Digest #233: When IaC can't recover your data, Postgres's 20-minute lookup, Kubernetes PVC tracking and IAM fixes in pull requests

Also inside: migrating production n8n on EKS, splitting AWS commitment discounts fairly, and monitoring Kubernetes with OpenObserve.

Sep 25, 2026

This week, you’ll learn how to find memory savings in your data structures and limit cache churn when rolling out the change. You’ll also find practical steps for scaling GKE workers from queue depth, turning unused IAM permissions into reviewable pull requests, and allocating AWS commitment discounts based on usage. And if your Postgres query uses the right index but still takes 20 minutes, you’ll see how deleted rows can get in the way.

If you’re interested in sponsoring this newsletter, reach out via LinkedIn or reply to this email.

Watch Luca Berton rebuild a production recovery live

Join us on October 7 (Wed) for a live session with Luca Berton, creator of Ansible Pilot and author of the backup and DR guides for Ansible Automation Platform. An automated change with valid permissions could pass every check and still corrupt production data. Re-running the playbook rebuilds the host, and a snapshot rolls back every good write since. Luca rebuilds the recovery for a real incident live, step by step, into a runbook live attendees keep. Bring an incident that broke production and passed every check, and it might be the one he runs - Save your seat