fix(docker): multiple fixes to docker_backend_local.yaml

configurations/nemesis/additional_configs/docker_backend_local.yaml

@@ -1,8 +1,8 @@
 test_duration: 90
 
 prepare_write_cmd:
-    - "cassandra-stress write cl=QUORUM n=1048576 -schema 'replication(strategy=NetworkTopologyStrategy,replication_factor=3)' -mode cql3 native -rate threads=80 -pop seq=1..5048570 -col 'n=FIXED(8) size=FIXED(128)' -log interval=5"
-    - "scylla-bench -workload=sequential -mode=write -replication-factor=3 -partition-count=10000 -clustering-row-count=100 -clustering-row-size=uniform:128..2048 -concurrency=10 -connection-count=10 -consistency-level=quorum -rows-per-request=10 -timeout=30s -validate-data"
+    - "cassandra-stress write cl=QUORUM n=1048576 -schema 'replication(strategy=NetworkTopologyStrategy,replication_factor=3)' -mode cql3 native -rate threads=30 throttle=2000/s -pop seq=1..5048570 -col 'n=FIXED(8) size=FIXED(128)' -log interval=5"
+    - "scylla-bench -workload=sequential -mode=write -replication-factor=3 -partition-count=10000 -clustering-row-count=100 -clustering-row-size=uniform:128..2048 -concurrency=10 -connection-count=10 -consistency-level=quorum -rows-per-request=10 -timeout=30s -validate-data -max-rate=1000"
 
 stress_cmd:
     - "cassandra-stress write cl=QUORUM duration=60m -schema 'replication(strategy=NetworkTopologyStrategy,replication_factor=3)' -mode cql3 native -rate threads=30 throttle=2000/s -pop seq=1..5048570 -col 'n=FIXED(8) size=FIXED(128)' -log interval=5"
@@ -21,11 +21,17 @@ nemesis_filter_seeds: false
 nemesis_during_prepare: true
 
 # NOTE: the parameters reduce footprint of scylla cluster for docker backend, comparing to default values
-append_scylla_args: '--smp 1 --memory 5G'
+append_scylla_args: '--smp 1 --memory 2G'
 use_mgmt: false
 monitor_swap_size: 0
 
 # NOTE: encryption should be disabled until
 #       https://github.com/scylladb/scylla-cluster-tests/issues/7287 is fixed
 server_encrypt: false
 client_encrypt: false
+
+# TODO: remove this when we'll run this in jenkins
+enable_argus: false
+
+# cause of issue https://github.com/scylladb/scylla-monitoring/issues/2246, we should use the older monitoring version
+monitor_branch: 'branch-4.5'

sdcm/cluster.py

@@ -12,6 +12,8 @@
 # Copyright (c) 2016 ScyllaDB
 
 # pylint: disable=too-many-lines
+from __future__ import annotations
+
 import contextlib
 import queue
 import logging
@@ -2602,7 +2604,7 @@ def check_node_health(self, retries: int = CHECK_NODE_HEALTH_RETRIES) -> None:
                          CHECK_NODE_HEALTH_RETRY_DELAY, self.name)
             time.sleep(CHECK_NODE_HEALTH_RETRY_DELAY)
 
-    def get_nodes_status(self):
+    def get_nodes_status(self) -> dict[BaseNode, dict]:
         nodes_status = {}
         try:
             statuses = self.parent_cluster.get_nodetool_status(verification_node=self)
@@ -2660,7 +2662,7 @@ def get_peers_info(self):
         return peers_details
 
     @retrying(n=5, sleep_time=10, raise_on_exceeded=False)
-    def get_gossip_info(self):
+    def get_gossip_info(self) -> dict[BaseNode, dict]:
         gossip_info = self.run_nodetool('gossipinfo', verbose=False, warning_event_on_exception=(Exception,),
                                         publish_event=False)
         LOGGER.debug("get_gossip_info: %s", gossip_info)
@@ -3180,7 +3182,7 @@ def tags(self) -> Dict[str, str]:
     def dead_nodes_ip_address_list(self):
         return [node.ip_address for node in self.dead_nodes_list]
 
-    def get_ip_to_node_map(self):
+    def get_ip_to_node_map(self) -> dict[str, BaseNode]:
         """returns {ip: node} map for all nodes in cluster to get node by ip"""
         return {ip: node for node in self.nodes for ip in node.get_all_ip_addresses()}
 
@@ -3219,8 +3221,12 @@ def get_rack_names_per_datacenter_and_rack_idx(self, db_nodes: list[BaseNode] |
         db_nodes = db_nodes if db_nodes else self.nodes
         status = db_nodes[0].get_nodes_status()
 
+        # intersection of asked nodes and nodes returned by nodetool status
+        # since topology might change during this command execution
+        actual_db_nodes = set(status.keys()).intersection(db_nodes)
+
         rack_names_mapping = {}
-        for (region, rack), nodes in self.nodes_by_racks_idx_and_regions(nodes=db_nodes).items():
+        for (region, rack), nodes in self.nodes_by_racks_idx_and_regions(nodes=actual_db_nodes).items():
             rack_names_mapping[(region, rack)] = status[nodes[0]]['rack']
 
         return rack_names_mapping

sdcm/utils/remote_logger.py

@@ -275,13 +275,6 @@ def __init__(self, node, target_log_file: str):
         super().__init__(target_log_file)
 
 
-class DockerScyllaLogger(CommandNodeLoggerBase):
-    # pylint: disable=invalid-overridden-method
-    @cached_property
-    def _logger_cmd(self) -> str:
-        return f'docker logs -f {self._node.name} 2>&1 | grep scylla >>{self._target_log_file}'
-
-
 class DockerGeneralLogger(CommandNodeLoggerBase):
     # pylint: disable=invalid-overridden-method
     @cached_property
@@ -577,8 +570,6 @@ def _logger_cmd(self) -> str:
 
 def get_system_logging_thread(logs_transport, node, target_log_file):  # pylint: disable=too-many-return-statements
     if logs_transport == 'docker':
-        if 'db-node' in node.name:
-            return DockerScyllaLogger(node, target_log_file)
         return DockerGeneralLogger(node, target_log_file)
     if logs_transport == 'kubectl':
         return KubectlGeneralLogger(node, target_log_file)

sdcm/utils/remotewebbrowser.py

@@ -64,7 +64,8 @@ def web_driver_docker_client(self) -> Optional[DockerClient]:
 class RemoteBrowser:
     def __init__(self, node, use_tunnel=True):
         self.node = node
-        self.use_tunnel = bool(self.node.ssh_login_info and use_tunnel)
+        backend = self.node.parent_cluster.params.get("cluster_backend")
+        self.use_tunnel = bool(self.node.ssh_login_info and use_tunnel and backend not in ('docker',))
 
     @cached_property
     def browser(self):