PostgreSQL, MySQL/MariaDB and MongoDB uptime monitoring — a real connection and a real query from our European regions, every 30 seconds to 5 minutes, with the credentials kept encrypted.
SELECT count(*) FROM jobs WHERE status = 'stuck' < 10 · 7 ✓postgres://, mysql:// or mongodb:// — a replica-set seed list works too. Any user:password in it is moved to encrypted fields; the URL is stored without it.
SELECT 1 or a ping by default, or your own — a stuck-jobs count, a replication lag, a replica-set status — and the value it must return.
Connection refused, authentication failed, query timing out or a result off its threshold — the same alert channels and escalation as every other monitor.
PostgreSQL, MySQL and MariaDB over their native protocols, MongoDB standalone or replica set (mongodb:// and mongodb+srv://), TLS on or off.
User and password are encrypted at rest and masked everywhere — API, exports, alerts, audit log, public status pages. Use a read-only user.
Connect + query time is recorded on every check, so a slow database shows up as a trend long before it becomes an outage.
A database that is not exposed to the internet is checked from a private agent running inside your network — nothing to open.