[Fleet] Add Elastic agent integration

[nchaulet]

Add a new integration for Elastic Agent

Still in progress

What does this PR do?

Create a new Elastic Agent integration that add a dashboard for agent metrics

Checklist

• I have reviewed tips for building integrations and this pull request is aligned with them.
• I have verified that all datasets collect metrics or logs.

Author's Checklist

How to test this PR locally

This PR depends on this PR of the elastic Agent elastic/beats#22793

Related issues

Screenshots

[elasticmachine]

💚 Build Succeeded

the below badges are clickable and redirect to their specific view in the CI or DOCS

Expand to view the summary

Build stats

• Build Cause: Branch indexing

• Start Time: 2020-12-15T22:27:59.891+0000

• Duration: 27 min 39 sec

Test stats 🧪

Test	Results
Failed	0
Passed	145
Skipped	0
Total	145

[elasticmachine]

Pinging @elastic/ingest-management (Team:Ingest Management)

[ph]

@ravikesarwani Can you take a look?

@nchaulet I am curious, the values looks weirds to me, the graph looks too perfect for me? Are the values aggregated?

[nchaulet]

@ph The graph are stacked per process, what does seems weird to you in the graph?

[ph]

what is the left axis for the graph?

[nchaulet]

The left axis is a percentage for CPU, memory size for Memory and the number of open handle

Maybe this screenshot makes more sense

[ruflin]

@nchaulet The part that makes me suspicious is that memory usage of all process is identical. My guess is, that it takes memory usage from the parent process, so maybe the reporting is off not the graph. @blakerouse might know more.

I did run and install the package on my end and it works. I suggest we get it in an then iterate on it.

[ravikesarwani]

Some initial comments from my side:

• I found this issue that Dimitri created and may help here [Discuss] Dashboard Creation Guidelines #269
• Can we label the Y axis?
• Can we add a filter for easy selection by user to select one or more Agent sub processes.
• For CPU usage and open handles I was thinking if the stacked graph is the best default representation. Line chart maybe more appropriate here. Thoughts?
• For memory usage stacked maybe good if we can show whole part relationship. Total available memory (8 GB) and Agent, filebeat & metricbeat is taking x, y and z from the total and F is free.
• We can change the CPU Usage to CPU Usage %
  @nchaulet Feel free to ping me any time for us to chat on Zoom and brainstorm any of the above points.
  Is there some place I can play with this?

[nchaulet]

@ravi-elastic the dashboard as updated in our Zoom conversation

[nchaulet]

I will merge it as it is and will update in new PRs if we have things to change.

[ph]

@nchaulet I've been trying to find a followup issue, did you create one to install this package by default?

[nchaulet]

I just created the follow up issue to install the package by default elastic/kibana#86112