Port monitoring

A lot of production infrastructure has no URL. Mail servers, databases, game servers, VPN concentrators, SSH bastions, message brokers — if it speaks TCP or UDP but not HTTP, a website checker is blind to it. Port monitoring closes that gap: connect to the socket, verify the service actually answers like itself, and alert when it doesn't.

The second half of that sentence matters. A port can accept connections while the service behind it is wedged — the process is alive enough for the TCP handshake but too broken to serve. That's why FourSight validates what the service says, not just whether the socket opens.

Anything with a listening socket. The most common production uses:

Most classic protocols announce themselves the moment you connect: an SMTP server says "220", SSH says "SSH-2.0-…", IMAP says "* OK". FourSight's banner validation reads that greeting and matches it against a regex — so a check only passes when the actual service answers, not merely the operating system's TCP stack. A wedged mail daemon whose port still accepts connections fails the check, which is exactly the failure other monitors miss.

For services that speak only when spoken to, send/expect mode transmits a string you define and matches the response against a regex — enough to health-check Redis with a PING, poke a custom line protocol, or validate a proprietary daemon's handshake. TLS and STARTTLS are supported, so implicit-TLS services like IMAPS (993) and opportunistic-TLS setups on SMTP are both first-class. UDP checks are available too, including a real DNS query mode for validating resolvers on port 53.

Built-in service presets (all editable)

Preset	Port / transport	Validation
SMTP	25 / TCP	Banner matches ^220
POP3	110 / TCP	Banner matches ^\+OK
IMAP	143 / TCP	Banner matches ^\* OK
IMAPS	993 / TCP + TLS	Banner matches ^\* OK over TLS
FTP	21 / TCP	Banner matches ^220
SSH	22 / TCP	Banner matches ^SSH-
DNS	53 / UDP	Real DNS query answered
Custom	Any / TCP or UDP	Connect-only, banner, or send/expect regex

Because connectivity is a property of the path, not just the server. A firewall rule change, a cloud security group edit, a BGP hiccup, or an overzealous DDoS mitigation can leave your port reachable from one continent and dead from another — while a single-vantage monitor happily reports green. FourSight probes every port monitor from four regions (US, Canada, Europe, APAC) and requires quorum agreement before opening an incident.

That design cuts both ways in your favor: a genuine regional block is visible in the per-region results, while a transient blip on one probe path doesn't page you. For mail servers especially — where an unreachable port silently defers inbound mail rather than bouncing it loudly — region-level visibility is the difference between a five-minute fix and discovering three days of queued email.

Use HTTP checks for anything that speaks HTTP — you get status codes, response-time data, and keyword validation for free. Use port checks for everything else: non-HTTP protocols, or cases where you specifically want to verify the listener layer (is sshd up?) independent of application behavior.

The practical pattern for a typical stack: HTTP checks on the app and API, SSL check on the certificate, DNS checks on the records, a port check each on the database, mail, and VPN, heartbeats on the cron jobs, and a domain expiry check on the root — eight monitor types, one dashboard, one escalation policy. All eight unlock together on the Growth plan at $40/mo flat, so covering the non-web half of your infrastructure doesn't require a second tool.

Tip: Timeouts are configurable per monitor (default 8 seconds). For latency-sensitive services, tighten the timeout so a degraded-but-listening service fails the check instead of limping past it.

Included in Growth — $40/mo flat

Cover the half of your stack that has no URL

Port monitoring is included in the Growth plan with all eight check types — TCP and UDP, banner and send/expect validation, presets for the classic protocols, and 4-region quorum before any alert.

• TCP and UDP checks against any host and port
• Banner regex, send/expect probes, TLS and STARTTLS support
• Presets for SMTP, POP3, IMAP, IMAPS, FTP, SSH, and DNS
• Per-region connection evidence on every incident

FAQ