Pipelines and Pizza
Real-world, hands-on experiences in infrastructure and DevOps. From junior fundamentals to senior-level architecture strategies — covering Terraform, Ansible, Azure, Nutanix, Git, and more.
★ Featured
-
First 100 Days of LGTM on Nutanix: What We Built, What We Got Wrong, What's Next
A hundred days into running the LGTM stack on Nutanix — what's actually deployed, the bugs that bit us, the surprises that didn't, and what's still on the roadmap. No fabricated war stories; every fix below is in our changelog.
-
Building an LGTM Observability Stack on Nutanix: Why We Did It and What It Looks Like
Architecture overview of a self-hosted LGTM observability stack (Loki, Grafana, Tempo, Mimir) running on Nutanix — why we built it with zero software budget on recycled hardware, what each component does, and the honest trade-offs five months in.
-
Welcome to Pipelines and Pizza
Kickoff post for my new technical blog — Terraform, Kubernetes, Nutanix, Azure, Ansible, and the journey from junior to senior DevOps engineer.
→ Recent Posts
-
Dashboards That Actually Get Used: 118 Across 18 Folders, Organized by Domain
The dashboard set we maintain — 118 dashboards across 18 domain-organized folders, NetApp Harvest + Pure FlashArray + Nutanix multi-vendor storage section, ConfigMap-sidecar GitOps pattern, and the rules that keep the set from sprawling.
-
Grafana 13 on CloudNativePG: The Real Upgrade Walkthrough
Running Grafana on Kubernetes backed by CloudNativePG instead of SQLite, plus the real Grafana 12.4.2 → 13.0.1 upgrade we ran last month, including the irreversible-moment that makes the pre-upgrade Postgres backup load-bearing.
-
Mimir on Kubernetes: 620K Active Series on Nutanix Objects
Deploying Grafana Mimir in SimpleScalable mode on Kubernetes with Nutanix Objects as the S3 backend — the real config running ~620K active series at 365-day retention, including why three of our Mimir components run as singletons on purpose.
-
Loki in Production: Labels, Per-Stream Retention, and the LogQL Alerts We Run
The production side of Loki — the label set we run, the 14-rule per-stream retention table, the LogQL alerts we actually rely on (audit, syslog, firewall), and the ingestion-rate gotcha that bit us early.