Add check_bootnode script and github action

paritytech · paritytech-processbot · Mar 21, 2023 · Mar 14, 2023 · Mar 14, 2023 · Mar 14, 2023
commit 0dfafb204993000d91284e9a7b50e212f3dfecfd
@@ -0,0 +1,18 @@
+# checks all runtimes we care about (kusama, polkadot, westend) and ensures
+# the bootnodes in their respective chainspecs are contactable
+
+name: Check all bootnodes
+on: push
+
+jobs:
+  check_bootnodes:
+    strategy:
+      matrix:
+        runtime: [westend, kusama, polkadot]
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout sources
+        uses: actions/checkout@v3
+      - name: Check ${{ matrix.runtime }} bootnodes
+        shell: bash
+        run: scripts/ci/github/check_bootnodes.sh ${{ matrix.runtime }}
@@ -0,0 +1,113 @@
+#!/usr/bin/env bash
+
+# In this script, we check each bootnode for each runtime and ensure they are contactable.
+# We do this by removing every bootnode from the chainspec with the exception of the one
+# we want to check. Then we spin up a node using this new chainspec, wait a little while
+# and then check our local node's RPC endpoint for the number of peers. If the node hasn't
+# been able to contact any other nodes, we can reason that the bootnode we used is not well-connected
+# or is otherwise uncontactable.
+
+# Root of the polkadot dir
+ROOT="$(dirname "${0}")/../../.."
+RUNTIME="$1"
+
+trap cleanup EXIT INT TERM
+
+cleanup(){
+    echo "[+] Script interrupted or ended. Cleaning up..."
+    # Kill all the polkadot processes
+    killall polkadot > /dev/null 2>&1
+}
+
+check_bootnode(){
+    BOOTNODE_INDEX=$1
+    TMP_CHAINSPEC_FILE="$RUNTIME.$BOOTNODE_INDEX.tmp.json"
+    FINAL_CHAINSPEC_FILE="$RUNTIME.$BOOTNODE_INDEX.final.json"
+    # Copy the chainspec file to a temporary location to avoid weird race conditions when running in parallel
+    cp "$CHAINSPEC_FILE" "$CHAINSPEC_TMPDIR/$TMP_CHAINSPEC_FILE"
+    pushd "$CHAINSPEC_TMPDIR" > /dev/null || exit 1
+        jq ".bootNodes |= [.[$BOOTNODE_INDEX]] " < "$TMP_CHAINSPEC_FILE" > "$FINAL_CHAINSPEC_FILE"
+        BOOTNODE=$( jq -r '.bootNodes[0]' < "$FINAL_CHAINSPEC_FILE" )
+        # Get the first ephemeral port
+        BASE_PORT=$(sysctl net.inet.ip.portrange.first | awk '{print $2}')
+        RPC_PORT=$((BASE_PORT + BOOTNODE_INDEX))
+        echo "[+] Checking bootnode $BOOTNODE"
+        polkadot --chain "$FINAL_CHAINSPEC_FILE" --no-mdns --rpc-port=$RPC_PORT --tmp > /dev/null 2>&1 &
+        POLKADOT_PID=$!
+        # We just spun up a bunch of nodes... probably want to wait a bit.
+        sleep 60
+    popd > /dev/null || exit 1
+    # Check the health endpoint of the RPC node
+    PEERS="$(curl -s -X POST -H "Content-Type: application/json" --data '{"jsonrpc":"2.0","method":"system_health","params":[],"id":1}' http://localhost:$RPC_PORT | jq -r '.result.peers')"
+    # Clean up the node
+    kill -9 $POLKADOT_PID
+    # Sometimes due to machine load or other reasons, we don't get a response from the RPC node
+    # If $PEERS is an empty variable, mark the node as unreachable
+    if [ -z "$PEERS" ]; then
+        PEERS=0
+    fi
+    if [ "$PEERS" -gt 0 ]; then
+        echo "[+] $PEERS peers found for $BOOTNODE"
+        echo "    Bootnode appears contactable"
+        GOOD_BOOTNODES+=("$BOOTNODE")
+        return 0
+    else
+        echo "[!] $PEERS peers found for $BOOTNODE"
+        echo "    Bootnode appears unreachable"
+        BAD_BOOTNODES+=("$BOOTNODE")
+        return 1
+    fi
+}
+
+# For each runtime
+CHAINSPEC_FILE="$ROOT/node/service/chain-specs/$RUNTIME.json"
+# count the number of bootnodes
+BOOTNODES=$( jq -r '.bootNodes | length' "$CHAINSPEC_FILE" )
+# Make a temporary dir for chainspec files
+CHAINSPEC_TMPDIR="$(mktemp -d -t "${RUNTIME}_chainspecs")"
+echo "[+] Using $CHAINSPEC_TMPDIR as temporary chainspec dir"
+# Store an array of the bad bootnodes
+BAD_BOOTNODES=()
+GOOD_BOOTNODES=()
+PIDS=()
+echo "[+] Checking $BOOTNODES bootnodes for $RUNTIME"
+for i in $(seq 0 $((BOOTNODES-1))); do
+    # Check each bootnode in parallel
+    check_bootnode "$i" &
+    PIDS+=($!)
+    # Hold off one second between attempting to spawn nodes
+    sleep 1
+done
+RESPS=()
+# Wait for all the nodes to finish
+for pid in "${PIDS[@]}"; do
+    wait "$pid"
+    RESPS+=($?)
+done
+echo
+# For any bootnodes that failed, add them to the bad bootnodes array
+for i in "${!RESPS[@]}"; do
+    if [ "${RESPS[$i]}" -ne 0 ]; then
+        BAD_BOOTNODES+=("$( jq -r .bootNodes["$i"] < "$CHAINSPEC_FILE" )")
+    fi
+done
+# For any bootnodes that succeeded, add them to the good bootnodes array
+for i in "${!RESPS[@]}"; do
+    if [ "${RESPS[$i]}" -eq 0 ]; then
+        GOOD_BOOTNODES+=("$( jq -r .bootNodes["$i"] < "$CHAINSPEC_FILE" )")
+    fi
+done
+
+# If we've got any uncontactable bootnodes for this runtime, print them
+if [ ${#BAD_BOOTNODES[@]} -gt 0 ]; then
+    echo "[!] Bad bootnodes found for $RUNTIME:"
+    for i in "${BAD_BOOTNODES[@]}"; do
+        echo "    $i"
+    done
+    cleanup
+    exit 1
+else
+    echo "[+] All bootnodes for $RUNTIME are contactable"
+    cleanup
+    exit 0
+fi