mgr/cephadm: fix error handling in deploy_given_daemons

In the case where _create_daemon just returns an exception
itself, which should only happen in the case deployment fails
in general rather than for a specific daemon, (can happen if
the host it wants to deploy on has gone offline for example)
deploy_given_daemons was not properly handling the exception
and instead referenced the unbound variable "failures".

Signed-off-by: Adam King <[email protected]>
(cherry picked from commit f0207ca)

Resolves: rhbz#2371418

Author: adk3798

src/pybind/mgr/cephadm/serve.py

@@ -1160,6 +1160,8 @@ async def deploy_given_daemons(self, to_deploy: List[CephadmDaemonDeploySpec]) -
             # deployment of an individual daemon
             msg = (f"Failed while placing {daemon_names} "
                    f"on {hostname}: {e}")
+            self.mgr.log.error(msg)
+            return (False, set(), [msg])
         for daemon_name, msg in failures.items():
             # these are failures specific to a particular daemon
             daemon_spec = [d for d in to_deploy if d.name() == daemon_name][0]