probe-endpoints.yml

# Daily health probe over every RPC endpoint in `pet-chain-endpoints.json`.
# Commits the new ranking directly to master if it differs from the current
# order (same convention as `update-known-good.yml`).
#
# This workflow stacks three behaviors to deliver a "use the healthiest
# endpoint" policy without any per-test or per-job change:
#
# 1. scripts/probe-endpoints.mjs probes each endpoint over a persistent
#    WebSocket with sustained traffic (matching production traffic shape).
# 2. scripts/apply-probe-ranking.mjs rewrites `pet-chain-endpoints.json`
#    so that each chain's array is sorted by score (best first).
# 3. Subway (when used, via `.github/subway-template.yml`) is configured
#    with `shuffle_endpoints: false`, so it picks array[0] at startup.
#    Chopsticks (when used directly, via @polkadot/rpc-provider) starts
#    at array[0] too because `WsProvider` initialises with
#    `#endpointIndex = -1` and `selectEndpointIndex` returns `(idx + 1) % N`.
#
# In both cases array[0] is the runtime preference and array[1..N-1] is
# the failover ladder on stall/disconnect. The probe never filters; it
# only reorders, so the ladder always has somewhere to escape to.
#
# Cadence: daily. Endpoint health drifts on the order of weeks, but a
# single-day cadence keeps the latency between "an endpoint goes bad"
# and "the JSON reflects it" bounded.

name: Probe RPC endpoints

on:
  workflow_dispatch:
    inputs:
      samples:
        description: 'RPC pings per endpoint (default 100)'
        required: false
        type: string
      window_ms:
        description: 'Wall-clock budget for the pings, in ms (default 10000)'
        required: false
        type: string
      inflight:
        description: 'Max concurrent pings per endpoint (default 4)'
        required: false
        type: string
      chains:
        description: 'Space-separated chain keys to probe (default: all)'
        required: false
        type: string
  schedule:
    - cron: '17 4 * * *'

permissions:
  contents: write

jobs:
  probe-and-rank:
    timeout-minutes: 30
    runs-on: ubuntu-latest
    steps:
      - uses: actions/create-github-app-token@v1
        id: app-token
        with:
          app-id: ${{ secrets.GH_APP_ID }}
          private-key: ${{ secrets.GH_APP_PRIVATE_KEY }}

      - uses: actions/checkout@v4
        with:
          token: ${{ steps.app-token.outputs.token }}

      - uses: actions/setup-node@v4
        with:
          node-version: 22.x
          cache: 'yarn'

      - run: yarn --immutable

      - name: Probe endpoints
        env:
          SAMPLES: ${{ inputs.samples }}
          WINDOW_MS: ${{ inputs.window_ms }}
          INFLIGHT: ${{ inputs.inflight }}
          CHAINS: ${{ inputs.chains }}
        run: node scripts/probe-endpoints.mjs

      - name: Apply ranking
        run: node scripts/apply-probe-ranking.mjs

      - name: Upload probe report
        if: always()
        uses: actions/upload-artifact@v4
        with:
          name: endpoint-probe
          path: endpoint-probe.json
          retention-days: 30

      - name: Commit and push changes
        run: |
          git config --global user.name 'github-actions[bot]'
          git config --global user.email '41898282+github-actions[bot]@users.noreply.github.com'
          git add packages/networks/src/pet-chain-endpoints.json
          if ! git diff --cached --quiet; then
            git commit -m "[ci skip] Re-rank RPC endpoints by health"
            git push
          else
            echo "No changes to commit"
          fi