Traffic has doubled, latency is climbing, and your autoscaler is sitting there at 4 replicas doing nothing. You are on the call with the business asking why. What do you do and what do you say?
What they are really testing: Whether you can operate a system that is refusing to help you, hold two threads at once, and communicate to non engineers without either panicking them or hiding the truth.
A real interview question
Traffic has doubled, latency is climbing, and your autoscaler is sitting there at 4 replicas doing nothing. You are on the call with the business asking why. What do you do and what do you say?
What most people say
drag me
“I would check the autoscaler configuration and see why the scaling policy is not triggering.”
It is not wrong, it is just slow, and the interviewer heard you spend the outage reading YAML while the queue built up. A senior scales by hand first and investigates second, and they say something useful to the business while they do it.
The follow-ups they ask next
Pods are Pending. What now?
The application autoscaler did its job and the node layer did not. Look at the cluster autoscaler, instance type availability, and account level quota, and be ready to shift to a different instance family.
The business asks for an ETA. What do you say?
Give a commitment you control: capacity is added, latency should recover in minutes, next update at a named time. Never promise a root cause on a timeline.
What the interviewer is listening for
- Overrides the autoscaler manually within the first minute
- Runs containment and communication in parallel
- Walks the scaling chain in order instead of guessing
- Names maxReplicas, missing metrics and quota as the usual culprits
- Closes the loop with an alert so the same ceiling cannot bite twice
What sinks the answer
- Debugs the autoscaler config while the service degrades
- Promises a root cause or an ETA they cannot control
- Assumes the autoscaler is correct because it is automation
- Never checks whether nodes actually have room for the pods
If you genuinely do not know
Say this instead of freezing. Reasoning out loud from what you do know beats silence every single time, and a good interviewer is listening for exactly that.
“I have not fought a stuck autoscaler live, but here is how I would reason: take the decision away from it and set the replicas myself, tell the business exactly what I did and when I will update, then check the scaling chain from metric to threshold to limit to actual capacity.”
Keep going with troubleshooting
Junior
You get paged at 2am. Checkout latency has gone from 200ms to 4 seconds and it started right after the 11pm deploy. Walk me through what you do.
Junior
It is 3am and you get paged: the primary application server is at 100% disk usage and the app is throwing write errors. You are the only one online. Walk me through exactly what you do.
Junior
You push a new version of a service and within two minutes the pods are all showing CrashLoopBackOff. The team is watching you share your screen. Walk me through exactly what you do.
Junior
A developer pings you: the checkout service cannot reach the payments service, the calls just hang and time out. You have never worked on either service. Walk me through what you do.
Junior
A developer messages you: 'My app suddenly gets Access Denied writing to the S3 bucket. Nothing changed.' It worked yesterday. Walk me through what you do.
Mid
Your load balancer shows 5xx errors jumping from 0.1 percent to 12 percent of requests over ten minutes. There was no deploy. What do you do?
Knowing the answer is not the same as recalling it under pressure
Sign in to send the questions you fumble to spaced recall, so they come back right before you would forget them, and learn the concepts behind them with hands-on labs.
Start free