Skip to main content

Metrics Collection Settings

You can customize how MetricsHub collects and processes metrics at different levels: globally, per resource group, or per resource.

Scheduling and Timing​

Collection Period​

MetricsHub collects metrics every minute by default. Use collectPeriod to change this:

# Global
collectPeriod: 2m

# Per resource
resources:
  myHost1:
    collectPeriod: 1m30s
warning

Collecting too frequently increases CPU load.

Discovery Cycle​

MetricsHub runs discovery every 30 collection cycles by default. Use discoveryCycle to change this:

# Global
discoveryCycle: 15

# Per resource
resources:
  myHost1:
    discoveryCycle: 5
warning

Running a discovery too frequently increases CPU load.

Duration Format​

All timeout and period values support these units:

UnitDescriptionExamples
sseconds120s
mminutes2m, 1m30s
hhours1h, 1h30m
ddays1d

Filtering​

Use monitorFilters to include or exclude specific monitor types and control your telemetry volume:

# Include only specific monitors
monitorFilters: [ +enclosure, +fan, +power_supply ]

# Exclude volumes
monitorFilters: [ "!volume" ]
warning

Critical monitors (batteries, power supplies, CPUs, fans, memory) should never be excluded.

Resource Identity​

Custom Hostname​

You can differentiate the hostname used for collection from the one reported in metrics:

resources:
  myHost1:
    attributes:
      host.name: reported-hostname # In metrics
    protocols:
      snmp:
        hostname: actual-ip-address # For collection

FQDN Resolution​

Set resolveHostnameToFqdn to true to resolve hostnames to fully qualified domain names:

# Global
resolveHostnameToFqdn: true

# Per resource
resources:
  myHost1:
    resolveHostnameToFqdn: true
warning

Resolution failures may change the host.name attribute and affect metric identity.

Custom Attributes​

Add labels to extend the Host Resource attributes:

resources:
  myHost1:
    attributes:
      host.name: my-host-01
      host.type: win
      app: Jenkins # Custom
      environment: prod # Custom

Performance​

Parallel Jobs​

Use jobPoolSize to control concurrent job execution (default: 20):

jobPoolSize: 40
warning

Too many parallel jobs may cause OutOfMemory errors.

Sequential Mode​

Use sequential to force serial execution for sensitive targets:

resources:
  myHost1:
    sequential: true
warning

Significantly slower. Use only when required by target system limitations.

StateSet Compression​

Use stateSetCompression to control how health status metrics are reported:

stateSetCompression: suppressZeros  # Default: only report non-zero states
stateSetCompression: none           # Report all states

Self-Monitoring​

Set enableSelfMonitoring to true to track MetricsHub's own performance:

enableSelfMonitoring: true

When enabled, MetricsHub emits metricshub.job.duration metrics for each job.