Dynamically show columns in a Discover table by wildcard matching

[melkamar (Martin Melka)]

Hi, I am wondering if Kibana supports my use case. I am creating a dashboard where I want to show app usage metrics data. The individual documents are pretty small (about 5-10 fields). Some of the fields are common to all the documents, but some are context-specific (and new may be added in the future).

To illustrate, this is an example document:

{
  "origin": {
    "inst": "foobar-demo",
    "env": "production"
  },
  "@timestamp": "2024-03-15T13:10:00+02:00",
  "metricPoint": {
    "metricSetName": "assignment_resolved",
    "userId": "5565d66d4809467a458b4567",
    "context": {
      "resolution_type": "done"
    }
  }
}

In my dashboard, I want to have a table where all relevant columns are shown. "Relevant columns" means:

• Common fields
  • origin.inst
  • origin.env
  • metricPoint.metricSetName
  • etc...
• "Context-specific" fields:
  • context.*

What I want is to show the statically defined columns for "common fields", but then to also show columns for all "context-specific" fields in the matched data. If the data in the selection would contain 10 different context.* fields, I would like there to be 10 extra columns.

The idea is that before looking at the columns in the dashboard, the data would be pre-filtered to a lower cardinality and at that point there would not be too many extra columns polluting the table.

---

At the moment, what I do is just show the statically defined fields, but when an analyst is interested in more details, they have to expand the document line with the <-> arrow (screenshot). This is clunky as it does not provide an easy overview of the data.

An alternative to this that I really want to avoid is having a concrete saved search per each metricSetName, which would show the appropriate context columns. But there would be dozens of saved searches with more being added on the fly.

---

To sum up, is there a way I can achieve this?

Without filter:

Filtering for just one metric type (notice the disappearing columns which are not contained in any matching document)

---

Thank you!