Content Ingestion & Podcast Video Incident Report | Spotify Engineering
7/20/2026
This post details a specific incident where video transcoding infrastructure reached maximum capacity, causing a publishing delay. It identifies four contributing factors: insufficient headroom for spikes, a concurrent batch processing job, increased per-item processing cost due to quality improvements, and a software bug underutilizing compute resources. Solutions implemented include stopping the batch job, fixing the bug, adding capacity, and improving monitoring. Future work involves enhancing capacity planning, prioritization, rate limiting, backpressure, and creator notification systems.