Skip to content
OnchainProbe Docs

Add Endpoints

Endpoints are the RPC URLs OnchainProbe actively tests.

Endpoint design basics

Section titled “Endpoint design basics”

Each endpoint belongs to one chain and supports:

  • Name and URL
  • Optional tags for filtering
  • Roles used by comparison suites
  • Optional auth (basic, bearer token, header, or query param)
  • Enabled/disabled state

Add an endpoint in the UI

Section titled “Add an endpoint in the UI”

Once your chain exists, you can start adding endpoints to it.

  1. Open Endpoints (/chains/:chainId/endpoints).
  2. Click Add Endpoint .
  3. Fill:
    • Name
    • URL (HTTP/HTTPS)
    • Tags (optional)
    • Roles (optional; keep defaults initially)
    • Auth (if required by provider)
  4. Save.

Tags

Section titled “Tags”

Tags are comma-separated labels you assign to endpoints. They help you organize, filter, and group endpoints when you have several on the same chain.

What tags are used for

Section titled “What tags are used for”
  • Filtering endpoint lists in the dashboard and API queries.
  • Grouping endpoints by shared characteristics (provider, client software, region, environment).
  • Public dashboards can filter which endpoints appear by tag.

Tags have no effect on test execution or scheduling. They are purely organizational.

Section titled “Recommended conventions”

Pick consistent naming patterns across your team:

  • Provider: alchemy, self-hosted, provider-b
  • Client: geth, nethermind, reth, besu, erigon
  • Environment: prod, staging, canary
  • Region: us-east, eu-west

Common mistakes

Section titled “Common mistakes”
  • Endpoint URLs must include a schema prefix: http://, https://, ws://, or wss://.
  • Using endpoint auth unsupported by your provider secret format
  • Mixing endpoints from different networks (chain ID mismatch is rejected automatically)

Auto-disable behavior

Section titled “Auto-disable behavior”

OnchainProbe protects your test results from permanently broken endpoints. If the block observer sees persistent connection or RPC failures from an endpoint for 24 continuous hours, it auto-disables the endpoint and fires an endpoint_disabled alert.

Recovery works asymmetrically: if the observer detects a successful block fetch during the 24-hour window, the failure timer resets. But once an endpoint is disabled, it stays disabled until you manually re-enable it.

Next steps

Section titled “Next steps”