Different CI context for similarly configured repositories

Summary

In oniro-core, we have among other repositories, two that behave differently in the way they run the CI pipelines for merge requests.

https://gitlab.eclipse.org/eclipse/oniro-core/HLaaS/ runs the merge request pipelines in the context of the parent. This uses the CI workers of the parent. In this way, a fork of this repository will be able to successfully run CI pipelines when raising an MR against the parent because the parent has CI workers configured. See for example eclipse/oniro-core/HLaaS!5 (closed)

https://gitlab.eclipse.org/eclipse/oniro-core/oniro/ runs the merge request pipelines in the context of the fork. This means that a fork created from this repository cannot run the MR pipelines because, given that it works in the context of the fork, it has no workers available to pick up the jobs (forks don't inherit any workers and shared workers are not configured). See for example eclipse/oniro-core/oniro!10 (merged). What is more confusing is that something changed in the past 2 days for this repository because a 2-day old merge request ran in the parent context. See eclipse/oniro-core/oniro!3 (merged). Since then, all the current MRs for this repository run in the fork's context without access to the parent's workers making them timeout on workers' availability.

As far as the settings we have access to, these two repositories have very similar (if not identical) configurations. Also, the repositories are in the same GitLab group.

Steps to reproduce

• Create a user fork from: https://gitlab.eclipse.org/eclipse/oniro-core/oniro/
• Propose a trivial change via a forked branch raised as an MR towards the parent repository

What is the current bug behaviour?

The CI pipelines fail/get stuck/timeout.

What is the expected correct behaviour?

The CI pipelines should run.

Relevant logs and/or screenshots

GitLab pipeline log: "There has been a timeout failure or the job got stuck. Check your timeout limits or try again."