#file systems

8 file systems interview questions, asked at Databricks, Uber, Amazon and 3 more companies. Last asked Sep 2026.

Design a Kafka-based file watcher for millions of filesSystem designDesign a file watcher that tracks millions of files and publishes change events to downstream consumers through Kafka. Cover the first full discovery of files, ongoing changes…
Last asked Sep 2026
Design the metadata service of a strongly consistent distributed file systemSystem designDesign the namespace layer for a distributed file system: make directories, list them, put and get files, delete, and walk the tree recursively. A listing must reflect every write…
Last asked Sep 2026
Design a distributed file system for large structured datasetsSystem designDesign a distributed file system or storage service for large structured datasets that data-processing jobs read and write. Go past the box diagram and walk through the concrete…
Last asked Sep 2026
Compute entity sizes and validate a hierarchical file systemCodingFiles and folders form a hierarchy. Compute the size of any entity, where a folder's size is the total of everything inside it, and validate the hierarchy: check that it really is…
Last asked Aug 2026
List every file path in a directory treeCodingGiven a directory tree (folders containing files and other folders), return the full path of every file. Write it with a depth-first search, then discuss an iterative version and…
Last asked Mar 2026
Design a crash-resilient file systemSystem designDesign a file system that can recover file contents correctly after a crash at any moment. Cover write ordering, journaling or copy-on-write, what is guaranteed after fsync, the…
Last asked Dec 2025
Print a directory tree with indentationCodingGiven a root directory, print it and everything under it, one entry per line, with each entry indented one step more than its parent directory.
Last asked Oct 2025
Concurrent caching client for byte-range file readsSystem designDesign a client that reads byte ranges of remote files and caches what it has fetched, so a later read of a range it already holds is served locally. Many threads use one client…
Last asked Aug 2025