For the complete documentation index, see llms.txt. This page is also available as Markdown.

Scaling

Configuration to scale Pipekit beyond defaults

The pages here cover configuration you should turn on once Pipekit's defaults stop holding up. The defaults are tuned for small to moderate deployments; heavy concurrent throughput or very large workflows will push past them.

When to read these pages

  • Vector Log Collection. Read this if you are losing logs, or if log streaming is putting load on your Kubernetes API server.

  • Node Status Offloading. Read this if you are running workflows large enough that Argo Workflows itself has node status offloading enabled.

For region-level continuity across separate clusters, see Disaster Recovery.

Last updated