Tooltip aggs

[theNewFlesh]

Problem:
I have some data that I either cannot graph (non-numeric, not meant to be counted) or do not want to. However, I would like to know what the data is when it is paired with data that I am graphing. For instance, let's say I have an index of render jobs submitted by a hundred different users, each document has a user name, total render time and time of completion. I wish to view this data with the y-axis set to render time and the x-axis set to submission time. I could split chart or split bars by user name, but that gives me 100 splits which is hard to look at, let alone work with. What I really want is a way to hover over some interesting bar in the graph and know what users it is associated with.

Recommendation:
A means of configuring what fields are displayed in the tooltip. This will likely involve an aggregation function for condensing them. The most usefull to me would be one that reduces a set of values to a list of unique ones. For instance, a set of multiple and duplicate user names, to a smaller list of unique user names.

[rashidkpc]

This could be done as say, tooltip aggregations, where the results of the aggregation are displayed in the tooltip for that point. This will probably take some serious thought due to the way aggregations are merged for display

[theNewFlesh]

I would lean toward non-numeric aggregations types, mapping to the x and y axes, as well as the split chart and split bar methods are generally the best methods for displaying numeric (including count and unique count) data.

[theNewFlesh]

Perhaps a simple list of unique values and a user provided reverse lookup hash for converting results to nicely formatted values. For instance, results of John (engineer), Bill (engineer) and Greg (engineer) could be converted from ["John", "Bill", "Greg"] to "engineering team" via a hash like this: {"engineering team": ["John", "Bill", "Greg"]}.

Having it referenced within the .kibana index but not statically assigned per visualization would be huge for maintenance and up to date information.

[chadrik]

+1
This is our single biggest problem with Kibana right now. We want users to be able to draw conclusions from the graphs we provide, but currently we can't show them crucial secondary data associated with each primary (graphed) data point.
The primary data helps tell the user "what", but the secondary data provides clues to "why"

[vikrim1]

+1

[babadofar]

+1

[sivy]

👍

[marcinkubica]

+1

Actually in need of this trying to bleed my brain and hacking in. Need ability to pump some fields into tooltip but they can't go into the actual chart.

Probably not so crucial for visualizing logs however useful for statistical data.

[staticfloat]

Agreed; we'd like to pair performance metrics with, for example, the commit version of the software running when the performance metric was taken. This would make it trivial to see if a performance degradation was due to a change in version; but right now, it's not so easy to see non numeric data that is paired with a particular document, when plotting numeric performance values.

[hackhowtofaq]

+1

[messias72]

+1

[graemechristie]

+1