Argo and Tekton: Pushing the Boundaries of the Possible on Kubernetes
Talkat KubeCon + CloudNativeCon North America 2021 · on Thursday, 2021-10-14· at 11:55· Room 403 AB + Online·
summary
Argo Workflows and Tekton run workflow engines on Kubernetes and show concrete workarounds and limits: they use entrypoint rewriting (emissary/commandlet) to control container lifecycle and sidecars, share data with emptyDir, HTTP or unix domain sockets for fast IPC, and hit scale with 20–30k pod graphs. Step results use /dev/termination-log 4KiB limit so both projects write results to configmaps or external DBs; they document CRD proliferation, API-server load, and propose container start/stop APIs and status offload.
AI-written summary · about
chapters (16)
- 0:00Intro
- 0:47What do we do?
- 1:21Why build on Kubernetes?
- 3:52For Example
- 4:49Container Lifecycle
- 6:11Ordering Container Shutdown
- 7:13Entrypoint Rewriting
- 7:49Ordering Container Start-Up
- 12:31Inject Sidecars: Istio or Vault
- 16:41Other ways?
- 17:40Cross-Pod Communication
- 18:23termination Message
- 21:00Write to ConfigMap / other CRD
- 22:37Custom Resource Proliferation
- 24:09Mitigations
- 27:55What can we do?
related talks
06mComparing Argo Workflows and Apache Airflow at Scale
25mOne Engine To Rule Them All: Unifying Cloud Workloads With Argo Workflows
39mBuild and Deploy Your Application with Argo on Kubernetes
27mThe Life of an Argo Workflows Salesperson
27mArgoCD as the Engine of a Brownfield Migration to Kubernetes
38mTekton and ArgoCD: The Dynamic Duo for Kube Native CI/CD
52mCloud-Native Kubernetes Workflows on AKS with Argo
52mCloud-Native Kubernetes Workflows on AKS with Argo