@keepur/hive 0.1.10 → 0.2.1

package/service/deploy-check.sh

@@ -0,0 +1,88 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+# =============================================================================
+# Deploy check — per-instance version compare against npm
+# =============================================================================
+#
+# Reads each instance's pinned tag from instances.conf. For each instance whose
+# installed version (from .hive/package.json) differs from the tag it's pinned to,
+# invokes deploy.sh --instance=<id> --tag=<tag>.
+#
+# Pinned "latest" compares against the npm `latest` dist-tag — so unpinned
+# instances autoupgrade whenever a new @keepur/hive is published.
+# =============================================================================
+
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+BUILD_DIR="${BUILD_DIR:-$HOME/build/hive}"
+DEPLOY_DIR="${DEPLOY_DIR:-$HOME/services/hive}"
+INSTANCES_CONF="$SCRIPT_DIR/instances.conf"
+
+cd "$BUILD_DIR"
+[[ "$(git branch --show-current)" == "deploy" ]] || { echo "ERROR: Build dir not on deploy branch"; exit 1; }
+
+echo "Checking for updates on deploy branch..."
+git fetch origin deploy --quiet
+
+# --- Load instances ---
+declare -a INSTANCES=()
+while IFS='|' read -r id config _agents _label logs_dir ports engine_tag; do
+  [[ "$id" =~ ^[[:space:]]*# ]] && continue
+  [[ -z "$id" ]] && continue
+  id=$(echo "$id" | xargs)
+  engine_tag=$(echo "${engine_tag:-}" | xargs)
+  INSTANCES+=("$id|${engine_tag:-latest}")
+done < "$INSTANCES_CONF"
+
+UPDATES_NEEDED=()
+for inst in "${INSTANCES[@]}"; do
+  IFS='|' read -r id tag <<< "$inst"
+  version="${tag#v}"
+  # Per-instance root: $DEPLOY_DIR/<id> if it exists (post-Phase-5 layout),
+  # else $DEPLOY_DIR (today's shared-dir primary layout).
+  if [[ -d "$DEPLOY_DIR/$id" ]]; then
+    instance_root="$DEPLOY_DIR/$id"
+  else
+    instance_root="$DEPLOY_DIR"
+  fi
+  installed=$(jq -r .version < "$instance_root/.hive/package.json" 2>/dev/null || echo "unknown")
+  if [[ "$version" == "latest" ]]; then
+    target=$(npm view @keepur/hive version 2>/dev/null || echo "unknown")
+  else
+    target=$(npm view "@keepur/hive@$version" version 2>/dev/null || echo "unknown")
+  fi
+  if [[ "$target" == "unknown" ]]; then
+    echo "  [$id] could not resolve target version (pinned: $tag). Skipping."
+    continue
+  fi
+  if [[ "$installed" == "$target" ]]; then
+    echo "  [$id] up to date ($installed)."
+  else
+    echo "  [$id] $installed → $target (pinned: $tag)"
+    UPDATES_NEEDED+=("$id|$target")
+  fi
+done
+
+if [[ ${#UPDATES_NEEDED[@]} -eq 0 ]]; then
+  echo "All instances up to date. Nothing to deploy."
+  exit 0
+fi
+
+echo ""
+echo "Updates needed:"
+for u in "${UPDATES_NEEDED[@]}"; do
+  IFS='|' read -r id target <<< "$u"
+  echo "  - $id → $target"
+done
+
+# Deploy each one. deploy.sh handles build once per invocation, so we loop.
+# Rationale for one-at-a-time: build-phase is shared, but the build is cheap
+# to re-run and keeping it inside deploy.sh keeps one script responsible for
+# the full flow. If this becomes a performance issue, split build into a
+# separate phase invoked once up front.
+for u in "${UPDATES_NEEDED[@]}"; do
+  IFS='|' read -r id target <<< "$u"
+  echo ""
+  echo "=== Deploying $id → $target ==="
+  "$SCRIPT_DIR/deploy.sh" --instance="$id" --tag="$target"
+done

package/service/deploy.sh

@@ -0,0 +1,476 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+# =============================================================================
+# Hive Deploy — build once, deploy to all instances
+# =============================================================================
+
+# --- Configuration ---
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+BUILD_DIR="${BUILD_DIR:-$HOME/build/hive}"
+DEPLOY_DIR="${DEPLOY_DIR:-$HOME/services/hive}"
+INSTANCES_CONF="$SCRIPT_DIR/instances.conf"
+
+# --- Flags ---
+DRY_RUN=false
+ROLLBACK=false
+FILTER_INSTANCE=""
+OVERRIDE_TAG=""
+for arg in "$@"; do
+  case "$arg" in
+    --dry-run) DRY_RUN=true ;;
+    --rollback) ROLLBACK=true ;;
+    --instance=*) FILTER_INSTANCE="${arg#--instance=}" ;;
+    --tag=*) OVERRIDE_TAG="${arg#--tag=}" ;;
+    *)
+      echo "ERROR: unknown arg: $arg" >&2
+      echo "Usage: deploy.sh [--dry-run] [--rollback] [--instance=<id>] [--tag=<tag>]" >&2
+      exit 2
+      ;;
+  esac
+done
+
+# --- Notification config (from dodi's .env — the primary instance) ---
+# shellcheck source=/dev/null
+source "$DEPLOY_DIR/.env"
+: "${SLACK_BOT_TOKEN:?SLACK_BOT_TOKEN not set in .env}"
+: "${DEVOPS_CHANNEL_ID:?DEVOPS_CHANNEL_ID not set in .env}"
+
+# --- Load instances ---
+declare -a INSTANCES=()
+while IFS='|' read -r id config _agents_path label logs_dir ports engine_tag; do
+  [[ "$id" =~ ^[[:space:]]*# ]] && continue  # skip comments
+  [[ -z "$id" ]] && continue                  # skip blank lines
+  # Trim whitespace
+  id=$(echo "$id" | xargs)
+  config=$(echo "$config" | xargs)
+  label=$(echo "$label" | xargs)
+  logs_dir=$(echo "$logs_dir" | xargs)
+  ports=$(echo "$ports" | xargs)
+  engine_tag=$(echo "${engine_tag:-}" | xargs)
+  INSTANCES+=("$id|$config|$label|$logs_dir|$ports|$engine_tag")
+done < "$INSTANCES_CONF"
+
+if [[ ${#INSTANCES[@]} -eq 0 ]]; then
+  echo "ERROR: No instances found in $INSTANCES_CONF"
+  exit 1
+fi
+
+echo "=== Hive Deploy (${#INSTANCES[@]} instances) ==="
+for inst in "${INSTANCES[@]}"; do
+  echo "  - $(echo "$inst" | cut -d'|' -f1)"
+done
+
+# --- Helpers ---
+run_cmd() {
+  if $DRY_RUN; then
+    echo "[DRY RUN] $*"
+  else
+    "$@"
+  fi
+}
+
+notify() {
+  local message="$1"
+  if $DRY_RUN; then
+    echo "[DRY RUN] notify: $message"
+    return
+  fi
+  local payload
+  payload=$(jq -n --arg channel "$DEVOPS_CHANNEL_ID" --arg text "$message" \
+    '{channel: $channel, text: $text}')
+  curl -s -X POST https://slack.com/api/chat.postMessage \
+    -H "Authorization: Bearer $SLACK_BOT_TOKEN" \
+    -H "Content-Type: application/json" \
+    -d "$payload" \
+    > /dev/null
+}
+
+health_check() {
+  local log_file="$1"
+  if $DRY_RUN; then
+    echo "[DRY RUN] health_check: would check $log_file"
+    return 0
+  fi
+  for _ in $(seq 1 30); do
+    sleep 1
+    if tail -5 "$log_file" 2>/dev/null | grep -q '"Hive is running"'; then
+      return 0
+    fi
+  done
+  return 1
+}
+
+kill_ports() {
+  local ports_str="$1"
+  if $DRY_RUN; then
+    echo "[DRY RUN] kill_ports: would scan $ports_str"
+    return
+  fi
+  for port in $ports_str; do
+    local pids
+    pids=$(lsof -i :"$port" -t 2>/dev/null || true)
+    if [[ -n "$pids" ]]; then
+      echo "  Killing stale process(es) on port $port: $pids"
+      echo "$pids" | xargs kill -9 2>/dev/null || true
+    fi
+  done
+  sleep 1
+}
+
+# --- Engine fetch/swap/rollback (KPR-53 / Phase 3) ---
+
+# Resolve a tag to an npm-facing bare-semver version string.
+# "v0.2.0" → "0.2.0"; "latest" → "latest"; "0.2.0" → "0.2.0"
+_normalize_tag() {
+  local tag="${1:-latest}"
+  if [[ "$tag" == "latest" ]]; then
+    echo "latest"
+  else
+    echo "${tag#v}"
+  fi
+}
+
+# _instance_root <id>
+# Returns the instance root dir: $DEPLOY_DIR/<id> if that dir exists (post-Phase-5
+# per-instance layout), else $DEPLOY_DIR (today's primary-shared-dir layout).
+# Lets deploy.sh be per-instance-dir aware now so Phase 5's migration is a no-op
+# at the script level.
+_instance_root() {
+  local id="$1"
+  if [[ -d "$DEPLOY_DIR/$id" ]]; then
+    echo "$DEPLOY_DIR/$id"
+  else
+    echo "$DEPLOY_DIR"
+  fi
+}
+
+# fetch_engine <instance_dir> <tag>
+# Populates <instance_dir>/.hive.next/ with the tag's contents.
+# Primary path: npm pack @keepur/hive@<version> + tar -xzf --strip-components=1.
+# Fallback: rsync from $BUILD_DIR (developer-ergonomics path), only when npm pack
+# actually fails (not merely when the tag isn't in the registry — transient
+# registry errors should surface, not silently swap to rsync).
+fetch_engine() {
+  local instance_dir="$1"
+  local tag="$2"
+  local version
+  version=$(_normalize_tag "$tag")
+
+  if $DRY_RUN; then
+    echo "[DRY RUN] fetch_engine: would populate $instance_dir/.hive.next/ from @keepur/hive@$version"
+    return 0
+  fi
+
+  rm -rf "$instance_dir/.hive.next"
+  mkdir -p "$instance_dir/.hive.next"
+
+  local packdir
+  packdir=$(mktemp -d)
+  local tarball
+  # npm pack prints the tarball filename on the last line of stdout.
+  # Run from a temp dir so the tarball doesn't litter $PWD.
+  echo "  fetch_engine: npm pack @keepur/hive@$version"
+  tarball=$(cd "$packdir" && npm pack "@keepur/hive@$version" 2>/dev/null | tail -n1)
+
+  if [[ -n "$tarball" && -f "$packdir/$tarball" ]]; then
+    tar -xzf "$packdir/$tarball" --strip-components=1 -C "$instance_dir/.hive.next/"
+    rm -rf "$packdir"
+  else
+    rm -rf "$packdir"
+    echo "  fetch_engine: npm pack failed; falling back to rsync from $BUILD_DIR" >&2
+    local src="$BUILD_DIR"
+    if [[ ! -f "$src/pkg/server.min.js" ]]; then
+      rm -rf "$instance_dir/.hive.next"
+      echo "ERROR: npm pack @keepur/hive@$version failed and fallback needs $src/pkg/server.min.js — run 'npm run bundle' in $src" >&2
+      return 1
+    fi
+    rsync -a --delete "$src/pkg/"       "$instance_dir/.hive.next/pkg/"
+    rsync -a --delete "$src/seeds/"     "$instance_dir/.hive.next/seeds/"
+    rsync -a --delete "$src/templates/" "$instance_dir/.hive.next/templates/"
+    rsync -a --delete "$src/install/"   "$instance_dir/.hive.next/install/"
+    rsync -a --delete "$src/service/"   "$instance_dir/.hive.next/service/"
+    # scripts/honeypot is a single binary, not the whole scripts/ dir
+    mkdir -p "$instance_dir/.hive.next/scripts"
+    [[ -f "$src/scripts/honeypot" ]] && cp "$src/scripts/honeypot" "$instance_dir/.hive.next/scripts/honeypot"
+    cp "$src/package.json" "$instance_dir/.hive.next/"
+  fi
+
+  # Sanity check — if the tarball/rsync was broken, catch it before the swap.
+  if [[ ! -f "$instance_dir/.hive.next/pkg/server.min.js" ]]; then
+    rm -rf "$instance_dir/.hive.next"
+    echo "ERROR: .hive.next/pkg/server.min.js missing after fetch_engine" >&2
+    return 1
+  fi
+}
+
+# install_engine_deps <instance_dir>
+# Runs `npm install --omit=dev` inside .hive.next/ so the bundle's runtime
+# externals (native modules, large SDKs, asset loaders — 14 of them) resolve
+# from .hive/pkg/. Node walks up to .hive/node_modules/ to find them.
+# Mirrors src/setup/populate-engine.ts so `hive init` and `hive update`
+# produce byte-identical .hive/ layouts.
+install_engine_deps() {
+  local instance_dir="$1"
+  if $DRY_RUN; then
+    echo "[DRY RUN] install_engine_deps: would npm install --omit=dev in $instance_dir/.hive.next/"
+    return 0
+  fi
+  if [[ ! -f "$instance_dir/.hive.next/package.json" ]]; then
+    echo "ERROR: install_engine_deps needs $instance_dir/.hive.next/package.json" >&2
+    return 1
+  fi
+  echo "  install_engine_deps: npm install --omit=dev in .hive.next/"
+  (cd "$instance_dir/.hive.next" && npm install --omit=dev --no-audit --no-fund --no-progress >&2)
+}
+
+# swap_engine <instance_dir>
+# Rotates: old .hive.prev → dropped; live .hive → .hive.prev; .hive.next → .hive.
+# Assumes the service is already stopped. The ~50ms window where .hive/ doesn't
+# exist is covered by the service being down.
+swap_engine() {
+  local instance_dir="$1"
+  if $DRY_RUN; then
+    echo "[DRY RUN] swap_engine: would rotate $instance_dir/.hive{,.prev,.next}"
+    return 0
+  fi
+  if [[ ! -d "$instance_dir/.hive.next" ]]; then
+    echo "ERROR: swap_engine called but $instance_dir/.hive.next does not exist" >&2
+    return 1
+  fi
+  if [[ -d "$instance_dir/.hive" ]]; then
+    rm -rf "$instance_dir/.hive.prev"           # drop older backup
+    mv "$instance_dir/.hive" "$instance_dir/.hive.prev"
+  fi
+  mv "$instance_dir/.hive.next" "$instance_dir/.hive"
+  # Clear any .hive.broken/ from a previous failed rollback — this is a successful deploy.
+  rm -rf "$instance_dir/.hive.broken"
+}
+
+# rollback_engine <instance_dir>
+# Manual rollback OR auto-rollback after health-check failure.
+# Moves the failed engine to .hive.broken/ (preserved for inspection) and
+# restores .hive.prev → .hive.
+rollback_engine() {
+  local instance_dir="$1"
+  if $DRY_RUN; then
+    if [[ ! -d "$instance_dir/.hive.prev" ]]; then
+      echo "[DRY RUN] rollback_engine: would fail — no $instance_dir/.hive.prev"
+      return 1
+    fi
+    echo "[DRY RUN] rollback_engine: would swap $instance_dir/.hive ↔ .hive.prev (failed engine → .hive.broken)"
+    return 0
+  fi
+  if [[ ! -d "$instance_dir/.hive.prev" ]]; then
+    echo "ERROR: no previous engine at $instance_dir/.hive.prev — rollback unavailable" >&2
+    return 1
+  fi
+  rm -rf "$instance_dir/.hive.broken"
+  if [[ -d "$instance_dir/.hive" ]]; then
+    mv "$instance_dir/.hive" "$instance_dir/.hive.broken"
+  fi
+  mv "$instance_dir/.hive.prev" "$instance_dir/.hive"
+}
+
+# --- Short-circuit: --rollback mode ---
+# Rollback is per-instance; requires --instance=<id> so we know which to roll.
+# No build phase, no notify until after the swap.
+if $ROLLBACK; then
+  if [[ -z "$FILTER_INSTANCE" ]]; then
+    echo "ERROR: --rollback requires --instance=<id>" >&2
+    exit 2
+  fi
+  # Find the matching instance row so we know its logs dir + label.
+  ROLLBACK_ROW=""
+  for inst in "${INSTANCES[@]}"; do
+    IFS='|' read -r id _config _label _logs _ports _tag <<< "$inst"
+    if [[ "$id" == "$FILTER_INSTANCE" ]]; then
+      ROLLBACK_ROW="$inst"
+      break
+    fi
+  done
+  if [[ -z "$ROLLBACK_ROW" ]]; then
+    echo "ERROR: no instance '$FILTER_INSTANCE' in $INSTANCES_CONF" >&2
+    exit 2
+  fi
+  IFS='|' read -r id _config label logs_dir ports _tag <<< "$ROLLBACK_ROW"
+  instance_root=$(_instance_root "$id")
+  echo "--- Rolling back $id (root: $instance_root) ---"
+  # Stop LaunchAgent BEFORE rotating .hive/. Without -kp, launchd auto-respawns
+  # mid-rollback and the restart picks up a partial state. Mirrors the deploy
+  # loop's stop sequence at the main branch.
+  echo "  Stopping $label..."
+  run_cmd launchctl kickstart -kp "gui/$(id -u)/$label" 2>/dev/null || true
+  kill_ports "$ports"
+  if ! rollback_engine "$instance_root"; then
+    notify "Rollback FAILED for \`$id\`: no previous engine (.hive.prev missing)."
+    exit 1
+  fi
+  run_cmd launchctl kickstart -k "gui/$(id -u)/$label"
+  if health_check "$instance_root/$logs_dir/hive.log"; then
+    rollback_version=$(jq -r .version < "$instance_root/.hive/package.json" 2>/dev/null || echo "unknown")
+    notify "Rollback succeeded for \`$id\` → \`$rollback_version\`."
+    echo "Rollback complete."
+    exit 0
+  else
+    notify "Rollback succeeded but health check failed for \`$id\`. Check logs."
+    exit 1
+  fi
+fi
+
+# --- Guard: shared instance_root with diverging pins ---
+# When multiple instances resolve to the same instance_root (today: dodi+personal
+# both share $DEPLOY_DIR until Phase 5 migrates them into per-instance dirs),
+# they share a single .hive/ — their ENGINE_TAG pins MUST agree. Otherwise
+# per-instance deploys silently clobber each other and deploy-check.sh
+# oscillates between versions on every poll. Compares the configured pins
+# (not the effective tag for this run) so even an explicit --tag override
+# can't sneak past a misconfigured pinning state. Fail fast before any work.
+declare -a _seen_roots=()
+declare -a _seen_pins=()
+for inst in "${INSTANCES[@]}"; do
+  IFS='|' read -r _id _config _label _logs _ports _engine_tag <<< "$inst"
+  _root=$(_instance_root "$_id")
+  _pin="${_engine_tag:-latest}"
+  for i in "${!_seen_roots[@]}"; do
+    if [[ "${_seen_roots[$i]}" == "$_root" && "${_seen_pins[$i]}" != "$_pin" ]]; then
+      echo "ERROR: instances share root '$_root' but pin different ENGINE_TAGs ('${_seen_pins[$i]}' vs '$_pin')." >&2
+      echo "       Set the same ENGINE_TAG for all instances under one root, or migrate them to per-instance dirs (Phase 5)." >&2
+      exit 2
+    fi
+  done
+  _seen_roots+=("$_root")
+  _seen_pins+=("$_pin")
+done
+unset _seen_roots _seen_pins _id _config _label _logs _ports _engine_tag _root _pin i
+
+# =============================================================================
+# Phase 1: Build (once, in $BUILD_DIR)
+# =============================================================================
+
+echo ""
+echo "--- Phase 1: Build ---"
+# Per-instance current versions are reported in Phase 2 after each health check.
+
+echo "Pulling latest in build dir..."
+cd "$BUILD_DIR"
+[[ "$(git branch --show-current)" == "deploy" ]] || { echo "ERROR: Build dir not on deploy branch"; exit 1; }
+run_cmd git pull --ff-only
+
+DEPLOY_SHA=$(git rev-parse --short HEAD)
+DEPLOY_MSG=$(git log -1 --pretty=%s)
+
+echo "Installing dependencies..."
+if ! run_cmd npm install; then
+  notify "Deploy aborted. \`npm install\` failed. Commit: \`$DEPLOY_SHA\`."
+  exit 1
+fi
+
+echo "Running checks..."
+if ! run_cmd npm run check; then
+  notify "Deploy aborted. \`npm run check\` failed. Commit: \`$DEPLOY_SHA\`."
+  exit 1
+fi
+
+echo "Building..."
+if ! run_cmd npm run build; then
+  notify "Deploy aborted. Build failed. Commit: \`$DEPLOY_SHA\`."
+  exit 1
+fi
+
+echo "Bundling..."
+if ! run_cmd npm run bundle; then
+  notify "Deploy aborted. Bundle failed. Commit: \`$DEPLOY_SHA\`."
+  exit 1
+fi
+
+# =============================================================================
+# Phase 2: Deploy each instance
+# =============================================================================
+
+FAILED_INSTANCES=()
+
+for inst in "${INSTANCES[@]}"; do
+  IFS='|' read -r id config label logs_dir ports engine_tag <<< "$inst"
+
+  # --instance=<id> filter
+  if [[ -n "$FILTER_INSTANCE" && "$FILTER_INSTANCE" != "$id" ]]; then
+    echo ""
+    echo "--- Skipping '$id' (filtered: --instance=$FILTER_INSTANCE) ---"
+    continue
+  fi
+
+  # --tag override > per-instance engine_tag > "latest"
+  tag="${OVERRIDE_TAG:-${engine_tag:-latest}}"
+  instance_root=$(_instance_root "$id")
+
+  echo ""
+  echo "--- Phase 2: Deploy instance '$id' @ $tag (root: $instance_root) ---"
+  echo "  config=$config label=$label logs=$logs_dir ports=$ports"
+
+  mkdir -p "$instance_root/$logs_dir"
+
+  echo "  Stopping $label..."
+  run_cmd launchctl kickstart -kp "gui/$(id -u)/$label" 2>/dev/null || true
+  kill_ports "$ports"
+
+  echo "  Fetching engine..."
+  if ! fetch_engine "$instance_root" "$tag"; then
+    notify "Deploy FAILED for \`$id\`: fetch_engine errored at tag \`$tag\`."
+    FAILED_INSTANCES+=("$id")
+    run_cmd launchctl kickstart -k "gui/$(id -u)/$label" || true  # bring old engine back up
+    continue
+  fi
+
+  echo "  Installing engine deps..."
+  if ! install_engine_deps "$instance_root"; then
+    notify "Deploy FAILED for \`$id\`: install_engine_deps errored at tag \`$tag\`."
+    FAILED_INSTANCES+=("$id")
+    rm -rf "$instance_root/.hive.next"
+    run_cmd launchctl kickstart -k "gui/$(id -u)/$label" || true  # bring old engine back up
+    continue
+  fi
+
+  echo "  Swapping engine..."
+  swap_engine "$instance_root"
+
+  echo "  Restarting $label..."
+  run_cmd launchctl kickstart -k "gui/$(id -u)/$label"
+
+  echo "  Checking health..."
+  if ! health_check "$instance_root/$logs_dir/hive.log"; then
+    echo "  Health check FAILED for $id — rolling back"
+    if rollback_engine "$instance_root"; then
+      run_cmd launchctl kickstart -k "gui/$(id -u)/$label"
+      notify "Deploy rolled back for \`$id\`: \`$tag\` failed health check, restored previous version."
+    else
+      notify "Deploy FAILED for \`$id\` and auto-rollback unavailable (.hive.prev missing). Manual intervention required."
+    fi
+    FAILED_INSTANCES+=("$id")
+    continue
+  fi
+
+  new_version=$(jq -r .version < "$instance_root/.hive/package.json" 2>/dev/null || echo "unknown")
+  echo "  Instance '$id' is healthy at version $new_version."
+done
+
+# =============================================================================
+# Phase 3: Report
+# =============================================================================
+
+echo ""
+echo "--- Phase 3: Report ---"
+
+if [[ ${#FAILED_INSTANCES[@]} -gt 0 ]]; then
+  failed_list=$(printf ", %s" "${FAILED_INSTANCES[@]}")
+  failed_list=${failed_list:2}
+  notify "Deploy partial. Build commit \`$DEPLOY_SHA\`: $DEPLOY_MSG. Failed instances: $failed_list."
+  echo "Deploy completed with failures: $failed_list"
+  exit 1
+else
+  # Count actual deploy targets (respecting --instance filter)
+  deployed=${#INSTANCES[@]}
+  [[ -n "$FILTER_INSTANCE" ]] && deployed=1
+  notify "Deploy succeeded ($deployed instance(s)). Build commit \`$DEPLOY_SHA\`: $DEPLOY_MSG."
+  echo "Deploy complete. $deployed instance(s) running."
+fi