Skip to main content

Supported technologies

Everything the agent runs on and everything it knows how to monitor, on one page. Each area links to its detailed page, and every metric behind these rows is listed in the metrics reference.

Operating systems​

One static binary per platform, amd64 and arm64, installed by the same one-liner everywhere. See Installation.

PlatformCoveredNotes
Debian, Ubuntuyesapt inventory and patch state.
RHEL, Rocky, AlmaLinux, CentOS Streamyesrpm inventory, dnf patch state, local CVE counts.
Fedorayesdnf5 patch state.
openSUSE, SLESyeszypper patch state.
Alpineyesmusl build, apk inventory, OpenRC service.
Devuan and other sysvinit systemsyesNo systemd required, init-aware service and self-update.
Windows (Server and desktop)yesNative service (SCM), Event Log, IIS, MSSQL, Hyper-V.
macOSyesIntel and Apple Silicon, launchd daemon.

Any other mainstream Linux with systemd, OpenRC or sysvinit is expected to work: host metrics need only /proc, and discovery understands dpkg, rpm and apk databases.

Web servers and websites​

TechnologyCoveredNotes
nginxyesPer-site checks, SSL, DNS liveness, domain lifecycle from nginx -T.
Apache httpdyesBoth Debian-family apache2 and RHEL-family httpd layouts.
IISyesSites, app pools, worker processes; sites are the same billing unit.
WordPressyesDetected per site; drives the WP-aware APM attribution.

Sites are discovered from the web server's real configuration, so non-standard layouts are found rather than assumed. See Websites and domains.

Services​

The signed catalog today, by category. New services are added server-side, with no agent update. See Service catalog; metric lists are in the reference.

CategoryCovered today
Databases, search, analyticsMySQL, MariaDB, PostgreSQL, Redis, MongoDB, MSSQL (default and named instances), Elasticsearch, ClickHouse
MailPostfix, Dovecot, Exim, Proxmox Mail Gateway
DNSBIND, PowerDNS, Unbound
Queues and brokersRabbitMQ, Kafka, NATS
Load balancers and proxiesHAProxy, Varnish, Traefik, keepalived
VirtualizationProxmox VE (single nodes and clusters: quorum, per-node guests, migrations and backups as events), KVM/libvirt, Hyper-V
Replicated storageDRBD (resources, devices, connections, out-of-sync and stalled resyncs), LINSTOR (nodes, storage pools, error reports), corosync
ContainersDocker (grouped by image), Kubernetes (DaemonSet)
FTPvsftpd, Pure-FTPd
PHPphp-fpm (including several versions side by side) and mod_php
Node.jssystemd, pm2 and Windows services: up/down, crash restarts, logs
OS healthsystemd state, cron activity, pending updates and security patches per package manager, reboot-required
Security signalsSSH brute-force with attacker sources, root logins, sshd config doors, OOM kills, file integrity, open ports, CIS-subset hardening scan
Backupsborg, restic, any artifact path by age

Anything exposing a Prometheus endpoint can be scraped by the prom probe mode, and apps can push their own app.* metrics to the local ingest port, so coverage does not end at this table.

APM and RUM​

Full details in APM and RUM.

RuntimeCoveredNotes
PHP 8.1 to 8.5yesphp-fpm and mod_php, glibc and musl, multiple versions on one host. Prebuilt signed artifacts, staged activation; the CPU profiler knows 8.5 as well.
WordPressyesHook-level spans (deep trace), N+1 insights, per-plugin CPU attribution.
PHP below 8.1partialZero-touch RED metrics and everything non-APM still work; method-level spans need 8.1+.
.NETyesASP.NET Core under systemd and IIS app pools, Windows and Linux, CPU flamegraphs via EventPipe.
Node.jsyesNode 18.19+ under systemd, pm2 or a Windows service, provisioned and staged like PHP; bare launches get a ready recipe. Runtime metrics (event loop, GC, heap) and per-service telemetry included; CPU flamegraph via an opt-in grant on Linux.
Java, Python, GopartialBring your own OpenTelemetry SDK; the agent's local OTLP receiver accepts the spans into the same pipeline.
CPU profiler (PHP)yesNon-thread-safe php-fpm builds, from outside the process. ZTS variant and arm64 artifacts are planned.
RUMyesAuto-injected on instrumented PHP sites; any other site via the hosted snippet.

Layer-1 RED needs eBPF (VM, bare metal, or a privileged container); in unprivileged containers the same numbers come from access logs automatically.

Kubernetes​

k3s clusters are covered end to end (that is where it runs live); kubeadm and k0s kubeconfig layouts are read when the agent runs on the node. Per-namespace and per-pod usage needs the metrics-server. See Kubernetes.

Network devices (SNMP)​

Agentless, through a collector host. See SNMP and network.

CapabilityCoveredNotes
SNMP v1, v2c, v3yesv3 USM with auth and privacy; credentials stay on the collector.
Standard MIBsyesIF-MIB (ports, traffic, errors), UPS-MIB, HOST-RESOURCES-MIB, ENTITY-SENSOR-MIB (temperatures, fans, PSU, optical dBm), PoE tables.
Vendor bundlesyesMikroTik health, APC PowerNet; more are just signed definitions.
Traps and informsyesv2c traps and v3 informs, with anti-spoof source filtering and a confirming poll.
SyslogyesCollector-side listener with severity rates and pattern counters.
Subnet discoveryyesScan of an approved range into candidates, never automatic polling.
TopologyyesLLDP and MikroTik neighbor tables, FDB, ARP, VLANs; map, MAC search and address book in the console.
Traffic flowsyesNetFlow v5 and v9, IPFIX; aggregated on the collector, raw flows never leave the site. sFlow planned.
Config backupyesOver SSH from the collector; RouterOS, Cisco IOS, Junos, FortiGate, EdgeOS recipes; versions and diffs in the console.

Not yet, honestly​

Planned

sFlow ingestion, arm64 APM artifacts, the ZTS variant of the PHP profiler, and a .deb package are planned and marked on their pages. If it is not marked Planned in this documentation, it is live.