bosh-dns fails to resolve FQDN associated with many IP addresses in Operations Manager

Products

• Operations Manager

Issue/Introduction

When attempting to resolve a fully qualified domain name (FQDN) from within an app container, it fails with the error message: 

connection timed out; no servers could be reached

The DNS resolution succeeds on first attempt but fails on subsequent tries because of a problem with the cache handler.

The problematic FQDN maps to numerous IP addresses (20+) such that response payload is greater than 512 bytes.

For example:

 $ nslookup fqdn.example.com  

name: fqdn.example.com
address: 10.212.150.1  

name: fqdn.example.com
address: 10.212.150.2
...
name: fqdn.example.com
address: 10.212.150.40  

$ nslookup fqdn.example.com  
connection timed out; no servers could be reached

Environment

Product Version: 2.5

Resolution

This is a known issue with bosh-dns, see this tracker story for more information. The issue is that bosh-dns cache handler does not handle TC/EDNS properly.

The recommended resolution is to upgrade bosh-dns to 1.16 or higher. The latest patched releases of Operations (Ops) Manager 2.6, 2.7, 2.8+ contains the bosh-dns release with the fix for this issue. Upgrading to the latest patch of any of these releases will resolve the issue.

Please reference Ops Manager release notes to identify if you have a bosh-dns version with this problem.