The feed generation of Python Podcast was kind of slow. Looking deeper into this using sentry, I found out that for each item in the feed, the size of the audio file has to be fetched via HTTP from S3. Ok, this might explain it 😅.

Here's a complete stream of me fixing the issue by caching the audio file sizes in a JSONField including the deployment to python-podcast.de. The stream is in german.

A few hours later I checked the slow HTTP-operations graph from sentry. Caching seems to work: