pr comments

lib/charms/mongodb/v0/mongos.py

@@ -180,12 +180,12 @@ def remove_shard(self, shard_name: str) -> None:
             cannot_remove_shard = (
                 f"cannot remove shard {shard_name} from cluster, another shard is draining"
             )
-            logger.info(cannot_remove_shard)
+            logger.error(cannot_remove_shard)
             raise NotReadyError(cannot_remove_shard)
 
         # TODO Follow up PR, there is no MongoDB command to retrieve primary shard, this is
         # possible with mongosh.
-        primary_shard = False
+        primary_shard = self.get_primary_shard()
         if primary_shard:
             # TODO Future PR, support removing Primary Shard if there are no unsharded collections
             # on it. All sharded collections should perform `MovePrimary`
@@ -196,8 +196,24 @@ def remove_shard(self, shard_name: str) -> None:
             raise RemovePrimaryShardError(cannot_remove_primary_shard)
 
         logger.info("Attempting to remove shard %s", shard_name)
-        self.client.admin.command("removeShard", shard_name)
-        logger.info("Shard %s, now draining", shard_name)
+        removal_info = self.client.admin.command("removeShard", shard_name)
+
+        # process removal status
+        remaining_chunks = (
+            removal_info["remaining"]["chunks"] if "remaining" in removal_info else "None"
+        )
+        dbs_to_move = (
+            removal_info["dbsToMove"]
+            if "dbsToMove" in removal_info and removal_info["dbsToMove"] != []
+            else ["None"]
+        )
+        logger.info(
+            "Shard %s is draining status is: %s. Remaining chunks: %s. DBs to move: %s.",
+            shard_name,
+            removal_info["state"],
+            str(remaining_chunks),
+            ",".join(dbs_to_move),
+        )
 
     def _is_shard_draining(self, shard_name: str) -> bool:
         """Reports if a given shard is currently in the draining state.
@@ -219,6 +235,12 @@ def _is_shard_draining(self, shard_name: str) -> bool:
             f"Shard {shard_name} not in cluster, could not retrieve draining status"
         )
 
+    def get_primary_shard(self) -> str:
+        """Processes sc_status and identifies the primary shard."""
+        # TODO Follow up PR, implement this function there is no MongoDB command to retrieve
+        # primary shard, this is possible with mongosh.
+        return False
+
     @staticmethod
     def _is_any_draining(sc_status: Dict, ignore_shard: str = "") -> bool:
         """Returns true if any shard members is draining.

lib/charms/mongodb/v0/shards_interface.py

@@ -46,6 +46,7 @@
 KEYFILE_KEY = "key-file"
 HOSTS_KEY = "host"
 OPERATOR_PASSWORD_KEY = MongoDBUser.get_password_key_name_for_user(OperatorUser.get_username())
+FORBIDDEN_REMOVAL_ERR_CODE = 20
 
 
 class RemoveLastShardError(Exception):
@@ -83,6 +84,7 @@ def __init__(
     def _on_relation_joined(self, event):
         """Handles providing shards with secrets and adding shards to the config server."""
         if not self.pass_hook_checks(event):
+            logger.info("Skipping relation joined event: hook checks did not pass")
             return
 
         # TODO Future PR, sync tls secrets and PBM password
@@ -132,6 +134,13 @@ def _on_relation_event(self, event):
         Updating of shards is done automatically via MongoDB change-streams.
         """
         if not self.pass_hook_checks(event):
+            logger.info("Skipping relation event: hook checks did not pass")
+            return
+
+        # adding/removing shards while a backup/restore is in progress can be disastrous
+        pbm_status = self.charm.backups._get_pbm_status()
+        if isinstance(pbm_status, MaintenanceStatus):
+            event.defer("Cannot add/remove shards while a backup/restore is in progress.")
             return
 
         departed_relation_id = None
@@ -150,7 +159,7 @@ def _on_relation_event(self, event):
             event.defer()
             return
         except OperationFailure as e:
-            if e.code == 20:
+            if e.code == FORBIDDEN_REMOVAL_ERR_CODE:
                 # TODO Future PR, allow removal of last shards that have no data. This will be
                 # tricky since we are not allowed to update the mongos config in this way.
                 logger.error(
@@ -208,7 +217,7 @@ def remove_shards(self, departed_shard_id):
                 self.charm.unit.status = MaintenanceStatus(f"Draining shard {shard}")
                 logger.info("Attempting to removing shard: %s", shard)
                 mongo.remove_shard(shard)
-                logger.info("Attempting to removing shard: %s", shard)
+                logger.info("Shard: %s, is now draining", shard)
 
                 if shard in mongo.get_shard_members():
                     shard_draining_message = f"shard {shard} still exists in cluster after removal, it is still draining."