2. Enable Mongoose autoReconnect behavior:

   Although Mongoose does try to reconnect by default, some advanced options are now passed through to the underlying MongoDB Node driver. You may try:

useUnifiedTopology: true, // implicit in newer versions, but good to confirm
autoReconnect: true, // legacy, but worth verifying behavior if using older driver

3. Avoid tlsAllowInvalidCertificates and tlsAllowInvalidHostnames in prod:

   These should only be used in development. Invalid certs/hostnames can sometimes cause renegotiation errors or get flagged by middleboxes or proxies.

Let me know if this helps.

Best,

Angelo

Hi Angelo,

Our application is now crashing because it cannot establish a connection to MongoDB Atlas. We've never had this issue before. Please advise.

Here is the screenshot

Hey Roman. We cannot guarantee a singular, static connection for multiple weeks on end. The reason being for this is we need to cycle our proxy instances for security reasons

You'll need reconnection logic at the application level to get around this. Alternatively, moving whatever workload that's querying your database onto Railway into the private network will guarantee a connection indefinitely.

Happy to help migrate if you need!