Monitoring system health by using the mmhealth command

The mmhealth command monitors the health status of a node and services that are hosted on the node. You can use the mmhealth command to view the health status of a whole cluster in a single view.

Every service hosted on an IBM Spectrum Scale™ node has its own health monitoring service. All the sub-components like the filesystem or network interfaces are monitored through the monitoring service of their main component. Only the sub-components of CES service such as NFS, SMB, Object, and authentication have their own health monitors. The mmhealth command gets the health details from these monitoring services. The role of a node in monitoring determines the components that need to be monitored. This is an internal node role and a node can have more than one role. For example, a CES node can also be a node with file systems and performance monitoring. The role of the node also determines the monitoring service that is required on a specific node. For example, you do not need a CES monitoring on a non-CES node. The monitoring services are only started if a specific node role is assigned to the node. Every monitoring service includes at least one monitor.

The following criteria must be met to use the health monitoring functionality on your GPFS™ cluster:
  • Only Linux and AIX® nodes are supported.
  • Only GPFS monitoring is supported on AIX.
  • The AIX nodes must have the Python 2.7.5 installed.
  • The cluster must have the minimum release level as 4.2.2.0 or higher to use mmhealth cluster show command.