Understanding Thresholds
How thresholds work for monitoring services
Certain service thresholds may be displayed on the Service Status Detail for All Hosts view of the Health tab. Each service thresholds subtopic provides information to help understand the different monitoring states. The services covered in the subtopics include current load, current users, network statistics, ping node, root partition, RAM usage, swap usage, total processes, CPU/memory usage, Lustre health, arrays and disk status, SES sensors, and FRUs.
Note: Use CSCLI to make changes to some thresholds.
Definitions
The following terms are used throughout the thresholds subtopics:
- mgmt
- The management node—for Icinga "management" is the localhost.
- all
- All physical nodes (except virtual enclosure nodes)
- all-but-mgmt
- All physical nodes except management nodes—for Icinga those are remote nodes.
- oss
- Object Storage Server (OSS)
- mds
- Metadata Server (MDS)
- enclosure
- A virtual node for the enclosures—they are virtual because there is no physical server that can be pinged