MetricHarbor
Start monitoring

AGENT-BASED INFRASTRUCTURE MONITORING

Know what's running. Know what's gone quiet.

Monitor Linux servers, databases, Kubernetes and devices from one place. No inbound firewall rules. No credentials on our servers. No guessing when an agent stops reporting.

Start monitoring How it works

Installs from apt or dnf · Outbound HTTPS only

Illustrative dashboard · sample data
Keep the firewall closedAgents connect outbound over HTTPS
Credentials remain yoursSecrets stay encrypted on your agent
Built for interruptionsLocal buffering during outages

A clearer kind of monitoring

Everything you need to notice. Nothing you need to expose.

Designed for the people responsible for keeping things running — and for knowing when they aren't.

01 / SECURITY FIRST

Your network stays yours.

The agent initiates outbound HTTPS connections and listens on nothing. No incoming ports, no firewall exceptions, and no database passwords or SSH keys stored in MetricHarbor — it keeps the name of a credential and the shape of it, never the secret.

02 / RESILIENT COLLECTION

A connection drop isn't a blind spot.

The agent schedules its own checks and buffers readings to disk when it cannot reach us — 64 MB and 72 hours by default, both adjustable. A design where the cloud drives every poll stops monitoring exactly when you need it most.

03 / EXPLICIT SIGNALS

Silence is an alert.

When an agent stops reporting, every check that depended on it stops too — so the absence of other alerts proves nothing. An agent going quiet is itself the loudest thing on the dashboard, never a blank space that looks like health.

Simple architecture

Monitoring without opening the door.

Put a small agent inside your network. It collects locally, keeps secrets on your infrastructure, and sends readings out through a connection it opens itself.

  • No inbound rules and no listening ports
  • No database passwords or SSH keys on the platform
  • Credentials encrypted at rest, on your own machine
  • Runs as an unprivileged user with one Linux capability
How information flows
Your infrastructure Linux, databases, Kubernetes and devices

Local MetricHarbor agent
OUTBOUND HTTPS ONLY →
MetricHarbor Central dashboards
Monitoring status
Alerts and graphs
🔒 Credentials stay on your side of the connection.

Broad visibility

One place for the systems that matter.

Whether you're watching a handful of servers or a mixed estate, the important details should be easy to find.

Linux hosts

Reached over SSH. Reads /proc, runs smartctl and mdadm, and any command you choose.

PostgreSQL

Connections, cache hit ratio, dead tuples, replication lag and database size.

MySQL and MariaDB

Connections, buffer pool, slow queries and replication status.

Redis and Valkey

Memory, clients, hit rate, evictions and command throughput.

Kubernetes

Node, pod and namespace metrics, labelled per pod, alongside the hosts underneath.

TLS certificates

Days to expiry, issuer, key size and chain validity, for any endpoint.

Network devices

Switches, routers, firewalls and appliances, reached by ICMP and SNMP.

Your first agent

Add the repository once. Install the agent anywhere.

Upgrades then arrive through the same tooling as everything else on the machine. Debian and Ubuntu below; RHEL, Rocky, Alma, CentOS Stream, Fedora and openSUSE install from the matching dnf repository.

Get an enrolment key

The full installation guide: every distribution, the tarball and the container.

1 · add the repository, once per machine
curl -fsSL https://apt.metricharbor.io/metricharbor.asc | gpg --dearmor -o /usr/share/keyrings/metricharbor.gpg
echo "deb [signed-by=/usr/share/keyrings/metricharbor.gpg] https://apt.metricharbor.io stable main" > /etc/apt/sources.list.d/metricharbor.list
apt update
Run as root. These pages are written without sudo, which is not installed on a minimal Debian.
2 · install and enrol
apt install metricharbor-agent
metricharbor-agent enrol --url https://api.metricharbor.io --key mhk_… --name db-primary
runuser -u metricharbor -- metricharbor-agent setup
systemctl enable --now metricharbor-agent
Replace mhk_… with an enrolment key from Fleet → Enrolment keys, and db-primary with a name you'll recognise in the fleet view. The service runs as the unprivileged metricharbor user and holds one Linux capability, CAP_NET_RAW, so that ICMP works.

Stay informed

Stop wondering whether silence means everything is fine.

Get a clear view of what's working, what's failing, and what has stopped checking in.

Start with your first agent