Lakehouse Admin
df5ec93dc3
- Trino catalogs, Grafana, Spark jobs, LDAP LDIF, NPM compose - Airflow DAG scripts, Proxmox VM inventory, network docs - Ansible playbook, Gitea CI validate workflow - Backup and health-check scripts, cron documentation - Homepage DOCS tab with links to all documentation - Extended collect-fleet-config.sh and populate-repo.py
35 lines
1.3 KiB
Bash
Executable File
35 lines
1.3 KiB
Bash
Executable File
#!/usr/bin/env bash
|
|
# Health check all ATC lab HTTP endpoints (from homepage siteMonitors).
|
|
set -u
|
|
|
|
FAIL=0
|
|
check() {
|
|
local name="$1" url="$2"
|
|
local code
|
|
code=$(curl -sk -o /dev/null -w "%{http_code}" --connect-timeout 5 --max-time 10 "$url" 2>/dev/null || echo "000")
|
|
if [[ "$code" =~ ^(200|301|302|401|403)$ ]]; then
|
|
printf "OK %-28s %s (%s)\n" "$name" "$url" "$code"
|
|
else
|
|
printf "FAIL %-28s %s (%s)\n" "$name" "$url" "$code"
|
|
FAIL=$((FAIL + 1))
|
|
fi
|
|
}
|
|
|
|
check "Homepage" "http://atc-docker01.dell-atc.lan/"
|
|
check "RSS proxy" "http://atc-docker01.dell-atc.lan:8090/health"
|
|
check "Superset" "http://atc-docker01.dell-atc.lan:8088/health"
|
|
check "Kafka UI" "http://atc-kafka01.dell-atc.lan:9000/"
|
|
check "Debezium" "http://atc-lake01.dell-atc.lan:8083/"
|
|
check "Spark" "http://atc-lake01.dell-atc.lan:8080/"
|
|
check "Trino" "http://atc-lake01.dell-atc.lan:8089/ui/"
|
|
check "Airflow" "http://10.0.21.55:8080/health"
|
|
check "Kibana" "http://atc-elastic01.dell-atc.lan:5601/"
|
|
check "Grafana" "http://atc-grafana.dell-atc.lan:3000/api/health"
|
|
check "Forgejo" "http://atc-mgt01.dell-atc.lan:3001/"
|
|
check "ObjectScale" "https://10.0.20.111/"
|
|
check "Proxmox" "https://10.0.10.65:8006/"
|
|
|
|
echo "---"
|
|
[[ $FAIL -eq 0 ]] && echo "All checks passed" && exit 0
|
|
echo "$FAIL check(s) failed" && exit 1
|