Windows container shows 502 intermittent error when accessing an app

Products

• VMware Tanzu Application Service for VMs

Issue/Introduction

Symptoms:

• Access to the application results in a 502 in the Gorouter logs
• Direct access to the app’s external IP and port fails to establish a connection
• Access to the app inside its container works fine
• Restarting the application temporarily resolves the issue

Environment

Cause

A race condition where one application is deployed to a Cell and before it receives any traffic a second application is deployed to the Cell. The routing rules for the first application are mistakenly removed. This manifests as 502 response codes on requests from the Gorouter, but the app is still running inside its container.

Resolution

The resolution is to apply Microsoft’s Windows Server 2016 cumulative update KB4074588 (Febuary 2018) or a more recent cumulative update to the Windows stemcell and deploy the updated stemcell.  Once the updated stemcell has been deployed to your Windows Cells, the error will be resolved and applications should no longer see the symptoms above.