DevOps Interview DropAdvanced TierScenario+50 XP on read
What metrics do you check before SSHing into an instance?
#Observability#Troubleshooting#Performance#SRE
Core Summary
The rule is simple: never SSH into an instance as your first troubleshooting step. Check metrics first. You should only go to the shell once you have narrowed the problem to a specific instance and you need to inspect logs, configuration, or running processes. The metrics you check are: CPU, memory, disk, network I/O, context switches, and running process list.
Hints
Hint 1: Metrics give you the fleet view; SSH gives you one instance
Hint 2: If the problem is one pod, find which one via metrics
Hint 3: You are looking for resource exhaustion or anomalies
Reported in interviews at Meta, Netflix, Amazon