Single pipeline - outdated deployment jobs

Is it possible to get an outdated job for just a single running pipeline?

According to Deployment safety | GitLab the outdated job can come only from the older pipeline running in parallel with another newer one. Is it so?

In our setup we have several jobs A,B,C using the same environment: staging. We encounter quite often a situation when just single pipeline is running but one of the A,B,C jobs gets cancelled because of:

“The deployment job is older than the previously succeeded deployment job, and therefore cannot be run.”

The jobs A,B,C can and are run in parallel. Moreover we use the resource groups with older_first policy to enqueue the jobs at the staging environment (using the staging resource group).

Are we missing something? How to avoid the “older” job cancellation within single running pipeline. Of course without disabling the “Skip outdated jobs” setting that we want to keep.

One more note: We used to use stages before and the setup worked as expected. But after removing stages and doing the job ordering strictly by the “needs” definitions we start to encounter this behavior.