The Databases page brings every database server in your fleet into one operational surface. Each row is one host running one or more database engines that the agent auto-discovered.

Open it from the sidebar under Monitoring → Databases, or at /databases.

What you see

The list contains every host where the agent has found MySQL, MariaDB, PostgreSQL, Redis, MongoDB, Elasticsearch, Memcached, CockroachDB, or ClickHouse — either via the running-services scan or via metric plugins reporting engine data.

Per row:

• Hostname with online / offline / pending status.
• Engines — one badge per detected engine.
• Latest metrics — RAM, disk, and connections for the primary engine on that host.
• Recent backups — number and freshness of database backups.
• Alerts — active alert count linking through to the Alerts page.

Click a row for the database detail page.

Database detail

The detail page opens at /databases/{server} with tabs for each engine detected on that host. Common sections:

• Runtime metrics — connections, active queries, database size, transactions per second (Postgres); questions, slow queries, threads (MySQL); ops/sec, hit rate, memory, connected clients (Redis).
• Charts — 24 h rolling charts pulled per engine.
• Databases / keyspaces — list of user databases with size.
• Backups — latest database dumps, with age and size.

Two dedicated sub-pages open from the detail header:

• Slow queries — at /databases/{server}/slow-queries. Surfaces the top slow queries captured by the engine (pg_stat_statements for Postgres, the slow log for MySQL / MariaDB). Sortable and filterable so you can see the biggest offender by mean time or total time.
• Lock waits — at /databases/{server}/lock-waits. Two tabs: Live shows the current blocking chains, and History shows lock events longer than five seconds captured over time. Each event includes the query being blocked, the blocker, the wait time, and (where available) the user and database.

Note

Slow queries and lock-wait monitoring depend on the engine being configured to expose that data. For Postgres this means enabling pg_stat_statements; for MySQL / MariaDB it means turning on the slow query log. HostAtlas will tell you when the required setup is missing.

What you can do

From the detail page:

• Run backup — triggers an on-demand dump of the selected database. The agent runs mysqldump / pg_dump / equivalent and stores the artifact locally.
• Download / delete backup — from the backups list.
• Ask the AI assistant to optimize — opens the AI Assistant with a preloaded question about the server’s slowest query and current runtime metrics.
• Kill a query (lock-waits page) — sends a kill command to the engine, scoped to the offending process.

Notifications

Database alerts flow through Alert Rules — pick from metric-based conditions (connections used, replication lag, slow query rate), lock-wait duration thresholds, and backup age. Route to any configured notification channel.

Related

• Servers — the host inventory. Every database server is also a server.
• Backups — file-level backup path monitoring for the dumps you already write to disk.
• Backup Runs — for hosts using the standalone hostatlas-backup CLI.
• Slow Requests — the app-layer counterpart to database slow queries.