Troubleshoot CPU, latency, and DNS issues
Company: Google
Role: Software Engineer
Category: Coding & Algorithms
Difficulty: medium
Interview Round: Onsite
A) A Linux server’s CPU usage suddenly spikes. What immediate checks and commands would you run (e.g., top/htop, pidstat, perf, iostat), how would you isolate the hot processes/threads, and how would you determine whether the bottleneck is user CPU, kernel CPU, or IO wait? Describe next steps and remediation. B) A service shows high latency. Explain how you would use netstat/ss, tcpdump, and traceroute/mtr to localize the issue (client, server, network), including which signals to capture (SYN backlog, retransmits, RTT, drops) and how to interpret them. C) Walk through the DNS resolution flow (stub resolver, recursive, authoritative, caching, TTLs) and explain the practical differences between dig and nslookup, including when you’d prefer each and sample commands for debugging.
Overview: This question evaluates system troubleshooting and diagnostic competencies, including Linux CPU and IO performance analysis, network latency and packet-level investigation, and DNS resolution and query tooling knowledge.
Read the full Google Software Engineer interview experience this question came from