Fix/first unit restart (#110)

* bump lib

* wait for unit in cluster before setting tls

* fix for single units restart

* removed unused import

* Add log + error handling

lib/charms/mysql/v0/mysql.py

@@ -69,7 +69,7 @@ def wait_until_mysql_connection(self) -> None:
 import logging
 import re
 from abc import ABC, abstractmethod
-from typing import Iterable, List, Optional, Set, Tuple
+from typing import Iterable, List, Optional, Tuple
 
 from tenacity import retry, retry_if_exception_type, stop_after_attempt, wait_random
 
@@ -83,7 +83,7 @@ def wait_until_mysql_connection(self) -> None:
 
 # Increment this PATCH version before using `charmcraft publish-lib` or reset
 # to 0 if you are raising the major API version
-LIBPATCH = 8
+LIBPATCH = 9
 
 UNIT_TEARDOWN_LOCKNAME = "unit-teardown"
 
@@ -1075,17 +1075,11 @@ def get_member_state(self) -> Tuple[str, str]:
         # MEMBER_ROLE is empty if member is not in a group
         return results[0], results[1] if len(results) == 2 else "unknown"
 
-    def reboot_from_complete_outage(self, instance_names: Set[str]) -> None:
-        """Wrapper for reboot_cluster_from_complete_outage command.
-
-        Args:
-            instance_names: set of instance names (e.g. `juju-e3f183-4:3306`)
-        """
-        options = {"rejoinInstances": list(instance_names)}
-
+    def reboot_from_complete_outage(self) -> None:
+        """Wrapper for reboot_cluster_from_complete_outage command."""
         rejoin_command = (
             f"shell.connect('{self.cluster_admin_user}:{self.cluster_admin_password}@{self.instance_address}')",
-            f"dba.reboot_cluster_from_complete_outage('{self.cluster_name}', {json.dumps(options)} )",
+            f"dba.reboot_cluster_from_complete_outage('{self.cluster_name}')",
         )
 
         try:

src/charm.py

@@ -13,6 +13,7 @@
     MySQLConfigureMySQLUsersError,
     MySQLCreateClusterError,
     MySQLGetMySQLVersionError,
+    MySQLRebootFromCompleteOutageError,
 )
 from charms.rolling_ops.v0.rollingops import RollingOpsManager
 from ops.charm import (
@@ -513,6 +514,17 @@ def _restart(self, _) -> None:
         container = self.unit.get_container(CONTAINER_NAME)
         container.restart(MYSQLD_SERVICE)
 
+        # when restart done right after cluster creation (e.g bundles)
+        # or for single unit deployments, it's necessary reboot the
+        # cluster from outage to restore unit as primary
+        if self.app_peer_data["units-added-to-cluster"] == "1":
+            try:
+                self._mysql.reboot_from_complete_outage()
+            except MySQLRebootFromCompleteOutageError:
+                logger.error("Failed to restart single node cluster")
+                self.unit.status = BlockedStatus("Failed to restart primary")
+                return
+
         unit_label = self.unit.name.replace("/", "-")
 
         try:
@@ -523,10 +535,11 @@ def _restart(self, _) -> None:
                         # `self.active_status_message` once it gets merged
                         self.unit.status = ActiveStatus()
                         return
+                    logger.debug("Restarted instance not yet in cluster")
                     raise Exception
         except RetryError:
             logger.error("Unable to rejoin mysqld instance to the cluster.")
-            self.unit.status = BlockedStatus("Restarted node unable to rejoin the cluster")
+            self.unit.status = BlockedStatus("Restarted instance unable to rejoin the cluster")
 
 
 if __name__ == "__main__":

src/relations/mysql_tls.py

@@ -66,6 +66,11 @@ def _on_tls_relation_joined(self, _) -> None:
 
     def _on_certificate_available(self, event: CertificateAvailableEvent) -> None:
         """Enable TLS when TLS certificate available."""
+        if not self.charm.unit_initialized:
+            logger.debug("Wait unit initialise before request certificate.")
+            event.defer()
+            return
+
         if (
             event.certificate_signing_request.strip()
             != self.charm.get_secret(SCOPE, "csr").strip()