ILM: Including "aging" information in Explain lifecycle API

[ppf2]

6.6.0

I can see this feature being useful for admins in the field.

Imagine an admin is not seeing indices getting deleted from the cluster per their ILM policy and are wondering how much longer is left on the clock before ILM will consider the indices meeting the "age" requirement for deletion.

Currently, we have this information in the TRACE logging level

checking for index age to be at least [20m] before performing actions in the "delete" phase. Now: 1550263112, lifecycle date: 1550262751, age: [6m/360s]

It will be helpful to include this in the explain lifecycle API output as well (i.e., how much time is left before it will attempt to execute the next action without requiring admins to do their own time calculations) so that enabling trace logging is not necessary.

[elasticmachine]

Pinging @elastic/es-core-features

[gwbrown]

You can determine this from the current explain response, although I can certainly see how it's not immediately obvious, and should probably be documented better than it is. The lifecycle_date and lifecycle_date_millis fields contain the "reference" date used.

GET /*/_ilm/explain?human:

{
  "indices": {
    "test-000001": {
      "index": "test-000001",
      "managed": true,
      "policy": "my_policy",
      "lifecycle_date": "2019-02-15T22:25:03.953Z", <1>
      "lifecycle_date_millis": 1550269503953,
      "phase": "new",
      "phase_time": "2019-02-15T22:25:04.098Z",
      "phase_time_millis": 1550269504098,
      "action": "complete",
      "action_time": "2019-02-15T22:25:04.098Z",
      "action_time_millis": 1550269504098,
      "step": "complete",
      "step_time": "2019-02-15T22:25:04.098Z",
      "step_time_millis": 1550269504098
    }
  }
}

<1> is the date that's checked against min_age of each phase from the policy. So enabling trace logging isn't strictly necessary to get this information, although including the current delta in the explain API output and/or documenting this better would probably be good to do.

[ppf2]

Thanks for the info! Adding something like a "time until next phase" field to the explain output will be nice.