[Request]: Document how to work around broken fleet policies after upgrading

[rockdaboot]

Description

It's about Universal Profiling.

After upgrading a cluster, it may happen that Collector and Symbolizer endpoints are not configured. Data ingestion will not be possible at that point. Here is a screenshot how it should look like in deployment overview:

When the issue happens, the last two entries are not visible, even though Universal Profiling has been set up in Kibana and the "Add data" instructions are shown on the UP landing page.

A work-around in this situation is to fire these queries from the Kibana dev console:

DELETE kbn:/api/fleet/package_policies/b01ea2ca-297f-4b49-9d8e-27228d03a508?force=true
DELETE kbn:/api/fleet/package_policies/818cd448-aff8-462e-971e-314b4b620bf4?force=true

Then perform the Universal Profiling setup again.

Resources

There is an internal link. Main contact should know (fallback contact is @travisestill).

Which documentation set does this change impact?

Stateful and Serverless

Feature differences

Experienced in stateful. Don't know is it happens in serverless.

What release is this request related to?

N/A

Collaboration model

The documentation team

Point of contact.

Main contact: @nchaulet

Stakeholders:
@elastic/obs-ds-intake-services
@elastic/obs-ds-ebpf-go