Add OOTB Kubernetes dashboard to Kubernetes integration & implement best dashboard creation practices

[MichaelKatsoulis]

Continuing work from elastic/beats#31591 which introduced new kubernetes OOTB dashboards for metricbeat 8.3.0, kubernetes package assets need to be updated as well.

Also dashboard creation best practises have been used as suggested in https://docs.google.com/document/d/1uyyFGx6xA5Kvl8c-ZdvXdvBGrHTylxU9F69TGqfzdmw/edit#heading=h.o6it0jm3zlmv:

1. Move from visualisations to lens whenever possible2.
2. Use Drilldown option.
3. Not too many visualizations per dashboard.
4. Dashboards changed to by value

These dashboards are compatible only with Kibana version 8.3.0 and above!

Related issues

• Closes Improve loading time of Kubernetes package Dashboards  beats#31021

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

• Start Time: 2022-06-14T09:15:32.312+0000

• Duration: 31 min 17 sec

Test stats 🧪

Test	Results
Failed	0
Passed	90
Skipped	0
Total	90

🤖 GitHub comments

To re-run your PR in the CI, just comment with:

• /test : Re-trigger the build.

[elasticmachine]

🌐 Coverage report

Name	Metrics % (covered/total)	Diff
Packages	100.0% (0/0)	💚
Files	100.0% (0/0)	💚 3.432
Classes	100.0% (0/0)	💚 3.432
Methods	94.872% (74/78)	👍 6.043
Lines	100.0% (0/0)	💚 10.104
Conditionals	100.0% (0/0)	💚

[flash1293]

First of all, thank you so much for this! It's great seeing all the new things being adopted.

It seems like there are still some convertable TSVB panels in the kubernetes integration (I might have missed some aspect though):

• API server
• Controller Manager
• Nodes. 👍
• Proxy
• Scheduler

Are those planned for later?

Some notes on converting:

• Simple bucket scripts (like doing a ratio) can be converted to formula
• The "derivative" agg is called "differences" in Lens formula (it has the same functionality)
• The "Unit" definition of derivative can be defined by using the "Normalize by unit" option from advanced options:

[flash1293]

As a quick side note - I like the markdown panels as an interim solution, however we are actively working on adding a nice looking metric visualization which allows to convey the same information (hopefully ready for 8.4)

[MichaelKatsoulis]

Are those planned for later?

No , probably I missed them. I will try to do everything in this PR

Some notes on converting:

• Simple bucket scripts (like doing a ratio) can be converted to formula
• The "derivative" agg is called "differences" in Lens formula (it has the same functionality)
• The "Unit" definition of derivative can be defined by using the "Normalize by unit" option from advanced options:

Thanks a lot. I will try it out !!

[MichaelKatsoulis]

As a quick side note - I like the markdown panels as an interim solution, however we are actively working on adding a nice looking metric visualization which allows to convey the same information (hopefully ready for 8.4)

Totally agree!

[MichaelKatsoulis]

I have update the Nodes dashboard to convert some visualisations to lens.

Regarding the rest (apiserver, proxy, scheduler, controllermanager) updating those dashboards was not part of this PR initially.

It seems that more work is needed than just converting visualisations to lens.
Some panels need to be reconsider as some metrics collected are deprecated in newer versions of kubernetes. See elastic/beats#31834.

They will be updated in a follow up PR.

Also a blocker while moving some visualisations to lens is that in lens there is no way currently, when grouping by a field to set the order_by to a specific filed value. Only alphabetical and rarity is possible which can hide important values.

[flash1293]

Thanks @MichaelKatsoulis ! It seems like there are a few visualizations on the "[Metrics Kubernetes] Pods" page which can be converted too, are they in-scope for this PR?

On "[Metrics Kubernetes] Overview" you are using TSVB markdown for the main menu, is there any special reason for that? Ideally a regular markdown panel should be used for that (accessible from the menu on the dashboard):

The TSVB markdown panel is still fetching data behind the scenes even if it's not used.

Also (and that's very important), these links won't work on a Kibana with a base path configured because they define an absolute path (starting with /). For maximum compatibility only specify the hash part (instead of /app/dashboards#/view/kubernetes-e0195ce0-bcaf-11ec-b64f-7dd6e8e82013, use #/view/kubernetes-e0195ce0-bcaf-11ec-b64f-7dd6e8e82013)

Should that menu be available on all dashboards? I like how you split it up but if they are not linked up among each other I can imagine navigating around will become difficult.

[MichaelKatsoulis]

It seems like there are a few visualizations on the "[Metrics Kubernetes] Pods" page which can be converted too, are they in-scope for this PR?

The reason for that is the problem with order by when grouping by pod.name. As soon as this is resolved I will move them as well.

On "[Metrics Kubernetes] Overview" you are using TSVB markdown for the main menu, is there any special reason for that? Ideally a regular markdown panel should be used for that (accessible from the menu on the dashboard):

Pure ignorance. I didn't even know that there was another option. I will move them.
Just to clarify @flash1293 you only refer to the Kubernetes Dashboards one. Because for some others I cannot convert it as I need to use fields and aggregations.

Also (and that's very important), these links won't work on a Kibana with a base path configured because they define an absolute path (starting with /). For maximum compatibility only specify the hash part (instead of /app/dashboards#/view/kubernetes-e0195ce0-bcaf-11ec-b64f-7dd6e8e82013, use #/view/kubernetes-e0195ce0-bcaf-11ec-b64f-7dd6e8e82013)

Thanks a lot. I will change that.

Should that menu be available on all dashboards? I like how you split it up but if they are not linked up among each other I can imagine navigating around will become difficult.

Good idea!

[flash1293]

@MichaelKatsoulis

you only refer to the Kubernetes Dashboards one

exactly

The reason for that is the #3115 (comment) with order by when grouping by pod.name. As soon as this is resolved I will move them as well.

The order by for pod.name grouping should work fine for average aggregations (it's only not working for differences, moving average, cumulative sum and formula). Seems like on the pod page the panels are just using average, right?

[MichaelKatsoulis]

The order by for pod.name grouping should work fine for average aggregations (it's only not working for differences, moving average, cumulative sum and formula). Seems like on the pod page the panels are just using average, right?

You are right again. I somehow thought it wasn't working for all of them. I will convert them.

Updated in a2de96a

[MichaelKatsoulis]

@flash1293 I see in Kibana that input controls which I use in all the dashboards will be soon deprecated. I guess they should be changed to the new controls. I missed that best practise in https://docs.google.com/document/d/1uyyFGx6xA5Kvl8c-ZdvXdvBGrHTylxU9F69TGqfzdmw/edit#heading=h.o6it0jm3zlmv.

Probably I can do those changes in this PR as well.

[flash1293]

@MichaelKatsoulis great point, I added that to the best practices doc, thanks for mentioning!

[MichaelKatsoulis]

I have update all the dashboards to use the new controls.

I will merge this PR and leave the rest of the updates as part of #3498