mgr/cephadm: Only start ranked daemon with highest rank gen, if many daemon exists...

If the removal of an extra NFS daemon fails for any reason, that daemon should not be started, as the NFS daemon with the highest rank generation is already running.

Fixes: https://tracker.ceph.com/issues/73442
Signed-off-by: Shweta Bhosale <Shweta.Bhosale1@ibm.com>

diff --git a/src/pybind/mgr/cephadm/serve.py b/src/pybind/mgr/cephadm/serve.py
index c518baa4d51af0934b99e99d751d85b99693a074..e82a6daf1ca9af092ad2f7659f947fc1b8d98ecd 100644 (file)
--- a/src/pybind/mgr/cephadm/serve.py
+++ b/src/pybind/mgr/cephadm/serve.py
@@ -1246,8 +1246,28 @@ class CephadmServe:
             # If no action is specified, check whether we need to start the daemon
             if not action and dd.daemon_type in DISABLED_SERVICES:
                 if dd.status == 0 and not dd.user_stopped:
-                    self.log.debug(f'Starting daemon {dd.name()}')
-                    action = 'start'
+                    should_start = False
+                    if spec and service_registry.get_service(daemon_type_to_service(dd.daemon_type)).ranked(spec):
+                        # For ranked services, check if we should start this daemon
+                        same_rank_daemons = [
+                            d for d in self.mgr.cache.get_daemons_by_service(dd.service_name())
+                            if d.rank == dd.rank and d.daemon_type == dd.daemon_type
+                        ]
+                        self.log.debug(
+                            f'Start hightest rank_gen daemon of same rank daemons {same_rank_daemons}'
+                        )
+                        if len(same_rank_daemons) == 1:
+                            should_start = True
+                        else:
+                            # Multiple daemons exist for this rank, only start the highest generation
+                            highest_gen_daemon = max(same_rank_daemons,
+                                                     key=lambda d: d.rank_generation or 0)
+                            should_start = (dd == highest_gen_daemon)
+                    else:
+                        should_start = True
+                    if should_start:
+                        self.log.debug(f'Starting daemon {dd.name()}')
+                        action = 'start'
 
             if action:
                 if scheduled_action == 'redeploy' and action == 'reconfig':