[Inventory][infra] Asset details shows blank metrics section for containers detected by EEM but without integrations data

[crespocarlos]

Summary

The Asset details page is unable to show the metrics section for non-monitored containers. This causes the metrics section and the metrics tab to just show an empty section.

It could perhaps hide the metrics section and tab or display a callout with an onboarding flow.

How to reproduce

• Start the opentelementry-demo
• From the Inventory, navigate to the Asset Details from any container

AC

• It should show the onboarding callout as per https://github.com/elastic/observability-dev/issues/3966

[elasticmachine]

Pinging @elastic/obs-ux-infra_services-team (Team:obs-ux-infra_services)

[crespocarlos]

cc: @roshan-elastic

[roshan-elastic]

Hey @crespocarlos - thanks for this.

Could you remind me which containers this happens to?

i.e. in what scenario is a container found but no metrics are observed?

If these containers will show and be accessed when the entityCentricExperience flag is on, I think we can put this as something to tackle later when we want to start polishing things up as we work on entity views.

@smith - Curious what your thoughts here are?

[smith]

It's ok to show the empty session if we're showing a callout telling them they need to add metric data. Shouldn't this be the case for something without integrations data?

[crespocarlos]

I wonder if the page should show an onboarding flow in these cases or inform why the metrics section is empty. For host, the page renders empty charts, which is not the case here.

Could you remind me which containers this happens to?

i.e. in what scenario is a container found but no metrics are observed?

EEM, for sure, but it's possible that a container listed in the service overview doesn't have metrics, which could cause this. I can't say if this scenario occurs very often.

[roshan-elastic]

EEM, for sure, but it's possible that a container listed in the service overview doesn't have metrics, which could cause this. I can't say if this scenario occurs very often.

Cool. For containers which only show when EEM is enabled, I'm not too worried about the UX here right now. If we show the call to action as per this issue - that's good enough for now.

If this happens for containers without EEM then presumably they've been seeing blank metrics for a while? I don't think solving that is a priority right now.

I wonder if the page should show an onboarding flow in these cases or inform why the metrics section is empty. For host, the page renders empty charts, which is not the case here.

This issue should show the call out for adding metrics?

Shouldn't this be the case for something without integrations data?

Good point. @crespocarlos - shouldn't this issue show the callout if metrics are missing and solve our problem here?

[crespocarlos]

@roshan-elastic

Good point. @crespocarlos - shouldn't https://github.com/elastic/observability-dev/issues/3966 show the callout if metrics are missing and solve our problem here?

Ideally, yes, but that doesn't seem to be working for containers.

[roshan-elastic]

Ah - I see. It sounds like we need a bug to follow up on https://github.com/elastic/observability-dev/issues/3966 as it's not meeting the AC.

Would you mind creating one if you've been able to replicate?

I can prioritise accordingly.

[crespocarlos]

@roshan-elastic, we can use this ticket to work on the problem.

[roshan-elastic]

@crespocarlos - oh yeah, duh!

I'll prioritise - good that we've captured here. Thanks!

[iblancof]

Hi, I'll be taking care of this!

I have some comments on this @crespocarlos and @roshan-elastic.

About the Metrics tab not showing the callout, I think there might have been a slight misalignment from this comment.

The callout is intentionally hidden for anything that's not Docker or Kubernetes.
In fact, the Metrics tab is not even showing the metrics charts/empty states:

If I understand correctly, the goal of this issue is to display the callout regardless of the container type, is that right? Would it make sense considering we were not even showing the metrics charts/empty states?

I might be missing some context here, but it seems that if we’re not even showing the charts, it’s because these types of containers can’t have metrics at all. In that case, what would be the point of prompting users to add metrics if it’s not actually possible?

[crespocarlos]

Hey @iblancof

The problem is that without signals data, we don't know whether the entity is a Docker or Kubernetes container. We want to show the callout in this case, as we know that the entity is a container because it has a container.id field

The metrics tab will be hidden as part of #201752

[iblancof]

After discussing all possible cases privately with @crespocarlos and confirming that everything is covered, the final solution will be, as previously mentioned, to always display the callout whenever metrics are not detected, both on the overview and in the metrics tab.

Additionally, the onboarding link will be updated.
Some recent changes from the onboarding team seem to have affected how the current link works, so we’ll adjust it to ensure everything works as expected.