who_need_help/scripts/load-stack-up.sh

#!/bin/sh
set -eu

ROOT=$(CDPATH='' cd -- "$(dirname -- "$0")/.." && pwd)
ENV_FILE=${WNH_LOAD_ENV_FILE:-"$ROOT/output/runtime/load.env"}
LOAD_TOOLS_IMAGE=${WNH_LOAD_TOOLS_IMAGE:-who-need-help:load-tools}
PYTHON_RUNTIME_IMAGE=${WNH_PYTHON_RUNTIME_IMAGE:-who-need-help:python-runtime}
REPLICAS=${1:-}

"$ROOT/scripts/ensure-local-load-env.sh"

set -a
# shellcheck source=/dev/null
. "$ENV_FILE"
set +a

: "${LOAD_PROJECT:?LOAD_PROJECT is missing from the generated load runtime environment}"
: "${LOAD_HOST:?LOAD_HOST is missing from the generated load runtime environment}"
: "${LOAD_WEB_REPLICAS:?LOAD_WEB_REPLICAS is missing from the generated load runtime environment}"
: "${LOAD_WORKER_REPLICAS:?LOAD_WORKER_REPLICAS is missing from the generated load runtime environment}"
: "${LOAD_RESILIENCE_RECOVERY_TIMEOUT_SECONDS:?LOAD_RESILIENCE_RECOVERY_TIMEOUT_SECONDS is missing from the generated load runtime environment}"
: "${LOAD_RESILIENCE_PROBE_INTERVAL_SECONDS:?LOAD_RESILIENCE_PROBE_INTERVAL_SECONDS is missing from the generated load runtime environment}"

if [ "$LOAD_PROJECT" = who_need_help ]; then
  echo "The load profile must not use the staging Compose project." >&2
  exit 1
fi

if [ -n "$REPLICAS" ]; then
  LOAD_WEB_REPLICAS=$REPLICAS
fi

case "$LOAD_WEB_REPLICAS:$LOAD_WORKER_REPLICAS" in
  *[!0-9:]* | 0:* | *:0 | :* | *:)
    echo "Web and worker replica counts must be positive integers." >&2
    exit 1
    ;;
esac

cd "$ROOT"
compose() {
  if [ "${WNH_LOAD_CPU_REPLAY:-false}" = true ]; then
    docker compose \
      --env-file "$ENV_FILE" \
      -p "$LOAD_PROJECT" \
      -f compose.yaml \
      -f compose.load.yaml \
      -f compose.cpu-replay.yaml \
      "$@"
  else
    docker compose \
      --env-file "$ENV_FILE" \
      -p "$LOAD_PROJECT" \
      -f compose.yaml \
      -f compose.load.yaml \
      "$@"
  fi
}

cleanup_failed_start() {
  status=$?
  trap - EXIT HUP INT TERM

  if [ "$status" -ne 0 ]; then
    # A cancelled Docker API request can finish creating a network just after
    # the first `compose down`. Repeat the exact project-scoped teardown until
    # the project has no containers or networks, or the bounded retry expires.
    attempt=0
    while [ "$attempt" -lt 5 ]; do
      compose down --remove-orphans >/dev/null 2>&1 || true
      remaining_containers=$(docker ps -aq \
        --filter "label=com.docker.compose.project=$LOAD_PROJECT")
      remaining_networks=$(docker network ls -q \
        --filter "label=com.docker.compose.project=$LOAD_PROJECT")

      if [ -z "$remaining_containers" ] && [ -z "$remaining_networks" ]; then
        break
      fi

      attempt=$((attempt + 1))
      sleep 1
    done

    if image_id=$(docker image inspect --format '{{.Id}}' "$LOAD_TOOLS_IMAGE" 2>/dev/null); then
      if [ -z "$(docker ps -aq --filter "ancestor=$image_id")" ]; then
        docker image rm "$LOAD_TOOLS_IMAGE" >/dev/null 2>&1 || true
      fi
    fi

    if image_id=$(docker image inspect --format '{{.Id}}' "$PYTHON_RUNTIME_IMAGE" 2>/dev/null); then
      if [ -z "$(docker ps -aq --filter "ancestor=$image_id")" ]; then
        docker image rm "$PYTHON_RUNTIME_IMAGE" >/dev/null 2>&1 || true
      fi
    fi

    if [ -n "$(docker ps -aq --filter "label=com.docker.compose.project=$LOAD_PROJECT")" ] || \
        [ -n "$(docker network ls -q --filter "label=com.docker.compose.project=$LOAD_PROJECT")" ]; then
      echo "Load-profile startup cleanup left project resources behind." >&2
      status=1
    fi
  fi

  exit "$status"
}

trap cleanup_failed_start EXIT HUP INT TERM

docker build --target load_tools --tag "$LOAD_TOOLS_IMAGE" .
docker build \
  --target python_runtime \
  --tag "$PYTHON_RUNTIME_IMAGE" \
  --file ops/external-boundaries/Dockerfile \
  ops/external-boundaries

compose up -d --build --wait \
  --scale "web=$LOAD_WEB_REPLICAS" \
  --scale "worker=$LOAD_WORKER_REPLICAS"

COMPOSE_PROJECT_NAME=$LOAD_PROJECT "$ROOT/scripts/verify-realtime-cluster.sh" compose
COMPOSE_PROJECT_NAME=$LOAD_PROJECT "$ROOT/scripts/verify-beam-runtime.sh" compose
deadline=$(( $(date +%s) + LOAD_RESILIENCE_RECOVERY_TIMEOUT_SECONDS ))
status=000

while [ "$(date +%s)" -le "$deadline" ]; do
  if ! status=$(
    curl --silent \
      --header "Host: $LOAD_HOST" \
      --output /dev/null \
      --write-out '%{http_code}' \
      "http://localhost:$HTTP_PORT/healthz/ready"
  ); then
    status=000
  fi

  if [ "$status" = 200 ]; then
    break
  fi

  sleep "$LOAD_RESILIENCE_PROBE_INTERVAL_SECONDS"
done

if [ "$status" != 200 ]; then
  echo "Load-profile readiness returned HTTP $status instead of 200." >&2
  exit 1
fi

curl --fail --silent --show-error \
  --header "Host: $LOAD_HOST" \
  "http://localhost:$HTTP_PORT/healthz/ready"
printf '\nLoad profile is ready with %s web and %s worker replicas.\n' \
  "$LOAD_WEB_REPLICAS" "$LOAD_WORKER_REPLICAS"

trap - EXIT HUP INT TERM