Harsh

> I still think the AuthError being thrown here is incorrect and should be fixed. That is true and acknowledged. I will look into fixing that in the upcoming releases....

> Is it possible that changes in how AppSync handles connection errors are also causing the spike? I highly doubt. The retrieval of the tokens happens before the connection is...

@MrAsterisco I think as I said in one of my previous messages, `noIdentityPool` in user pool tokens is a bug that we should fix. And I think I missed that...

Hmm.. you might have to enable verbose logging for the affected users and then you will see a lot more data in the logs. The scenario I can think of...

@priva-byte Would you be able to share this implementation that you talked about in your description? > we subsequently implemented custom client-side retry logic If you are trying to reuse...

@priva-byte Also created a small fix to handle the error gracefully. Hopefully you'll be able to resubscribe without any errors. Would you be able to test with and validate if...

@priva-byte Would you be able to share code snippets so that I can try to recreate this locally?

@priva-byte Thanks for the information. I will use your code snippet to try and recreate the issue locally.

@almasios Thanks for providing that information, I will try to see if I can repro the issue with VPN and the switch you provided.

@almasios I have a potential fix made in the attached PR. Would you be able to validate if this fixes your issue? Branch to test: https://github.com/aws-amplify/amplify-swift/tree/subscription-failures (`subscription-failures`)