Help: Just rolled my first 3‑stage CI/CD on GKE and love how sushi vibes echo my process 🍣

I’ve set up a Jenkins 2.378 on a B450‑Tomahawk chassis, plugged in three Docker‑hosted micro‑services, and crunched them all on GKE‑EKS‑Free! My CI stack looks like this:

• Stage 1: Source check‑out – J‑Hash version tagging from GitHub via Jenkins.
• Stage 2: Build – Docker compose build – 30 sec, a.k.a sushi roll prep.
• Stage 3: Deploy – Helm chart rollout on GKE (k8s v1.27, nodes N1–P400) with autoscaling.

The result? Zero rollback in prod after first go live! Each micro‑service consumes <200 mlCPU and <512 MiB RAM when running – efficient like sashimi slices.

What I really wanna circle back over is the rolling update strategy. My team originally rolled with 30% of pods then 70%, but I tweaked it to 50% and it shaved the wait on sticky sessions down to almost nothing – feels like marinating in soy sauce, instant. Any one with a similar lightweight Kubernetes roll plan (no operators, pure Helm) to recommend? Or tricks to keep pod logs pristine but not swamp Jenkins?

TL;DR: Made 3‑stage pipeline using Jenkins‑2.378, Docker build ‑o, Kubernetes Helm 3.12, autoscaling N1‑P400 nodes – stack performing like sushi without over‑fished resources. Happy to share detailed config or hop on a call if you’re curious.

Just wanted to toss this out there to get feedback and maybe swap ideas with any sushi‑devops aficionados out there.