This release contains breaking changes, make sure to read these release notes before updating.
- Support for configuration reload has been greatly improved and several edge cases have been fixed. Anything under
hosts,groups, andmetricscan currently be reloaded during runtime. - There are two new fallbacks for the storage metrics. These will be attempted when there is no information under the normal storage endpoint. See #218 and #212 for more information.
- The released binaries are now built without CGO enabled.
- Several minor fixes for better compatibility with different vendors.
Configuration
The configuration file (and the associated environment variables) has been restructured. You must update your configuration file before upgrading to the new release. See the sample-config.yml for a full overview of the new format.
-
The
authssection has been renamed togroupsand the associated query parameter has also been renamed fromauthtogroup. You will need to adjust your Prometheus configuration accordingly. -
The entire
metricssection has been completely restructured. -
When configuring targets under
hostsandgroups, there is a newhttpsubsection for advanced HTTP client options. Theallow_legacy_rsa_kexoption has been moved here. -
You can now override the global
metricsconfiguration for individual hosts and groups.
The metrics configuration was changed to include all related configuration options in the same section, which made it easier to allow overrides for hosts and groups.
Metrics
There are no breaking changes to existing metrics, but several new metrics have been added to the exporter.
For the manager group there are two new metrics related to the remote console (enabled status and max concurrent sessions). There is also a new option to export the event log from the manager itself. Currently this is only supported on HPE (iLO Event Log) and Dell (Lifecycle Logs). This must be explicitly enabled in the configuration file.
For the sensors group there is a new configuration option, called unified, which will export sensor metrics from another endpoint. When enabling this, the following metrics will be exported instead.
idrac_sensor_info{id,context,name,type,units}
idrac_sensor_health{id,status}
idrac_sensor_value{id}
idrac_sensor_lower_critical{id}
idrac_sensor_upper_critical{id}
When the unified sensors endpoint is unsupported, the exporter will automatically export the legacy sensor metrics instead.
Important
Some testing is advised before enabling this feature in production.
The unified sensor metrics are exported from a more modern Redfish endpoint, which is not supported on older machines. Tests also show that even on some newer machines it does not work properly, often resulting in failed scrapes. Moreover, scraping this endpoint generally takes longer, so you might need to adjust timeout values accordingly.
Full Changelog: v2.6.3...v3.0.0