mgr/rook: get running pods, auth rm, better error checking for orch nfs

This commit updates orch ls to show the age and the number of running nfs
pods, removes auth entities when removing an nfs service and implements
better error checking when creating nfs daemons.

Signed-off-by: Joseph Sawaya <jsawaya@redhat.com>

src/pybind/mgr/rook/module.py

@@ -1,5 +1,6 @@
-from logging import error
+import datetime
 import logging
+import re
 import threading
 import functools
 import os
@@ -348,6 +349,7 @@ def describe_service(self,
         if service_type == 'nfs' or service_type is None:
             # CephNFSes
             all_nfs = self.rook_cluster.get_resource("cephnfses")
+            nfs_pods = self.rook_cluster.describe_pods('nfs', None, None)
             for nfs in all_nfs:
                 if nfs['spec']['rados']['pool'] != NFS_POOL_NAME:
                     continue
@@ -356,13 +358,16 @@ def describe_service(self,
                 if svc in spec:
                     continue
                 active = nfs['spec'].get('server', {}).get('active')
+                creation_timestamp = datetime.datetime.strptime(nfs['metadata']['creationTimestamp'], '%Y-%m-%dT%H:%M:%SZ')
                 spec[svc] = orchestrator.ServiceDescription(
                     spec=NFSServiceSpec(
                         service_id=nfs_name,
                         placement=PlacementSpec(count=active),
                     ),
                     size=active,
                     last_refresh=now,
+                    running=len([1 for pod in nfs_pods if pod['labels']['ceph_nfs'] == nfs_name]),
+                    created=creation_timestamp.astimezone(tz=datetime.timezone.utc)
                 )
         if service_type == 'osd' or service_type is None:
             # OSDs
@@ -506,6 +511,15 @@ def remove_service(self, service_name: str, force: bool = False) -> str:
         elif service_type == 'rgw':
             return self.rook_cluster.rm_service('cephobjectstores', service_id)
         elif service_type == 'nfs':
+            ret, out, err = self.mon_command({
+                'prefix': 'auth ls'
+            })
+            matches = re.findall(rf'client\.nfs-ganesha\.{service_id}\..*', out)
+            for match in matches:
+                self.check_mon_command({
+                    'prefix': 'auth rm',
+                    'entity': match
+                })
             return self.rook_cluster.rm_service('cephnfses', service_id)
         elif service_type == 'rbd-mirror':
             return self.rook_cluster.rm_service('cephrbdmirrors', service_id)
@@ -558,7 +572,11 @@ def apply_rgw(self, spec):
     @handle_orch_error
     def apply_nfs(self, spec):
         # type: (NFSServiceSpec) -> str
-        return self.rook_cluster.apply_nfsgw(spec, self)
+        try:
+            return self.rook_cluster.apply_nfsgw(spec, self)
+        except Exception as e:
+            logging.error(e)
+            return "Unable to create NFS daemon, check logs for more traceback\n" + str(e.with_traceback(None))
 
     @handle_orch_error
     def remove_daemons(self, names: List[str]) -> List[str]:

src/pybind/mgr/rook/rook_cluster.py

@@ -776,7 +776,7 @@ def predicate(item):
                             "osd": ("ceph-osd-id", service_id),
                             "mon": ("mon", service_id),
                             "mgr": ("mgr", service_id),
-                            "ceph_nfs": ("ceph_nfs", service_id),
+                            "nfs": ("nfs", service_id),
                             "rgw": ("ceph_rgw", service_id),
                         }[service_type]
                     except KeyError: