Every Tuesday - Deep dives, architecture lessons, and real engineering stories.
Every Saturday - High Value - the best DevOps, SRE, Cloud, AI, tools, tutorials, and projects from the week.
SRE/DevOps Interview Prep Guide
Welcome to this week's Uptime Sync. We start with wfx-loot, a campaign that used stolen GitHub credentials to push a CI secret stealer into tens of thousands of repositories, and with how GitHub rebuilt its Git storage to handle 7,200 pushes a second. Then Sentry explains how a few broken release names overloaded its database, and we look at why one empty resources: {} block makes Kubernetes 1.34 to 1.36 pods BestEffort. We also cover the GCP europe-west2-b outage, how Cloudflare halved DNS build delay and cut memory use by 85%, Uber's Shard Manager, how Yelp moved its Cassandra fleet between Kubernetes clusters without copying data, how Atlassian scaled Jira to about 6 million versions, and a large-transaction binary log optimization AWS contributed to MySQL.
On the tutorial front: raise and verify file-descriptor limits for a self-hosted LLM gateway, and triage ImagePullBackOff from kubelet events to a fix. We also cover checking PostgreSQL indexes and constraints after a glibc upgrade, a default-deny baseline with GKE ClusterNetworkPolicy, sizing Java container memory for heap plus non-heap, and DNS failover between AWS and Azure with Route 53 and Terraform. In the videos, a retired Microsoft engineer explains what broke in the Outlook outage, Hello Interview shares a framework for any system design interview, and Hussein explains why you are not replaceable.
This week's projects include Restic for backing up and restoring a working directory, and Kumo for running AWS SDK integration tests locally without AWS credentials. Ggshield scans your repository for leaked secrets, Coolify hosts apps and their databases on your own server without a managed PaaS, and Llmgateway puts one OpenAI-compatible endpoint in front of several LLM providers.
Newsworthy Reads
Tutorials of the Week
Projects of the Week
Restic - Initialize a local restic repository, back up a working directory, then restore a file from the snapshot to verify your backup and recovery path.
Kumo - Point an AWS SDK v2 integration test suite at a local kumo container on port 4566 so S3, SQS or DynamoDB tests run in CI without AWS credentials or cloud cost.
Ggshield - Run ggshield secret scan repo . to scan your repository for leaked secrets.
Coolify - Install it on a spare Linux server with the documented one-line script to host a static site or containerized app and its database without a managed PaaS.
Llmgateway - Deploy the unified Docker image on your own infrastructure to put one OpenAI-compatible endpoint in front of several LLM providers, then review per-model token usage and cost data.
👋 Find me on Twitter | Linkedin | Instagram | Connect 1:1
Thank you for supporting this newsletter. Consider sharing this post with your friends.
Y’all are the best.
