Windows Failover Cluster
Monitor nodes, roles, resources, witness and CSV usage of your Windows Failover Cluster, for example on Hyper-V and HCI hosts.
The Windows Failover Cluster check monitors the local Windows Failover Cluster read-only: nodes, roles, resources, the configured witness and the usage of Cluster Shared Volumes (CSV). It runs on a cluster node on Windows (x64 and Arm64), every 5 minutes by default. The check installs nothing, runs no cluster validation, and does not move or repair roles or resources.
It complements the Microsoft Hyper-V Host and DataCore SANsymphony checks but does not replace them.
Prerequisites
- The octoja agent runs on a node of the failover cluster.
- Windows PowerShell 5.1 and the FailoverClusters PowerShell module are present on the node. The check does not install the module.
- You need the Monitoring Check Management permission.
Steps
- Go to Devices and open a cluster node.
- Click the Checks tab → Add Check → Windows Failover Cluster.
- Under Required roles, enter the roles that must always be online. Leave the field empty if intentionally offline roles are acceptable.
- Adjust the CSV usage thresholds if needed.
- Click Add Check.
To cover several clusters, assign the check through a config package.
Settings
Thresholds are limited to 0–100%. If the warning threshold is higher than the critical threshold, the check swaps the two.
Results
The overall result is the worst status across all detected issues.
- Redirected I/O on a CSV volume alone is allowed, for example during a backup.
- A cluster without a witness (for example with node majority) is valid. An unavailable witness does not by itself mean that the cluster lost quorum; the check does not calculate quorum.
- A cluster without CSV volumes is valid.
The result shows the cluster name, the number of nodes up and in total, and the number of roles, resources and CSV volumes. Below that, each issue is listed with its severity, the affected object, its state and, for CSV usage, the measured value and threshold. Healthy objects are not stored individually.
Limits
The check queries the cluster from the node it runs on. If that node itself is unreachable, it cannot report anything about the rest of the cluster. Also monitor the nodes with Device online.
Was this helpful?
More in Virtualization & containers
Docker ContainerHyper-V VM StatusMicrosoft Hyper-V HostProxmox Cluster & QuorumStill need help? Ask the team