Description of problem:

When you try to sync more AC repos concurrently to a Capsule than available pulpcore workers, the sync fails leaving OOMs in logs.

How reproducible:

Always for the given setup.

Is this issue a regression from an earlier version:

Hard to say, previous version suffered from deadlock so the task failed for a different reason: https://issues.redhat.com/browse/SAT-27939

Steps to Reproduce:

1. Start with fresh Satellite and Capsule. I used standard flavour SatLab VMs - Capsule with 24 GB RAM, 4 GB SWAP, 6 CPUs -> 6 pulpcore-workers running.
2. Set the Capsule with Immediate download policy and disable the autosync on CV promote.
3. Create a product with some mid-size Ansible Collection repo and sync it. I used this one  from galaxy.ansible.com (470 collections):
    - name: prometheus.prometheus
      version: "0.25.0"
4. Create 2 Lifecycle Environments and assign them to Capsule.
5. Create 3 Content Views, add the repo, publish and promote them to LCEs from 4. Now you have 6 repos, one for each pulpcore-worker, which is the minimal setup to hit the issue.
6. Trigger the Complete sync on the Capsule.

Actual behavior:
The sync task fails with "Errors: Pulp task error"

Expected behavior:
One or more of the followings:

• ability to successfully sync more AC repos than workers count
• lower RAM usage per worker
• minimal requirements RAM / CPU redefined
• better hint what happened in the Sat task result