Deployment Patterns for Consul in Kubernetes
summary
A live demo builds a Go app with two endpoints and deploys Consul via Consul servers + agents via Helm to show that watch delivery depends on agent topology: running two agents on a three‑node cluster causes HTTP watch callbacks to reach only one replica, scaling the app to five pods still updates only the nodes with local agents (three). Switching agents to a DaemonSet fixes per‑node delivery but still misses cross‑node replicas placed multiple per node; use sidecars for exec handlers and full per‑pod guarantees.
AI-written summary · about
chapters (5)
related talks
22mA Hands-on Approach to Consul Deployment Strategies in a Hybrid Environment
40mIntroducing A New Architecture for Simplified Service Mesh Deployments
19mService Mesh and Your Legacy Apps: Connecting to Kubernetes with Consul
22mConsul Use Cases At Stripe: Service Mesh and More
37mDeploying applications to Kubernetes: never send a human to do a machine's job
43mConsul Learn Lab: Deploy Resilient Applications with Service Mesh and AWS Lambda
44mScaling Microservices with Kubernetes
28mDeploying hybrid topologies with k8s & Envoy: a look at service discovery