mgr/cephadm: allow osd spec removal

OSD specs/drivegroups are essentially templates for OSD creation but do
not map to the full lifecycle of the OSDs that they create.  When a spec
is removed, remove it immediately.

If no --force is provided, the error lists which OSDs will be left behind.
If --force is passed, the service is removed.

This leaves behind a few oddities:

- When you list services, OSDs that were created by the drivegroup may
  still exist, causing the drivegroup to appear in the list as
  unmanaged services.
- If you create a new drivegroup with the same name, the prior OSDs will
  appear to belong to the new spec instance, regardless of whether the
  spec/drivegroup parameters are the same.

Signed-off-by: Sage Weil <sage@newdream.net>
(cherry picked from commit bda5c304f33cb7a401a289d92f35b5d15f2fce15)

diff --git a/src/pybind/mgr/cephadm/module.py b/src/pybind/mgr/cephadm/module.py
index 665ee1a231b6ab99590a7f3b0b3233bfecd68bdf..235b16682c7847d46f4520b9d3f84fb31619f6f1 100644 (file)
--- a/src/pybind/mgr/cephadm/module.py
+++ b/src/pybind/mgr/cephadm/module.py
@@ -2062,9 +2062,9 @@ Then run the following:
                 return f'Unable to remove {service_name} service.\n' \
                        f'Note, you might want to mark the {service_name} service as "unmanaged"'
 
-        # Report list of affected OSDs
-        osds_msg = {}
-        if service_name.startswith('osd.'):
+        # Report list of affected OSDs?
+        if not force and service_name.startswith('osd.'):
+            osds_msg = {}
             for h, dm in self.cache.get_daemons_with_volatile_status():
                 osds_to_remove = []
                 for name, dd in dm.items():
@@ -2072,23 +2072,17 @@ Then run the following:
                         osds_to_remove.append(str(dd.daemon_id))
                 if osds_to_remove:
                     osds_msg[h] = osds_to_remove
-
-        found = self.spec_store.rm(service_name) or osds_msg
-        if found:
-            self._kick_serve_loop()
             if osds_msg:
-                return f'The service {service_name} will be deleted once the following OSDs: {osds_msg} ' \
-                       f'will be deleted manually.'
-            else:
-                return f'Removed service {service_name}'
-        else:
-            # must be idempotent: still a success.
-            try:
-                self.cache.get_daemon(service_name)
-                return (f'Failed to remove service <{service_name}>. "{service_name}" is the name of a daemon, not a service. '
-                        + 'Running service names can be found with "ceph orch ls"')
-            except OrchestratorError:
-                return f'Failed to remove service. <{service_name}> was not found. Running service names can be found with "ceph orch ls"'
+                msg = ''
+                for h, ls in osds_msg.items():
+                    msg += f'\thost {h}: {" ".join([f"osd.{id}" for id in ls])}'
+                raise OrchestratorError(f'If {service_name} is removed then the following OSDs will remain, --force to proceed anyway\n{msg}')
+
+        found = self.spec_store.rm(service_name)
+        if found and service_name.startswith('osd.'):
+            self.spec_store.finally_rm(service_name)
+        self._kick_serve_loop()
+        return f'Removed service {service_name}'
 
     @handle_orch_error
     def get_inventory(self, host_filter: Optional[orchestrator.InventoryFilter] = None, refresh: bool = False) -> List[orchestrator.InventoryHost]: