Better handling of comma-separated CL_NODE

Author: yorickdowne

default.env

@@ -297,7 +297,7 @@ LOG_LEVEL=info
 JWT_SECRET=
 # Authenticated execution client endpoint. This default uses the execution node container.
 EL_NODE=http://execution:8551
-# Consensus client address. This could be comma-separated for Lighthouse, Nimbus or Teku VC clients, with failover,
+# Consensus client address. This could be a comma-separated list with failover,
 # or could just be a remote consensus client URL for "validator only" setups.
 CL_NODE=http://consensus:5052
 # Used by "ethd keys", adjust this if you have multiple Eth Docker stacks connected to the same Docker bridge network

ethdo/Dockerfile

@@ -9,6 +9,7 @@ RUN apt-get update && DEBIAN_FRONTEND=noninteractive TZ=Etc/UTC apt-get install
   adduser \
   bash \
   jq \
+  curl \
   && gosu nobody true \
   && apt-get clean \
   && rm -rf /var/lib/apt/lists/*

ethdo/docker-entrypoint.sh

@@ -83,8 +83,20 @@ if [[ "$*" =~ "--offline" ]]; then
     chown ethdo:ethdo /app/offline-preparation.json
   fi
 else
-  # Get just the first CL_NODE
-  __args=( "${__args[@]:0:1}" "--connection" "$(cut -d, -f1 <<<"${CL_NODE}")" "${__args[@]:1}" )
+  nodes=$(echo "${CL_NODE}" | tr ',' ' ')
+  for node in ${nodes}; do
+    if curl -s -m 5 -o /dev/null -w "%{http_code}" "${node}" | grep -q "^[23]"; then
+      node_reachable=1
+      break
+    fi
+  done
+
+  if [[ "${node_reachable}" -eq 0 ]]; then
+    echo "No consensus client node is reachable via any URL in ${CL_NODE}"
+    exit 1
+  fi
+
+  __args=( "${__args[@]:0:1}" "--connection" "${node}" "${__args[@]:1}" )
 fi
 
 set +e

lighthouse-vc-only.yml

@@ -93,6 +93,8 @@ services:
     depends_on:
       lighthouse-builder:
         condition: service_completed_successfully
+    environment:
+      - CL_NODE=${CL_NODE}
     volumes:
       - lhvalidator-data:/var/lib/lighthouse
       - ./.eth/validator_keys:/validator_keys
@@ -103,8 +105,6 @@ services:
       - account
       - validator
       - exit
-      - --beacon-node
-      - ${CL_NODE:-http://consensus:5052}
       - --datadir
       - /var/lib/lighthouse
       - --network

lighthouse.yml

@@ -177,6 +177,8 @@ services:
         condition: service_started
       lighthouse-builder:
         condition: service_completed_successfully
+    environment:
+      - CL_NODE=${CL_NODE}
     volumes:
       - lhvalidator-data:/var/lib/lighthouse
       - ./.eth/validator_keys:/validator_keys
@@ -187,8 +189,6 @@ services:
       - account
       - validator
       - exit
-      - --beacon-node
-      - http://consensus:5052
       - --datadir
       - /var/lib/lighthouse
       - --network

lighthouse/validator-exit.sh

@@ -9,4 +9,21 @@ if [[ "$(id -u)" -eq 0 ]]; then
   exec gosu lhvalidator "${BASH_SOURCE[0]}" "$@"
 fi
 
+nodes=$(echo "${CL_NODE}" | tr ',' ' ')
+for node in ${nodes}; do
+  if curl -s -m 5 -o /dev/null -w "%{http_code}" "${node}" | grep -q "^[23]"; then
+    node_reachable=1
+    break
+  fi
+done
+
+if [[ "${node_reachable}" -eq 0 ]]; then
+  echo "No consensus client node is reachable via any URL in ${CL_NODE}"
+  sleep 30
+  exit 1
+fi
+
+# Insert the --beacon-node after the 4th position
+set -- "${@:1:4}" "--beacon-node" "${node}" "${@:5}"
+
 exec "$@"

siren.yml

@@ -25,7 +25,12 @@ services:
       - NODE_OPTIONS="--dns-result-order=ipv4first"
     <<: *logging
     entrypoint:
-      - docker-entrypoint.sh
+      - /bin/bash
+      - -c
+    command:
+      - |
+        export BEACON_URL="$${BEACON_URL%%,*}"
+        exec /app/docker-assets/docker-entrypoint.sh
     labels:
       - traefik.enable=true
       - traefik.http.routers.$(SIREN_HOST:-siren}.service=${SIREN_HOST:-siren}

teku-vc-only.yml

@@ -97,7 +97,7 @@ services:
     entrypoint:
       - /opt/teku/bin/teku
       - voluntary-exit
-      - --beacon-node-api-endpoint=${CL_NODE:-http://consensus:5052}
+      - --beacon-node-api-endpoints=${CL_NODE:-http://consensus:5052}
       - --validator-keys=/var/lib/teku/validator-keys:/var/lib/teku/validator-passwords
       - --validator-keys=/var/lib/teku/validator/key-manager/local:/var/lib/teku/validator/key-manager/local-passwords
 

teku.yml

@@ -176,7 +176,7 @@ services:
     entrypoint:
       - /opt/teku/bin/teku
       - voluntary-exit
-      - --beacon-node-api-endpoint=${CL_NODE:-http://consensus:5052}
+      - --beacon-node-api-endpoints=${CL_NODE:-http://consensus:5052}
       - --validator-keys=/var/lib/teku/validator-keys:/var/lib/teku/validator-passwords
       - --validator-keys=/var/lib/teku/validator/key-manager/local:/var/lib/teku/validator/key-manager/local-passwords
 

vc-utils/keymanager.sh

@@ -96,19 +96,35 @@ else
 
 __call_cl_api() {
   local exitstatus
+  local nodes
+  local node
+  local node_reachable=0
+
+  nodes=$(echo "${CL_NODE}" | tr ',' ' ')
+  for node in ${nodes}; do
+    if curl -s -m 5 -o /dev/null -w "%{http_code}" "${node}" | grep -q "^[23]"; then
+      node_reachable=1
+      break
+    fi
+  done
+
+  if [[ "${node_reachable}" -eq 0 ]]; then
+    echo "No consensus client node is reachable via any URL in ${CL_NODE}"
+    exit 1
+  fi
 
   set +e
   if [[ -z "${__api_data}" ]]; then
     __code=$(curl -m 60 -s --show-error -o /tmp/result.txt -w "%{http_code}" -X "${__http_method}" -H "Accept: application/json" \
-        "${CL_NODE}"/"${__api_path}")
+        "${node}"/"${__api_path}")
   else
     __code=$(curl -m 60 -s --show-error -o /tmp/result.txt -w "%{http_code}" -X "${__http_method}" -H "Accept: application/json" -H "Content-Type: application/json" \
-        --data "${__api_data}" "${CL_NODE}"/"${__api_path}")
+        --data "${__api_data}" "${node}"/"${__api_path}")
   fi
   exitstatus=$?
   if [[ "${exitstatus}" -ne 0 ]]; then
     echo "Error encountered while trying to call the consensus client REST API via curl."
-    echo "Please make sure the ${CL_NODE} URL is reachable."
+    echo "Please make sure the ${node} URL is reachable."
     echo "Error code ${exitstatus}"
     exit ${exitstatus}
   fi
@@ -585,6 +601,9 @@ validator-list() {
 
 
 validator-count() {
+  local nodes
+  local node
+  local node_reachable=0
   local vc_api_container
   local vc_service
   local vc_api_port
@@ -635,9 +654,22 @@ validator-count() {
     fi
   fi
 
+  nodes=$(echo "${CL_NODE}" | tr ',' ' ')
+  for node in ${nodes}; do
+    if curl -s -m 5 -o /dev/null -w "%{http_code}" "${node}" | grep -q "^[23]"; then
+      node_reachable=1
+      break
+    fi
+  done
+
+  if [[ "${node_reachable}" -eq 0 ]]; then
+    echo "No consensus client node is reachable via any URL in ${CL_NODE}"
+    exit 1
+  fi
+
   echo "Querying validator state, this may take a minute"
   for __pubkey in $(echo "${vals}" | jq -r '.data[].validating_pubkey'); do
-    val_state=$(curl -k -m 60 -s --show-error "${CL_NODE}/eth/v1/beacon/states/head/validators/${__pubkey}" | jq -r .data.status)
+    val_state=$(curl -k -m 60 -s --show-error "${node}/eth/v1/beacon/states/head/validators/${__pubkey}" | jq -r .data.status)
     case "${val_state}" in
       active_ongoing) ((vals_active++));;
       active_exiting) ((vals_exiting++));;