A Google Cloud support engineer solves a tough DNS case (cloud.google.com)

Article summary

A Google Cloud support engineer solved a complex DNS case by gathering information, formulating hypotheses, and testing them. The issue was caused by a large receive buffer size set for UDP packets, which led to packet loss. The engineer used various tools and techniques, including raw sockets and kernel configuration, to identify and resolve the issue. The case highlights the importance of detailed troubleshooting and collaboration between support engineers and customers.

Main themes

• DNS troubleshooting
• Google Cloud support
• Raw sockets
• Runbooks and troubleshooting
• Follow-the-sun support
• Customer support quality
• Contract negotiations and discounts

What commenters say

• Raw sockets can bypass flawed code and do not use the same buffering as normal sockets, which can aid in troubleshooting.
• Having a runbook for common issues can speed up the troubleshooting process, but it is not a replacement for skilled engineers.
• Follow-the-sun support may not be effective for complex issues that require continuous attention from a single team or engineer.
• The quality of Google Cloud support varies depending on the customer's spend and contract, with larger customers potentially receiving better support.
• Some customers have had negative experiences with Google Cloud support, including being blamed for issues and receiving inadequate assistance.
• Contract-level commits and dedicated sales teams can provide better support and discounts for large customers.
• The effectiveness of Google Cloud support may depend on the customer's ability to prove that an issue is not their fault, rather than the support team's willingness to help.