fix: don't remove from quorum if application is going down (#149)

requirements.txt

@@ -1,13 +1,13 @@
 cffi==1.16.0 ; python_version >= "3.10" and python_version < "4.0" and platform_python_implementation != "PyPy"
 cosl==0.0.11 ; python_version >= "3.10" and python_version < "4.0"
-cryptography==42.0.5 ; python_version >= "3.10" and python_version < "4.0"
+cryptography==42.0.7 ; python_version >= "3.10" and python_version < "4.0"
 kazoo==2.9.0 ; python_version >= "3.10" and python_version < "4.0"
 ops==2.13.0 ; python_version >= "3.10" and python_version < "4.0"
 pure-sasl==0.6.2 ; python_version >= "3.10" and python_version < "4.0"
 pycparser==2.22 ; python_version >= "3.10" and python_version < "4.0" and platform_python_implementation != "PyPy"
 pydantic==1.10.15 ; python_version >= "3.10" and python_version < "4.0"
 pyyaml==6.0.1 ; python_version >= "3.10" and python_version < "4.0"
 six==1.16.0 ; python_version >= "3.10" and python_version < "4.0"
-tenacity==8.2.3 ; python_version >= "3.10" and python_version < "4.0"
-typing-extensions==4.11.0 ; python_version >= "3.10" and python_version < "4.0"
+tenacity==8.3.0 ; python_version >= "3.10" and python_version < "4.0"
+typing-extensions==4.12.0 ; python_version >= "3.10" and python_version < "4.0"
 websocket-client==1.8.0 ; python_version >= "3.10" and python_version < "4.0"

src/charm.py

@@ -10,7 +10,7 @@
 from charms.grafana_agent.v0.cos_agent import COSAgentProvider
 from charms.rolling_ops.v0.rollingops import RollingOpsManager
 from charms.zookeeper.v0.client import QuorumLeaderNotFoundError
-from kazoo.exceptions import BadVersionError, ReconfigInProcessError
+from kazoo.exceptions import BadArgumentsError, BadVersionError, ReconfigInProcessError
 from ops import (
     ActiveStatus,
     CharmBase,
@@ -225,7 +225,7 @@ def _on_cluster_relation_departed(self, event: RelationDepartedEvent) -> None:
         # likely due to a controller upgrade or a cloud maintenance with machines being reshuffled
         # periodically, juju would emit a LeaderElected event, and would return no peer units
         # the leader would then remove all other units from the quorum, which when restarted, would fail
-        if not event.departing_unit:
+        if not event.departing_unit or not self.model.app.planned_units():
             return
 
         departing_server_id = (
@@ -238,6 +238,7 @@ def _on_cluster_relation_departed(self, event: RelationDepartedEvent) -> None:
             ReconfigInProcessError,  # another unit already handling
             BadVersionError,  # another unit handled
             QuorumLeaderNotFoundError,  # this unit is departing, can't find leader in peer data
+            BadArgumentsError,  # already handled
         ):
             pass
 

tests/integration/test_tls.py

@@ -26,6 +26,7 @@ async def test_deploy_ssl_quorum(ops_test: OpsTest):
             channel="edge",
             num_units=1,
             config={"ca-common-name": "zookeeper"},
+            revision=163,
         ),
     )
     await ops_test.model.add_relation(APP_NAME, TLS_NAME)
@@ -70,6 +71,7 @@ async def test_add_tls_provider_succeeds_after_removal(ops_test: OpsTest):
             channel="edge",
             num_units=1,
             config={"ca-common-name": "zookeeper"},
+            revision=163,
         ),
     )