Description of problem:

When podman creates a container, it creates two additional systemd
scopes dynamically:

  - libpod-conmon-<CONTAINERID>.scope - runs a conmon process that
    tracks a container's pid1 into a dedicated pidfile.
  - libpod-<CONTAINERID>.scope - created dynamically by runc,
    for cgroups accounting.

The liveness of the scopes is directly tied to that of the podman
container.

On shutdown, systemd stops both services like pacemaker _and_ scopes,
in no particular order. Which means that it can happen that systemd
stops the container's scopes _before_ pacemaker itself. When that
it the case, the following :

  - the stopped scope sends a SIGTERM to the pid1 of the pacemaker
    container (i.e. pacemaker remote).
  - this makes the pacemaker container stops.
  - pacemaker notices that the pacemaker remote node stopped
    unexpectedly and pacemaker may try to restart it.

Ultimately the whole sequence of shutdown gets polluted and sometimes
yields an error that blocks the regular shutdown, or block the restart
once the node has rebooted.

There's no option in podman to configure the scope file to not stop
before pacemaker is stopped. The only workaround so far is to inject
an additional drop-in file for each scope, with extra dependencies
that prevents systemd from stopping the scopes file before pacemaker
is stopped (so effectively, giving back the entire control to
pacemaker).


Version-Release number of selected component (if applicable):
resource-agents-4.1.1-17.el8_0.3.x86_64

How reproducible:
Always

Steps to Reproduce:
1. create a bundle ocf resource in the cluster.
2. shutdown one cluster node
3. check in the journal whether the pacemaker-remote process received a SIGTERM before the pacemaker service was stopped by pacemaker

Actual results:
A pacemaker-managed podman container is stopped by systemd instead of pacemaker

Expected results:
pacemaker should be the only one to stop pacemaker-managed container.