Production Engineer · Phone Screen · Coding

Meta — Production Engineer ❌ Failed

Level: Senior-Level

Round: Phone Screen · Type: Coding · Difficulty: 6/10 · Duration: 60 min · Interviewer: Neutral

Topics: Coding, Troubleshooting

Location: San Francisco Bay Area

Interview date: 2026-01-15

Got offer: False

Summary

I had a phone screen with two coding questions and a troubleshooting question.

Details

I was contacted by a recruiter and had a phone screen. There was no self-introduction at the beginning. I was asked to share my entire screen and disable camera filters, using CoderPad.

The first question was a 'dinosaur' problem that can be found online. The second coding question was to write a small monitoring script. This script receives continuous data output from vmstat and checks if a certain metric exceeds the limit too frequently (vmstat | myscript us 100 50 300). For example:

• Monitor CPU user time (us).
• Anomalies are defined as exceeding a threshold, like 100.
• Maximum allowed exceedances within a time window, like 50.
• Time window duration, like 300 seconds.

The script should monitor the number of exceedances within the last 300 seconds. If the number is too high, print out the anomalous data.

After the coding questions, the interviewer introduced themself, and then asked me to introduce myself.

I was then asked to troubleshoot a web server down issue. The server allows customers to submit orders and receive confirmations. I received an error message at night stating the server was down, without further details. The problem was ongoing and affected all users. There were no recent code changes or traffic changes. I was asked what I would check and was given the results of my checks.

My (incorrect) approach: I checked the application and system logs, finding 500 errors for an API that writes information to the database. The error indicated a failure to write. I checked disk storage and database storage, which were not full. Then I checked the queue, which was also not full. The final answer was that the database transaction log was full, preventing writes. The interviewer kept providing guidance, but I didn't arrive at the correct answer and there were no follow-up questions.