While you can configure the rollup policies in the Splunk UI, you can also do it using Splunk’s REST API or using spunk configuration files (metric_rollups.conf).

Here is the interesting part: your rollup span value is limited only to the values available in the UI drop-down (no matter which method of configuration you are using).

The options that you can use are:

If you will try to use any other value you will see the following error in the splunkd.log: 12-12-2022 14:04:08.645 +1100 ERROR MetricsRollupPolicy [12181677 TcpChannelThread] - Failed to apply rollup policy to index='...'. Summary span='7d' cannot be cron scheduled.

This is not (as of 15/12/2022) a documented “feature”.

Other posts about Splunk

The post Splunk Failed to apply rollup policy to index… Summary span… cannot be cron scheduled appeared first on ISbyR.

The Problem

Let’s rewind a little.

Below is the payload that is sent to Splunk HEC and you will notice that there are 2 “statuses”:

• "status": "success" – which is one of the dimensions and it can represent a collector/monitor status
• "metric_name:status": 0 – which is the actual metric value that was collected by the collector/monitor

{
    "time": 1664970920,
    "event": "metric",
    "host": "host_5.splunk.com",
    "index": "d_telegraf_metrics",
    "fields": {
        "collector": "collector_a",
        "status": "success",
        "metric_name:query_time_seconds": 10.869,
        "metric_name:status": 0
    }
}

In the perfect world where you would probably rename one of these not to confuse the end-user in Splunk, but that (living in a perfect world) is not always the case.

As a result, we end up with NULLs in the graphs

The Solution

Lucky for us Splunk’s search language (SPL) is very powerful and flexible and with two little modifications to the “original” SPL (that was produced by the Metrics Analyzer), we can solve the issue.

All you need to do is :

1. instead of prestats=true rename the metric function result using as command.
2. update the avg function in the timechart command to use the renamed field name.

Original SPL:

Revised SPL:

The Result

More posts about Splunk

The post Splunk Failed to apply rollup policy to index… Summary span… cannot be cron scheduled appeared first on ISbyR.

While you can configure the rollup policies in the Splunk UI, you can also do it using Splunk’s REST API or using spunk configuration files (metric_rollups.conf).

Here is the interesting part: your rollup span value is limited only to the values available in the UI drop-down (no matter which method of configuration you are using).

The options that you can use are:

If you will try to use any other value you will see the following error in the splunkd.log: 12-12-2022 14:04:08.645 +1100 ERROR MetricsRollupPolicy [12181677 TcpChannelThread] - Failed to apply rollup policy to index='...'. Summary span='7d' cannot be cron scheduled.

This is not (as of 15/12/2022) a documented “feature”.

Other posts about Splunk

The post Splunk Failed to apply rollup policy to index… Summary span… cannot be cron scheduled appeared first on ISbyR.

The Problem

Let’s rewind a little.

Below is the payload that is sent to Splunk HEC and you will notice that there are 2 “statuses”:

• "status": "success" – which is one of the dimensions and it can represent a collector/monitor status
• "metric_name:status": 0 – which is the actual metric value that was collected by the collector/monitor

{
    "time": 1664970920,
    "event": "metric",
    "host": "host_5.splunk.com",
    "index": "d_telegraf_metrics",
    "fields": {
        "collector": "collector_a",
        "status": "success",
        "metric_name:query_time_seconds": 10.869,
        "metric_name:status": 0
    }
}

In the perfect world where you would probably rename one of these not to confuse the end-user in Splunk, but that (living in a perfect world) is not always the case.

As a result, we end up with NULLs in the graphs

The Solution

Lucky for us Splunk’s search language (SPL) is very powerful and flexible and with two little modifications to the “original” SPL (that was produced by the Metrics Analyzer), we can solve the issue.

All you need to do is :

1. instead of prestats=true rename the metric function result using as command.
2. update the avg function in the timechart command to use the renamed field name.

Original SPL:

Revised SPL:

The Result

More posts about Splunk

The post Plotting Splunk with the same metric and dimension names shows NULL appeared first on ISbyR.