Port monitoring for services that don't speak HTTP
Your database, mail server, and SSH box can go down without a single web request failing. A TCP connect check every minute tells you they're accepting connections, and alerts you the moment they're not.
Any TCP service
Databases, mail servers, game servers, SSH: if it accepts a TCP connection, you can monitor it. No HTTP endpoint required.
Host, port, timeout: that's it
A TCP monitor takes three settings. No probes to script, no agents to install: point it at a host and port and you're done.
Checks every minute
Connections are attempted every minute by default, and the interval is configurable per monitor. Results show as up or down.
Alerts when connections fail
A refused or timed-out connection marks the monitor down and triggers SMS, email, or Discord alerts, with recovery notices when it's back.
How it works
- 1
Add a TCP monitor
Enter the host, the port your service listens on, and a connection timeout.
- 2
Connections are checked every minute
Infra Agent opens a TCP connection on schedule. Accepted means up; refused or timed out means down.
- 3
Get alerted
Route down alerts and recovery notices per monitor to SMS, email, or Discord.
Frequently asked questions
What can I monitor with a TCP port check?
Anything that speaks TCP but not HTTP: Postgres and other databases, SMTP and IMAP mail servers, game servers, SSH, Redis, message queues, and any other service that listens on a port. If it accepts connections, a TCP monitor can watch it.
What do I need to configure for a port monitor?
Just three things: a host, a port, and a timeout. There are no scripts to write and no agents to install; Infra Agent attempts the connection from the outside, the same way your clients do.
How often is the port checked?
Every minute by default, and you can configure the interval per monitor. Each check records an up or down result, so you know within about a minute when a port stops accepting connections.
What happens when a connection is refused or times out?
The monitor goes down, an incident is tracked from the first failed check to recovery (including duration and the last error), and your alerts fire. Choose SMS, email, or Discord per monitor, for down alerts, recovery notices, or both.