Description of problem:
with rgw-multisite, dashboard deployment fails on secondary site because task tries to run task "get radosgw system user" on all mon's but through the container named ceph-mon-site2m1 on all 3 mons.

Version-Release number of selected component (if applicable):
container image rhceph:4-27 and ceph-ansible 4.0.23

How reproducible:
100%

Steps to Reproduce:
1. deploy a rgw-multiste setup without deploying dashboard initially
2. deploy dashboard on secondary site, collocated on osd node
3. fails with the result below.

Actual results:
2020-06-28 17:24:26,943 p=99004 u=cephadmin n=ansible | TASK [ceph-dashboard : get radosgw system user] *******************************
********************************************************
2020-06-28 17:24:26,943 p=99004 u=cephadmin n=ansible | Sunday 28 June 2020  17:24:26 +0300 (0:00:00.119)       0:10:26.424 ***********
2020-06-28 17:24:27,340 p=99004 u=cephadmin n=ansible | FAILED - RETRYING: get radosgw system user (3 retries left).
2020-06-28 17:24:27,411 p=99004 u=cephadmin n=ansible | FAILED - RETRYING: get radosgw system user (3 retries left).
2020-06-28 17:24:27,701 p=99004 u=cephadmin n=ansible | changed: [site2m1]
2020-06-28 17:24:32,595 p=99004 u=cephadmin n=ansible | FAILED - RETRYING: get radosgw system user (2 retries left).
2020-06-28 17:24:32,671 p=99004 u=cephadmin n=ansible | FAILED - RETRYING: get radosgw system user (2 retries left).
2020-06-28 17:24:37,850 p=99004 u=cephadmin n=ansible | FAILED - RETRYING: get radosgw system user (1 retries left).
2020-06-28 17:24:37,929 p=99004 u=cephadmin n=ansible | FAILED - RETRYING: get radosgw system user (1 retries left).
2020-06-28 17:24:43,116 p=99004 u=cephadmin n=ansible | fatal: [site2m2]: FAILED! => changed=true
  attempts: 3
  cmd:
  - timeout
  - --foreground
  - -s
  - KILL
  - '20'
  - podman
  - exec
  - ceph-mon-site2m1
  - radosgw-admin
  - --cluster
  - ceph
  - user
  - info
  - --uid=ceph-dashboard
  delta: '0:00:00.055607'
  end: '2020-06-28 17:24:44.044413'
  msg: non-zero return code
  rc: 125
  start: '2020-06-28 17:24:43.988806'
  stderr: 'Error: no container with name or ID ceph-mon-site2m1 found: no such container'
  stderr_lines: <omitted>
  stdout: ''
  stdout_lines: <omitted>
2020-06-28 17:24:43,192 p=99004 u=cephadmin n=ansible | fatal: [site2m3]: FAILED! => changed=true
  attempts: 3
  cmd:
  - timeout
  - --foreground
  - -s
  - KILL
  - '20'
  - podman
  - exec
  - ceph-mon-site2m1
  - radosgw-admin
  - --cluster
  - ceph
  - user
  - info
  - --uid=ceph-dashboard
  delta: '0:00:00.055752'
  end: '2020-06-28 17:24:44.121714'
  msg: non-zero return code
  rc: 125
  start: '2020-06-28 17:24:44.065962'
  stderr: 'Error: no container with name or ID ceph-mon-site2m1 found: no such container'
  stderr_lines: <omitted>
  stdout: ''
  stdout_lines: <omitted>

Expected results:
successfull playthrough

Additional info:
Affected task is found in roles/ceph-dashboard/tasks/configure_dashboard.yml
Initial workaround has been to add run_once: true to the task.