The humio-metrics Repository

The humio-metric repository is a system repository that holds metrics generated by LogScale itself. There are metrics on both system-wide scope, such as total number of live queries active, and metrics for each repository.

Note

Special access restrictions apply: a user can get access to search this repository using the same set of rules as any other repository. But any user who does not have access through those rules can search this repo, being restricted to searching only the events that directly relate to other repositories, that said user has access to. You can thus not see what other repositories may exist in LogScale through this access, and neither can you query the "global" metrics without being a member of the repository.

By default, a user can only query the humio-metrics repository for the repositories they can search. A user must be a member of the humio-metrics repository to query about all repositories.

The metrics logged into this repository are also available on JMX and Prometheus endpoints provided those are enabled. By default those listeners are disabled, whereas logging to this repository is always enabled.

On cloud, the metrics for an organization can be accessed through the humio-organization-metrics view. An organization root can query metrics for all repositories in the organization.

Metric Structure

An event within the humio-metric repository will include the following core fields:

  • name

    The name of the metric being reported. A list of possible metrics is shown in Node-Level Metrics.

  • type

    The type of the metric, which provides an indication of the other fields that will be included within the event. For more information, see Metric Types.

Additional fields within an event will depend on the underlying type.

Metric Types

There are two major types of metrics in LogScale; node level metrics and object level metrics. Objects include repositories, ingest listeners, or storage partitions. For example ingest metrics for a given repository can be obtained by viewing ingest-bytes.

Within the individual metric types there are a different event entries for different metrics that include different fields. The underlying type can be identified using the type field:

Counter

The Counter event is a count of the corresponding metric event as an integer. A metric event of this type may contain the following fields:

Field Type Description
count Integer An incremental count of the metric.
Gauge

A Gauge event stores either a value or ratio. A metric event of this type may contain the following fields:

Field Type Description
value Integer or Float Any value appropriate for the metric.
Histogram

The Histogram type tracks counter and statistical information for a given metric within a specific time window (typically 1 minute).

Field Type Description
count Integer Counter of the number of events since LogScale was started
max Float Maximum value of the metric within the time period
mean Float Mean value of the metric within the time period
median Float Median value of the metric within the time period
min Float Minimum value of the metric within the time period
p75 Float 75th percentile of the metric within the time period
p95 Float 95th percentile of the metric within the time period
p98 Float 98th percentile of the metric within the time period
p99 Float 99th percentile of the metric within the time period
p999 Float 99.9th percentile of the metric within the time period
rate_unit String A text description of the rate unit for the values.
stddev Float Standard deviation of the metric within the time period
window_count Float Counter of the number of events within the time period
Meter

A Meter measures the rate of a given metric, and includes moving averages and the rate description. A metric event of this type may contain the following fields:

Field Type Description
count Integer Count of the corresponding metric
m1 Float 1 minute moving average
m15 Float 15 minute moving average
m5 Float 5 minute moving average
mean_rate String Moving mean rate (per minute)
Timer

A Timer metric measures the duration of a given metric event, and includes moving averages, statistical values and when appropriate a measured rate. A metric event of this type may contain the following fields:

Field Type Description
count Integer Count of the corresponding metric
duration_unit String A description of the unit used to track the duration.
m1 Float 1 minute moving average
m15 Float 15 minute moving average
m5 Float 5 minute moving average
mean_rate String A text description of the rate unit for the values.
max Float Maximum value of the metric within the time period
mean Float Mean value of the metric within the time period
mean_rate Float Mean rate of the metric within the time period
median Float Median value of the metric within the time period
min Float Minimum value of the metric within the time period
p75 Float 75th percentile of the metric within the time period
p95 Float 95th percentile of the metric within the time period
p98 Float 98th percentile of the metric within the time period
p99 Float 99th percentile of the metric within the time period
p999 Float 99.9th percentile of the metric within the time period
rate_unit String A text description of the rate unit for the values.
stddev Float Standard deviation of the metric within the time period