Splunk Metrics

StackState Self-hosted v5.1.x

Overview

This page describes the Splunk Metrics check running on StackState Agent V2.

If you are currently running the Splunk Metrics check on Agent V1 (legacy), it's advised that you migrate to Agent V2.

When the Splunk StackPack is installed in StackState, you can configure the Splunk Metrics check on StackState Agent V2 to begin collecting Splunk metrics data.

Metrics are collected from Splunk by executing Splunk saved searches that have been specified in the StackState Agent V2 Splunk Metrics check configuration. In order to receive Splunk metrics data in StackState, you will therefore need to add configuration to both Splunk and StackState Agent V2.

  • Splunk saved search - there should be at least one saved search that generates the metrics data you want to retrieve. Each saved search can retrieve one metric.

  • StackState Agent V2 Splunk Metrics check - a Splunk Metrics check should be configured to connect to your Splunk instance and execute the relevant Splunk saved searches.

The Splunk Metrics check on StackState Agent V2 will execute all configured Splunk saved searches periodically. Data will be requested from the last received metric timestamp up until now.

Prerequisites

To run the Splunk Metrics Agent check, you need to have:

Fields used

StackState Agent V2 executes the Splunk saved searches configured in the Splunk Metrics Agent check configuration file and pushes retrieved data to StackState as a telemetry stream. The following fields from the results of a saved search are sent to StackState:

FieldDescription

_time (long)

Data collection timestamp, millis since epoch.

metric (string)

The name of the metric. Taken from the configured metric_name_field.

value (numeric)

The value of the metric. Taken from the configured metric_value_field.

Example Splunk query

index=vms MetricId=cpu.usage.average
| table _time VMName Value    
| eval VMName = upper(VMName)
| rename VMName as metricCpuUsageAverage, Value as valueCpuUsageAverage
| eval type = "CpuUsageAverage"

The example Splunk saved search above would result in the following metric data in StackState:

FieldData

_time

Splunk _time field.

metric

Splunk <metricCpuUsageAverage> field.

value

Splunk <valueCpuUsageAverage> field.

Agent check

Configure the Splunk Metrics check

To enable the Splunk Metrics integration and begin collecting metrics data from your Splunk instance, the Splunk Metrics check must be configured on StackState Agent V2. The check configuration provides all details required for the Agent to connect to your Splunk instance and execute a Splunk saved search.

Example Agent V2 Splunk Metrics check configuration file: splunk_metric/conf.yaml.example (github.com)

To configure the Splunk Metrics Agent check:

  1. Edit the StackState Agent V2 configuration file /etc/stackstate-agent/conf.d/splunk_metric.d/conf.yaml.

  2. Under instances, add details of your Splunk instance:

    • url - The URL of your Splunk instance.

    • authentication - How the Agent should authenticate with your Splunk instance. Choose either token-based (recommended) or basic authentication. For details, see authentication configuration details.

    • tags - Optional. Can be used to apply specific tags to all reported metrics in StackState.

  3. Under saved_searches, add details of each Splunk saved search that the check should execute. Each saved search can retrieve one metric:

    • name - The name of the Splunk saved search to execute.

      • metric_name_field - The field in the Splunk results that will contain the metric name. Default "metric".

      • metric_value_field - The field in the Splunk results that will contain numerical data. Default "value".

      • match: - Regex used for selecting Splunk saved search queries. Default metrics.*.

      • app - The Splunk app in which the saved searches are located. Default "search".

      • request_timeout_seconds - Default 10

      • search_max_retry_count - Default 5

      • search_seconds_between_retries - Default 1

      • batch_size - Default 1000

      • initial_history_time_seconds - Default 0

      • max_restart_history_seconds - Default 86400

      • max_query_chunk_seconds - Default 3600

      • unique_key_fields - The fields to use to uniquely identify a record. Default _bkt and _cd.

      • parameters - Used in the Splunk API request. The default parameters provided make sure the Splunk saved search query refreshes. Default force_dispatch: true and dispatch.now: true.

  4. More advanced options can be found in the example configuration (github.com).

  5. Save the configuration file.

  6. Restart StackState Agent V2 to apply the configuration changes.

  7. Once the Agent has restarted, wait for the Agent to collect data and send it to StackState.

  8. Metrics retrieved from splunk are available in StackState as a metrics telemetry stream in the stackstate-metrics data source. This can be mapped to associated components.

Uniquely identify a record

To prevent sending duplicate metrics over multiple check runs, received saved search records must be uniquely identified for comparison. By default, a record is identified by the Splunk default fields _bkt and _cd. This behavior can be customized for each saved search by specifying unique_key_fields in the Splunk Metrics Agent check configuration. Note that the specified unique_key_fields fields are mandatory fields for each record returned by the Splunk saved search.

If it isn't possible to uniquely identify a record by a combination of specific fields, the whole record can be used by setting unique_key_fields: [] (an empty list).

Disable the Agent check

To disable the Splunk Metrics Agent check:

  1. Remove or rename the Agent integration configuration file, for example:

     mv /etc/stackstate-agent/conf.d/splunk_metric.d/conf.yaml /etc/stackstate-agent/conf.d/splunk_metric.d/conf.yaml.bak
  2. Restart the StackState Agent to apply the configuration changes.

Splunk metrics in StackState

Metrics retrieved from splunk are available in StackState as a metrics telemetry stream in the stackstate-metrics data source. This can be mapped to associated components.

See also

Last updated