Add fallback for host.id in Cloud

[christos68k]

trafficstars

This PR changes the semantics behind host.id calculation for Cloud.

Changes

In Cloud, it is possible for instance_id to not be retrievable from inside the instance (e.g. security concerns / restricted instance metadata). In this case, the host.id can be set from machine-id (same steps as non-containerized systems).

Without a fallback, the must wording has strong implications for clients and users in cases where the instance_id can not be automatically retrieved:

1. Do not automatically populate host.id
2. Urge the user to either provide instance_id or change the configuration to make it retrievable

A (breaking) alternative to the current approach is to always populate host.id from machine-id, and use a different field for instance_id (e.g. service.instance.id)

Merge requirement checklist

• [x] CONTRIBUTING.md guidelines followed.
• [x] CHANGELOG.md updated for non-trivial changes.
• [ ] schema-next.yaml updated with changes to existing conventions.

[AlexanderWert]

@christos68k please do the change in the corresponding yaml file: https://github.com/open-telemetry/semantic-conventions/blob/ed7d3e0f939cbdcd6c0b6093d157d3f5e8bbc9bd/model/registry/host.yaml#L8

The tables in markdown files are generated from the model yaml files. So once you did the change in the yaml file you need to run: make table-generation. This will automatically update the tables in the markdown files.

[AlexanderWert]

@open-telemetry/semconv-system-approvers @open-telemetry/semconv-container-approvers PTAL

[github-actions[bot]]

This PR was marked stale due to lack of activity. It will be closed in 7 days.

[christos68k]

Adding a comment to keep this PR from being closed.

[github-actions[bot]]

This PR was marked stale due to lack of activity. It will be closed in 7 days.

[github-actions[bot]]

Closed as inactive. Feel free to reopen if this PR is still being worked on.

[christos68k]

Could someone with rights (@mx-psi @AlexanderWert maybe?) to do so reopen this pull request?

[ChrsMark]

@christos68k I have reopened this one. You will need to update the changelog entry to follow the new format: https://github.com/open-telemetry/semantic-conventions/blob/main/CONTRIBUTING.md#adding-a-changelog-entry

[github-actions[bot]]

This PR was marked stale due to lack of activity. It will be closed in 7 days.

[github-actions[bot]]

This PR was marked stale due to lack of activity. It will be closed in 7 days.

[jsuereth]

Given activity happening in Entity SIG - I think it's best to "wait" on major changes to resource related attributes. My own opinion on this Pr is that:

1. Yes, using machine id makes sense as the default for host.id
2. Having cloud-specific ids land in host.id seems problematic from a consistency/implementation standpoint. We need to define mechanisms where we would understand how a cloud-provider Resource detector would interact with a Host-specific resource detector. This change would force the ability for us to understand merge conflcts and overrides between these two disjoint components.

As such - (while I think machine id should be the default for host.id) I think we need to hold off on this PR until more progress has been made in Entity SiG - particularly the mapping between Resource <-> Entity and conflict resolution.

[github-actions[bot]]

This PR was marked stale due to lack of activity. It will be closed in 7 days.

[github-actions[bot]]

Closed as inactive. Feel free to reopen if this PR is still being worked on.