Application responds with intermittent 404 errors after binding a route service

Products

• VMware Tanzu Application Service for VMs

Issue/Introduction

Summary

When binding or unbinding a route service to an application, the application starts responding with intermittent 404 errors. The application hasn't crashed and is still running. However, the route is continuously being registered and unregistered at a regular interval for 10 minutes. After 10 minutes, the system stabilizes and the system no longer unregisters the route.

Symptoms

• Access to the application results in intermittent unknown route errors (404's) for 10 minutes.
• Checking the status of the application shows it running.
• A route service was recently bound or unbound from this application.

Cause

When a route service is bound or unbound from an application the system believes the route was unregistered. When the system unregisters a route, it will continuously send unregister messages to the Gorouter every 30 seconds for 10 minutes because the path these messages take to the Gorouters is lossy.

At the same time, the system will send register messages for the application instances too. This results in this regular interval of 404's and successful application responses.

Affects

This known issue affects the following TAS for VMs Versions:

• TAS for VMs 2.8.0+
• TAS for VMs 2.7.0+
• TAS for VMs 2.6.5+, released: 09/17/2019
• TAS for VMs 2.5.10+, released: 09/17/2019
• TAS for VMs 2.4.14+, released: 09/17/2019

This may also affect similarly released TAS for VMs for Windows tiles and Isolated Segment tiles as they're running the same Diego Release that this bug was detected in.

Environment

Product Version: 2.4

Resolution

Mitigation

To mitigate the issue from occurring you can wait 10 minutes for the issue to
resolve itself. Or you can restart your application and your route service application after binding or unbinding the route.

Permanent Fix (Not Available Yet)

A permanent fix is currently in progress and will be resolved in a future patch release of the affected product versions.

Reference

This issue is also reported in the OSS CloudFoundry Diego-Release Github issues, see: https://github.com/cloudfoundry/diego-release/issues/500