fix(ignore_topology_change_coordinator_errors): support sct_config parameter

PR scylladb/scylla-cluster-tests/scylladb#10386 filter some expected
raft error messages globally. But this change broke integration
unit test. Issue scylladb#10676.

This fix 2 things:
 - change function signature to run with passed sct config for
unit tests or get sct running config
 - correctly filter error events.

ignore_topology_change_coordinator_errors is context manager
created as generator wrapped contextlib.contextmanager
and DBEventsFilter are created using ExitStack as ContextManagers.
To call this contextmanagers, was used
`ignore_topology_change_coordinator_errors().__enter__()` call

but once this call finished, all DBEventsFilters contextmangers
executed theirs __exit__() method and thus all event filters marked
as expired and appriate events are not filterd out. To avoid that,
ExitStack object recreated and all __exit__ methods will be triggered
after `yield`. This allow to run `ignore_topology_change_coordinator_errors`
as context manager to wrap some code/functions/method
or execute it globally withou events expiring

Fixes scylladb#10676

Author: aleksbykov

sdcm/sct_events/group_common_events.py

@@ -72,14 +72,15 @@ def ignore_operation_errors():
 
 
 @contextmanager
-def ignore_topology_change_coordinator_errors():
+def ignore_topology_change_coordinator_errors(params: dict | None = None):
+    params = params or TestConfig().tester_obj().params
     with ExitStack() as stack:
         if SkipPerIssues(
                 issues=[
                     "https://github.com/scylladb/scylladb/issues/20754",
                     "https://github.com/scylladb/scylladb/issues/20950",
                 ],
-                params=TestConfig().tester_obj().params,
+                params=params,
         ):
             # @piodul:
             #
@@ -98,17 +99,19 @@ def ignore_topology_change_coordinator_errors():
                 line=r".*raft_topology - topology change coordinator fiber got error exceptions::unavailable_exception "
                      r"\(Cannot achieve consistency level for cl ALL\.",
             ))
-            stack.enter_context(DbEventsFilter(
-                db_event=DatabaseLogEvent.RUNTIME_ERROR,
-                line=r".*raft_topology - topology change coordinator fiber got error std::runtime_error"
-                     r" \(raft topology: exec_global_command\(barrier\) failed with seastar::rpc::closed_erro"
-                     r"r \(connection is closed\)\)"
-            ))
-            stack.enter_context(DbEventsFilter(
-                db_event=DatabaseLogEvent.RUNTIME_ERROR,
-                line=r".*raft_topology - drain rpc failed, proceed to fence old writes:.*connection is closed",
-            ))
+        stack.enter_context(DbEventsFilter(
+            db_event=DatabaseLogEvent.RUNTIME_ERROR,
+            line=r".*raft_topology - topology change coordinator fiber got error std::runtime_error"
+                 r" \(raft topology: exec_global_command\(barrier\) failed with seastar::rpc::closed_error"
+                 r" \(connection is closed\)\)"
+        ))
+        stack.enter_context(DbEventsFilter(
+            db_event=DatabaseLogEvent.RUNTIME_ERROR,
+            line=r".*raft_topology - drain rpc failed, proceed to fence old writes:.*connection is closed",
+        ))
+        new_stack = stack.pop_all()
         yield
+        new_stack.close()
 
 
 @contextmanager

sdcm/sct_events/setup.py

@@ -154,7 +154,7 @@ def enable_default_filters(sct_config: SCTConfiguration):  # pylint: disable=unu
     # upgrades and any place where the race between raft global barrier and gossipier could
     # take place. So ignore such messages globally for any sct test.
     # TODO: this should be removed after gossiper will be removed.
-    ignore_topology_change_coordinator_errors().__enter__()
+    ignore_topology_change_coordinator_errors(sct_config).__enter__()
 
 
 __all__ = ("start_events_device", "stop_events_device", "enable_default_filters")

unit_tests/test_events.py

@@ -412,7 +412,7 @@ def test_default_filters(self):
         with environment(SCT_CLUSTER_BACKEND='docker'):
             enable_default_filters(SCTConfiguration())
 
-        with self.wait_for_n_events(self.get_events_logger(), count=4):
+        with self.wait_for_n_events(self.get_events_logger(), count=5):
             DatabaseLogEvent.BACKTRACE() \
                 .add_info(node="A",
                           line_number=22,
@@ -439,6 +439,13 @@ def test_default_filters(self):
                      ".c.sct-project-1.internal/10.142.1.155:9042: Cannot achieve consistency level for cl ONE. Requires 1, alive 0",
             ).publish()
 
+            DatabaseLogEvent.RUNTIME_ERROR().add_info(
+                node="A",
+                line_number=22,
+                line="ERROR 2023-12-18 12:45:25,673 [shard 0: gms] raft_topology - topology change coordinator fiber got error std::runtime_error "
+                     "(raft topology: exec_global_command(barrier) failed with seastar::rpc::closed_error (connection is closed))"
+            ).publish()
+
         log_content = self.get_event_log_file("events.log")
 
         self.assertIn("other back trace", log_content)
@@ -451,6 +458,7 @@ def test_default_filters(self):
         error_log_content = self.get_event_log_file("error.log")
         assert 'data_dictionary::no_such_column_family' not in error_log_content
         assert 'Authentication error' in error_log_content
+        assert 'topology change coordinator fiber got error' not in error_log_content
 
     def test_failed_stall_during_filter(self):
         with self.wait_for_n_events(self.get_events_logger(), count=5, timeout=3):