Harvester Special Agent

Community MKP for monitoring Harvester and Harvester-based SUSE Virtualization environments with Checkmk.

This package adds a Checkmk special agent for Harvester platform, Longhorn storage and Harvester node monitoring. It collects data from the Kubernetes API, Harvester custom resources, Longhorn custom resources, Prometheus metrics and, optionally, a node-local DaemonSet collector.

It is intended for Harvester platform monitoring. For KubeVirt VM monitoring, install and configure the separate checkmk-kubevirt extension in addition.

What you get

• Harvester version and settings monitoring
• Harvester backup target status
• Harvester VM image import status
• Harvester platform pod summary
• Harvester node information and conditions
• Harvester node configuration monitoring
• Harvester block device discovery
• Longhorn node status
• Longhorn disk status and usage
• Longhorn volume robustness and attachment status
• Longhorn replica summary
• Prometheus connection health
• Optional node-local OS context through a DaemonSet collector
• SUSE Linux Micro / SLES identity labels on Harvester node hosts
• Native Checkmk node services for filesystems, CPU load, memory, uptime, interfaces, disk I/O and process discovery when node collector output is enabled

Created services

The package can create source-host services such as:

• Harvester Prometheus connection
• Harvester version
• Harvester settings
• Harvester backup target default
• Harvester platform pods
• Harvester VM image namespace/name
• Longhorn replicas
• Longhorn volumes summary
• Longhorn volume name

On Harvester node piggyback hosts, it can create services such as:

• Harvester node info
• Harvester node conditions
• Harvester node configuration
• Harvester block device /dev/...
• Longhorn node status
• Longhorn disk status disk-name
• Longhorn disk usage disk-name
• Harvester node local OS context
• Harvester node local mounts
• Harvester node local processes
• Filesystem /...
• CPU load
• Memory
• Uptime
• Disk IO SUMMARY
• Interface ...

The exact number of services depends on the Harvester cluster, Longhorn volumes, node count, mounted filesystems and enabled collection options.

Requirements

• Checkmk 2.5
• Harvester cluster API access
• Kubernetes service account / bearer token
• Read access to Harvester and Longhorn resources
• Access to a Prometheus endpoint exposing Kubernetes, Harvester, Longhorn and node metrics
• Optional: node collector DaemonSet from docs/harvester_node_collector.yaml

Installation

Install via mkp add / mkp enable, or use Setup > Extension packages.

After installation, configure the Harvester special agent rule with:

• Kubernetes API endpoint
• Bearer token from the Checkmk password store
• Prometheus endpoint
• Cluster name
• Node piggyback hostname mode
• Optional node collector settings

A minimal Kubernetes RBAC example is included in the repository under docs/harvester_rbac.yaml.

The optional node collector manifest is included under docs/harvester_node_collector.yaml. Deploy it only if you want Checkmk to receive node-local context such as OS release details, selected mounts, selected kernel modules and process data.

Notes

This extension focuses on Harvester platform, Longhorn and Harvester node monitoring. For VM-level KubeVirt monitoring, use the separate checkmk-kubevirt package.