managarten/scripts/mac-mini/health-check.sh
Till JS 85e38176d8 chore(macmini/scripts): runbook hardening — status diff + ingress walk
Two failures during the 2026-04-07 production outage triage were caused
not by the underlying outage but by `status.sh` and `health-check.sh`
hiding the broken state. Both scripts hardened so the same outage
shape can't reoccur invisibly.

status.sh — compose-vs-running diff
  The old script printed "X containers running / Y total" without
  noticing that some compose-defined containers were never started in
  the first place. The Mac Mini was running 37 of 42 declared
  containers and the script reported "37 running" with no indication
  of the gap — `mana-core-sync` and `mana-api-gateway` were silently
  missing for hours.

  New behaviour: read every service from `docker compose config`,
  diff its `container_name` against `docker ps`, and report each
  declared service whose container is not currently up. The same
  outage state would have been flagged on the very first run.

health-check.sh — public-hostname walk via Cloudflare DNS
  The old script probed ~50 hardcoded `localhost:<port>/health`
  endpoints across Chat, Todo, Calendar, etc. — but the per-app
  HTTP backends those endpoints expected don't exist anymore (the
  ghost-API cleanup removed them entirely). Every probe returned
  HTTP 000 / connection refused, generating a wall of false-positive
  alerts that drowned out the real signal.

  The block was replaced with a dynamic walk of every `hostname:`
  entry in `~/.cloudflared/config.yml`. Each hostname is probed via
  the public Cloudflare tunnel, so DNS gaps, missing tunnel routes,
  502/530 origin failures and timeouts surface as failures the same
  way real users would experience them. On its first run after the
  cleanup it surfaced eighteen previously-invisible hostname failures
  (no DNS, 502, or 530) — every one of them a real production issue.

  DNS resolution intentionally goes through `dig +short HOST @1.1.1.1`
  instead of the local resolver. The Mac Mini's home-router DNS keeps
  a negative cache for hours after the first failed lookup, so newly
  added CNAMEs (like the post-outage sync/media records) appeared as
  "no response" from inside the script for hours even though external
  users saw them resolve immediately. Asking Cloudflare's DNS directly
  gives the script the same view the public internet has.

  The Matrix, Element, GPU-LAN-redundant and monitoring port-by-port
  blocks were removed — the public-hostname walk covers all of them
  via their `*.mana.how` hostnames going through the actual tunnel.

  The "stuck container" detector now ignores `*-init` containers
  (one-shot init pods, Exit 0 = success, intentionally never re-run).

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-07 22:31:53 +02:00

373 lines
12 KiB
Bash
Executable file

#!/bin/bash
# Mana Health Check Script
# Checks all services and sends notifications on failure
#
# Notification channels (configure via environment or .env.notifications):
# - Telegram: TELEGRAM_BOT_TOKEN + TELEGRAM_CHAT_ID
# - Email: EMAIL_TO + EMAIL_FROM + SMTP_* settings
# - ntfy: NTFY_TOPIC
# Ensure PATH includes docker
export PATH="/usr/local/bin:/opt/homebrew/bin:$PATH"
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
PROJECT_ROOT="$(cd "$SCRIPT_DIR/../.." && pwd)"
# Load notification config if exists
if [ -f "$PROJECT_ROOT/.env.notifications" ]; then
source "$PROJECT_ROOT/.env.notifications"
fi
# Colors for terminal output
RED='\033[0;31m'
GREEN='\033[0;32m'
YELLOW='\033[1;33m'
NC='\033[0m' # No Color
# Track failures
FAILURES=()
# ============================================
# Notification Functions
# ============================================
send_telegram() {
local message="$1"
if [ -z "$TELEGRAM_BOT_TOKEN" ] || [ -z "$TELEGRAM_CHAT_ID" ]; then
return 0
fi
curl -s -X POST "https://api.telegram.org/bot${TELEGRAM_BOT_TOKEN}/sendMessage" \
-d "chat_id=${TELEGRAM_CHAT_ID}" \
-d "text=${message}" \
-d "parse_mode=HTML" \
>/dev/null 2>&1
if [ $? -eq 0 ]; then
echo " [Telegram] Notification sent"
else
echo " [Telegram] Failed to send"
fi
}
send_email() {
local subject="$1"
local body="$2"
if [ -z "$EMAIL_TO" ]; then
return 0
fi
# Use msmtp if available, otherwise try mail command
if command -v msmtp &> /dev/null; then
echo -e "Subject: ${subject}\nFrom: ${EMAIL_FROM:-mana@localhost}\nTo: ${EMAIL_TO}\n\n${body}" | \
msmtp -a default "$EMAIL_TO" 2>/dev/null
elif command -v mail &> /dev/null; then
echo "$body" | mail -s "$subject" "$EMAIL_TO" 2>/dev/null
elif command -v sendmail &> /dev/null; then
echo -e "Subject: ${subject}\nFrom: ${EMAIL_FROM:-mana@localhost}\nTo: ${EMAIL_TO}\n\n${body}" | \
sendmail "$EMAIL_TO" 2>/dev/null
else
echo " [Email] No mail client available (install msmtp)"
return 1
fi
if [ $? -eq 0 ]; then
echo " [Email] Notification sent to $EMAIL_TO"
else
echo " [Email] Failed to send"
fi
}
send_ntfy() {
local message="$1"
if [ -z "$NTFY_TOPIC" ]; then
return 0
fi
curl -s -d "$message" \
-H "Title: Mac Mini Alert" \
-H "Priority: high" \
-H "Tags: warning" \
"https://ntfy.sh/$NTFY_TOPIC" >/dev/null 2>&1
if [ $? -eq 0 ]; then
echo " [ntfy] Notification sent"
else
echo " [ntfy] Failed to send"
fi
}
send_all_notifications() {
local failed_services="$1"
local timestamp=$(date '+%Y-%m-%d %H:%M:%S')
# Telegram message (HTML format)
local telegram_msg="🚨 <b>Mana Health Check Failed</b>
<b>Time:</b> ${timestamp}
<b>Host:</b> $(hostname)
<b>Failed Services:</b>
${failed_services}
Check logs: <code>ssh mac-mini</code>"
# Email message
local email_subject="[ALERT] Mana Health Check Failed"
local email_body="Mana Health Check Failed
=============================
Time: ${timestamp}
Host: $(hostname)
Failed Services:
${failed_services}
To investigate:
ssh mac-mini
cd ~/projects/mana-monorepo
./scripts/mac-mini/status.sh
docker logs <container-name>"
# Plain text for ntfy
local ntfy_msg="Mana Failed: ${failed_services}"
echo ""
echo "Sending notifications..."
send_telegram "$telegram_msg"
send_email "$email_subject" "$email_body"
send_ntfy "$ntfy_msg"
}
# ============================================
# Health Check Functions
# ============================================
check_service() {
local name=$1
local url=$2
local timeout=${3:-5}
local status=$(curl -s -o /dev/null -w "%{http_code}" --max-time "$timeout" "$url" 2>/dev/null)
if [ "$status" = "200" ]; then
echo -e " ${GREEN}[OK]${NC} $name"
return 0
else
echo -e " ${RED}[FAIL]${NC} $name (HTTP $status)"
FAILURES+=("$name")
return 1
fi
}
# ============================================
# Main Health Check
# ============================================
echo ""
echo "=== Mana Health Check ==="
echo "Time: $(date)"
echo ""
echo "Infrastructure:"
# Check postgres via docker
if docker exec mana-infra-postgres pg_isready -U postgres >/dev/null 2>&1; then
echo -e " ${GREEN}[OK]${NC} PostgreSQL"
else
echo -e " ${RED}[FAIL]${NC} PostgreSQL"
FAILURES+=("PostgreSQL")
fi
# Check redis via docker
if docker exec mana-infra-redis redis-cli ping >/dev/null 2>&1; then
echo -e " ${GREEN}[OK]${NC} Redis"
else
echo -e " ${RED}[FAIL]${NC} Redis"
FAILURES+=("Redis")
fi
# Check for stuck containers (Created/Exited status). The only exclusion
# is *-init containers, which are one-shot init pods by design — Exit 0
# is success and the container intentionally never re-runs.
STUCK_CONTAINERS=$(docker ps -a --filter "status=created" --filter "status=exited" --format "{{.Names}}" \
| grep "^mana-" \
| grep -vE -- "-init$" \
|| true)
if [ -n "$STUCK_CONTAINERS" ]; then
echo -e " ${RED}[FAIL]${NC} Stuck containers detected:"
echo "$STUCK_CONTAINERS" | while read c; do echo " - $c"; done
FAILURES+=("Stuck containers: $(echo $STUCK_CONTAINERS | tr '\n' ' ')")
fi
echo ""
echo "Local services (auth + unified web):"
check_service "Auth API" "http://localhost:3001/health"
check_service "Unified Web" "http://localhost:5000/health"
# ────────────────────────────────────────────────────────────
# Public hostnames via Cloudflare Tunnel
# ────────────────────────────────────────────────────────────
# Walk every `hostname:` entry in cloudflared's ingress config and
# probe the live public URL. This catches the failure modes that the
# old port-by-port probes missed:
#
# - Tunnel ingress points at a port that no container listens on (502)
# - DNS CNAME for a tunnel hostname is missing (530 / NXDOMAIN)
# - Cloudflared is misrouted or stale on a different config
# - The container is healthy on LAN but unreachable from the public side
#
# A 200/204/301/302/308 from the public hostname is OK; anything else
# (including 404/502/530/timeout) is reported as a failure.
#
# Hostnames matching the patterns in TUNNEL_INGRESS_SKIP are excluded
# (typically internal-only routes or things we know don't expose
# /health, like raw IMAP / SSH ingress).
CLOUDFLARED_CONFIG="${HOME}/.cloudflared/config.yml"
TUNNEL_INGRESS_SKIP_REGEX='^(ssh|smtp|imap|pop3)\.'
check_public_hostname() {
local host=$1
local timeout=${2:-8}
# Resolve via Cloudflare's public DNS (1.1.1.1) instead of the local
# resolver. The Mac Mini's home-router DNS keeps a negative cache
# for hostnames that didn't exist when first queried — newly added
# CNAMEs like the 2026-04-07 sync/media records take hours to clear
# there, even though they resolve fine for external users. Asking
# 1.1.1.1 directly gives us the same view the public internet has.
local ip
ip=$(dig +short "$host" @1.1.1.1 2>/dev/null | head -1)
if [ -z "$ip" ]; then
echo -e " ${RED}[FAIL]${NC} ${host} (no DNS record on Cloudflare zone)"
FAILURES+=("${host} (no DNS)")
return 1
fi
# Try /health, accept anything <500. We use --resolve to bypass any
# local DNS cache and pin the lookup to the IP we just got back.
local status
status=$(curl -sk -o /dev/null -w "%{http_code}" --max-time "$timeout" \
--resolve "${host}:443:${ip}" \
"https://${host}/health" 2>/dev/null)
if [ -z "$status" ] || [ "$status" = "000" ]; then
echo -e " ${RED}[FAIL]${NC} ${host} (no response — tunnel timeout?)"
FAILURES+=("${host} (no response)")
return 1
fi
case "$status" in
2*|3*|404)
# 404 is OK for hostnames whose backend has no /health route
# but the tunnel + DNS are working.
echo -e " ${GREEN}[OK]${NC} ${host} (HTTP ${status})"
return 0
;;
5*)
echo -e " ${RED}[FAIL]${NC} ${host} (HTTP ${status} — origin / tunnel)"
FAILURES+=("${host} (HTTP ${status})")
return 1
;;
*)
echo -e " ${YELLOW}[WARN]${NC} ${host} (HTTP ${status})"
return 0
;;
esac
}
if [ -f "$CLOUDFLARED_CONFIG" ]; then
echo ""
echo "Public hostnames (Cloudflare Tunnel ingress):"
HOSTNAMES=$(awk '/^[[:space:]]*-[[:space:]]*hostname:/{print $3}' "$CLOUDFLARED_CONFIG" \
| grep -vE "$TUNNEL_INGRESS_SKIP_REGEX" \
| sort -u)
for host in $HOSTNAMES; do
check_public_hostname "$host"
done
else
echo ""
echo -e " ${YELLOW}[SKIP]${NC} cloudflared config not found at $CLOUDFLARED_CONFIG"
fi
echo ""
echo "GPU Server (192.168.178.11, LAN-only probe):"
# Direct LAN check — catches GPU box being down even when the public
# tunnel hostnames in the ingress walk above happen to time out or
# return cached errors.
check_service "GPU Ollama" "http://192.168.178.11:11434/api/version" 3
check_service "GPU STT" "http://192.168.178.11:3020/health" 3
check_service "GPU TTS" "http://192.168.178.11:3022/health" 3
check_service "GPU Image Gen" "http://192.168.178.11:3023/health" 3
# GPU Video Gen (LTX) is intentionally not probed — it's planned but
# not deployed yet, so its absence is expected and shouldn't page.
echo ""
echo "Monitoring (LAN, not exposed via tunnel):"
check_service "VictoriaMetrics" "http://localhost:9090/health"
echo ""
echo "Alerting:"
check_service "vmalert" "http://localhost:8880/health"
check_service "Alertmanager" "http://localhost:9093/-/healthy"
check_service "Alert Notifier" "http://localhost:9095/health"
echo ""
echo "Disk Space:"
check_disk() {
local name=$1
local path=$2
local warn_pct=${3:-80}
local crit_pct=${4:-90}
if [ ! -d "$path" ]; then
echo -e " ${YELLOW}[SKIP]${NC} $name ($path not found)"
return 0
fi
local usage_pct=$(df "$path" | tail -1 | awk '{gsub(/%/,""); print $5}')
local avail=$(df -h "$path" | tail -1 | awk '{print $4}')
if [ "$usage_pct" -ge "$crit_pct" ]; then
echo -e " ${RED}[CRIT]${NC} $name: ${usage_pct}% used ($avail free)"
FAILURES+=("Disk $name: ${usage_pct}% (critical)")
elif [ "$usage_pct" -ge "$warn_pct" ]; then
echo -e " ${YELLOW}[WARN]${NC} $name: ${usage_pct}% used ($avail free)"
FAILURES+=("Disk $name: ${usage_pct}% (warning)")
else
echo -e " ${GREEN}[OK]${NC} $name: ${usage_pct}% used ($avail free)"
fi
}
check_disk "System (/)" "/"
check_disk "ManaData" "/Volumes/ManaData"
echo ""
echo "Cloudflare Tunnel:"
if pgrep -x "cloudflared" >/dev/null; then
echo -e " ${GREEN}[OK]${NC} cloudflared running"
else
echo -e " ${RED}[FAIL]${NC} cloudflared not running"
FAILURES+=("cloudflared")
fi
echo ""
echo "=== Summary ==="
if [ ${#FAILURES[@]} -eq 0 ]; then
echo -e "${GREEN}All services healthy!${NC}"
exit 0
else
echo -e "${RED}Failed services (${#FAILURES[@]}):${NC}"
FAILED_LIST=""
for f in "${FAILURES[@]}"; do
echo " - $f"
FAILED_LIST="${FAILED_LIST}- ${f}\n"
done
# Send notifications
send_all_notifications "$(echo -e "$FAILED_LIST")"
exit 1
fi