Pular para o conteúdo

Nutanix AHV

Este conteúdo não está disponível em sua língua ainda.

Nutanix clusters, read through the v4 REST API of Prism Central. One Prism Central covers every cluster registered to it.

From the DumbMonit server (or its relay agent) to the device:

PortProtocolPurpose
9440TCPPrism Central API (HTTPS, default)

Every minute DumbMonit reads, with HTTP basic authentication on port 9440:

WhatPath
Clusters/api/clustermgmt/v4.0/config/clusters
Hosts/api/clustermgmt/v4.0/config/hosts
AHV VMs/api/vmm/v4.0/ahv/config/vms
Storage containers/api/clustermgmt/v4.0/config/storage-containers
Fill level of each container/api/clustermgmt/v4.0/stats/storage-containers/{extId}
Unresolved alerts/api/monitoring/v4.0/serviceability/alerts

Lists are read 100 items per page, up to 1,000 items each. The fill level and the alerts are optional: if the account’s role refuses them, the rest is still collected. Nutanix declared the v1, v2.0 and v3 APIs deprecated; only v4 keeps evolving, and it needs Prism Central pc.2024.3 or later.

All metrics are prefixed dumbmonit_nutanix_.

MetricLabelsMeaning
clustersClusters, without Prism Central itself.
cluster_infocluster, versionValue 1: the AOS version of each cluster.
cluster_availablecluster1 when Prism Central reaches the cluster.
hostsHosts of every cluster.
host_infohost, cluster, hypervisorValue 1: the hypervisor each host runs.
host_cpu_cores, host_memory_byteshost, clusterCapacity.
host_degradedhost, cluster1 when Nutanix marks the host as degraded.
host_maintenancehost, cluster1 when the host is entering or in maintenance mode.
vms, vms_runningAHV VMs, and how many are powered on.
vm_power_state_infovm, cluster, stateValue 1: on, off, paused or undetermined.
vm_runningvm, cluster1 when the VM is powered on.
vm_vcpus, vm_memory_bytesvm, clusterSockets × cores × threads, and configured memory.
storage_containersStorage containers, without Nutanix’s internal ones.
storage_container_capacity_bytes, storage_container_used_bytes, storage_container_used_percentcontainer, clusterFill level, from the latest statistics of the last thirty minutes.
alerts_unresolved, alerts_critical, alerts_warningUnresolved alerts, all and by severity.

The built-in rules that apply:

  • Nutanix host degraded: a host is marked degraded for five minutes (Warning).
  • Nutanix storage container almost full: a container is 90% full for fifteen minutes (Advisory).
  • Nutanix critical alerts: Prism Central reports at least one unresolved critical alert for five minutes (Warning).
  1. In Prism Central, open Admin Center → IAM → Identities and add a local user with a long random password.

    dumbmonit
  2. Open Admin Center → IAM → Authorization Policies and create a policy with the Prism Viewer role, for all entities, assigned to that user.

  3. In DumbMonit, enter the Prism Central address (for example prism-central.lan) with that user name and password.

  • Prism Viewer, all entities: Sees clusters, hosts, VMs, storage containers, alerts

Vendor documentation: Nutanix developer portal

Prism Viewer reads clusters, hosts, VMs, storage containers and alerts, and changes nothing. One Prism Central reports every cluster registered to it.

The v4 API is served by Prism Central pc.2024.3 or later; a cluster managed by Prism Element alone, or by an older Prism Central, is not supported. DumbMonit only reads the v4 API (clustermgmt, vmm and monitoring, version 4.0): it never powers a VM on or off, never resolves an alert and never changes a storage container.

CredentialFields
User name and passwordThe dumbmonit local user with the Prism Viewer role.

Address: the host name or IP of Prism Central, host:port, or a URL. Options:

  • Port (9440).
  • Accept an unverifiable certificate: Prism Central’s own certificate is self-signed.
  • Timeout per request (15 s).

“Nutanix Prism Central refused the credentials … (401)”: wrong user name or password.

“Nutanix Prism Central has no /api/clustermgmt/v4.0/…”: the address is a Prism Element, or a Prism Central older than pc.2024.3.

No storage_container_used_percent: the role does not allow reading statistics; the containers are still counted, with their configured capacity.