GET _rollup/data/<index>

This API is useful because a rollup job is often configured to rollup only a subset of fields from the source index. Furthermore, only certain aggregations can be configured for various fields, leading to a limited subset of functionality depending on that configuration.

This API enables you to inspect an index and determine:

Imagine we have an index named sensor-1 full of raw data. We know that the data will grow over time, so there will be a sensor-2, sensor-3, etc. Let’s create a rollup job that targets the index pattern sensor-* to accommodate this future scaling:

PUT _rollup/job/sensor
{
  "index_pattern": "sensor-*",
  "rollup_index": "sensor_rollup",
  "cron": "*/30 * * * * ?",
  "page_size": 1000,
  "groups": {
    "date_histogram": {
      "field": "timestamp",
      "fixed_interval": "1h",
      "delay": "7d"
    },
    "terms": {
      "fields": [ "node" ]
    }
  },
  "metrics": [
    {
      "field": "temperature",
      "metrics": [ "min", "max", "sum" ]
    },
    {
      "field": "voltage",
      "metrics": [ "avg" ]
    }
  ]
}

We can then retrieve the rollup capabilities of that index pattern (sensor-*) via the following command:

GET _rollup/data/sensor-*

Which will yield the following response:

{
  "sensor-*" : {
    "rollup_jobs" : [
      {
        "job_id" : "sensor",
        "rollup_index" : "sensor_rollup",
        "index_pattern" : "sensor-*",
        "fields" : {
          "node" : [
            {
              "agg" : "terms"
            }
          ],
          "temperature" : [
            {
              "agg" : "min"
            },
            {
              "agg" : "max"
            },
            {
              "agg" : "sum"
            }
          ],
          "timestamp" : [
            {
              "agg" : "date_histogram",
              "time_zone" : "UTC",
              "fixed_interval" : "1h",
              "delay": "7d"
            }
          ],
          "voltage" : [
            {
              "agg" : "avg"
            }
          ]
        }
      }
    ]
  }
}

The response that is returned contains information that is similar to the original rollup configuration, but formatted differently. First, there are some house-keeping details: the rollup job ID, the index that holds the rolled data, and the index pattern that the job was targeting.

Next it shows a list of fields that contain data eligible for rollup searches. Here we see four fields: node, temperature, timestamp and voltage. Each of these fields list the aggregations that are possible. For example, you can use a min, max or sum aggregation on the temperature field, but only a date_histogram on timestamp.

Note that the rollup_jobs element is an array; there can be multiple, independent jobs configured for a single index or index pattern. Each of these jobs may have different configurations, so the API returns a list of all the various configurations available.

We could also retrieve the same information with a request to _all:

GET _rollup/data/_all

But note that if we use the concrete index name (sensor-1), we’ll retrieve no rollup capabilities:

GET _rollup/data/sensor-1

{

}

Why is this? The original rollup job was configured against a specific index pattern (sensor-*) not a concrete index (sensor-1). So while the index belongs to the pattern, the rollup job is only valid across the entirety of the pattern not just one of it’s containing indices. So for that reason, the get rollup capabilities API only returns information based on the originally configured index name or pattern.