Deep dives into the tools you run in production.
Tutorials, failure post-mortems, and toolchain breakdowns — each article references specific versions, real configs, and the failure modes worth knowing about.
Kubernetes 1.30: What broke in our staging cluster and how we traced it.
Instructor: Ravi Nair · Senior SRE · 9 min read
A walk through a real node-pressure eviction cascade triggered by the new sidecar container graduation. Includes the kubectl debug commands and the admission webhook patch that resolved it.
Tags: Kubernetes 1.30 · containerd · admission webhooks · node pressure
Current toolchain. Specific configs. No filler.
Pinning AWS provider versions without breaking modules.
Alert fatigue: cutting noise with recording rules.
EC2 Savings Plans vs Reserved Instances in 2024.
How to pre-compute cardinality-heavy queries so your on-call rotation stops seeing false positives at 2 a.m. Runnable PromQL and a recording rule YAML you can drop in.
The pricing model changed again. Here is what the new compute commitment terms mean for mixed-workload accounts and when Spot still beats both options outright.
Version constraint syntax that actually holds across a monorepo. Includes a .terraform.lock.hcl workflow and the one edge case where required_providers silently loses.
Priya Anand · 7 min read
Kiran Desai · 9 min read
Ravi Nair · 6 min read