Remove unnecessary restarts of metricsets while using Pod/Service autodiscover

[ChrsMark]

Describe the enhancement:
Improve update checks for pod and service resources to avoid having unnecessary restarts of metricsets.
Similar to #19974.

[elasticmachine]

Pinging @elastic/integrations-platforms (Team:Platforms)

[ChrsMark]

I did some first investigation for this and it looks like Pods (and most probably services) are properly handled on Updates:

beats/libbeat/autodiscover/providers/kubernetes/pod.go

Line 140 in bcb4e0c

// If Pod is in a phase where all containers in the have terminated emit a stop event

I also added debug messages to observe the updates and I didn't catch any update that we don't handle properly and can cause unnecessary restarts. I think that we are ok with the current state.

@jsoriano do you have any ideas around this? Otherwise we can close it.

[jsoriano]

I did some first investigation for this and it looks like Pods (and most probably services) are properly handled on Updates:

beats/libbeat/autodiscover/providers/kubernetes/pod.go

Line 140 in bcb4e0c

// If Pod is in a phase where all containers in the have terminated emit a stop event

I also added debug messages to observe the updates and I didn't catch any update that we don't handle properly and can cause unnecessary restarts. I think that we are ok with the current state.

@jsoriano do you have any ideas around this? Otherwise we can close it.

Yep, I am ok with closing this, thanks for the investigation.