Weaviate Diagnostics 🩺

🚧 This diagnostics tool is still in beta, feedback appreciated at customer-success@weaviate.io. 🚧

The purpose of this tool is to collect diagnostics from Weaviate clusters and also warn against any common misconfigurations.

The diagnostics are collected by making a series of requests to a Weaviate instance and then generating a self-contained html report.

Diagnostics are collected for:

• Weaviate Schema, Meta, Module, and Node config
• pprof CPU profile
• Basic Memory / Disk / CPU info
• Prometheus metrics
• Weaviate specific environment variables

Dependencies

Optional recommended dependency is graphviz for the cpu profiling graph.

brew install graphviz # if on mac
apk add graphviz ttf-freefont # if inside weaviate alpine container

Usage

Basic usage for local Weaviate instance:

./weaviate-diagnostics diagnostics -u "http://localhost:8080" -o weaviate-report.html

WCS cluster authenticating using api key

./weaviate-diagnostics diagnostics -a "$WEAVIATE_API_KEY" -u "https://cluster-name.weaviate.cloud" -o weaviate-report.html

Run -h for more options:

./weaviate-diagnostics diagnostics -h
A tool to help diagnose issues with Weaviate

Usage:
  weaviate-diagnostics diagnostics [flags]

Flags:
  -a, --apiKey string       API key authentication
  -h, --help                help for diagnostics
  -m, --metricsUrl string   full URL plus path of the Weaviate metrics endpoint (default "http://localhost:2112/metrics")
  -o, --output string       File to write the report to (default "weaviate-report.html")
  -w, --pass string         Password for OIDC authentication (defaults to prompt)
  -p, --profileUrl string   URL of the Weaviate pprof endpoint (default "http://localhost:6060/debug/pprof/profile?seconds=5")
  -u, --url string          URL of the Weaviate instance (default "http://localhost:8080")
  -n, --user string         Username for OIDC authentication