This is an automated email from the ASF dual-hosted git repository.
av pushed a commit to branch ignite-ducktape
in repository https://gitbox.apache.org/repos/asf/ignite.git
The following commit(s) were added to refs/heads/ignite-ducktape by this push:
new a44d5a3 IGNITE-13645 : Discovery ducktape test should detect failed
nodes by asking the cluster. (#8409)
a44d5a3 is described below
commit a44d5a3695663475563b661b305421efcd6f3422
Author: Vladsz83 <vlads...@gmail.com>
AuthorDate: Thu Nov 5 18:23:55 2020 +0300
IGNITE-13645 : Discovery ducktape test should detect failed nodes by asking
the cluster. (#8409)
---
.../services/utils/templates/log4j.xml.j2 | 4 +++
.../tests/ignitetest/tests/discovery_test.py | 31 +++++++++-------------
2 files changed, 16 insertions(+), 19 deletions(-)
diff --git
a/modules/ducktests/tests/ignitetest/services/utils/templates/log4j.xml.j2
b/modules/ducktests/tests/ignitetest/services/utils/templates/log4j.xml.j2
index b2a9a7b..a0eeff0 100644
--- a/modules/ducktests/tests/ignitetest/services/utils/templates/log4j.xml.j2
+++ b/modules/ducktests/tests/ignitetest/services/utils/templates/log4j.xml.j2
@@ -48,6 +48,10 @@
<level value="WARN"/>
</category>
+ <category name="org.apache.ignite.spi.discovery.tcp">
+ <level value="INFO"/>
+ </category>
+
<root>
<level value="INFO"/>
<appender-ref ref="CONSOLE_ERR"/>
diff --git a/modules/ducktests/tests/ignitetest/tests/discovery_test.py
b/modules/ducktests/tests/ignitetest/tests/discovery_test.py
index f2be35b..fbd392c 100644
--- a/modules/ducktests/tests/ignitetest/tests/discovery_test.py
+++ b/modules/ducktests/tests/ignitetest/tests/discovery_test.py
@@ -163,8 +163,7 @@ class DiscoveryTest(IgniteTest):
results['Ignite cluster start time (s)'] = start_servers_sec
- failed_nodes, survived_node = choose_node_to_kill(servers,
test_config.nodes_to_kill,
-
test_config.sequential_failure)
+ failed_nodes = choose_node_to_kill(servers, test_config.nodes_to_kill,
test_config.sequential_failure)
if test_config.load_type is not ClusterLoad.NONE:
load_config = ignite_config._replace(client_mode=True) if
test_config.with_zk else \
@@ -181,12 +180,11 @@ class DiscoveryTest(IgniteTest):
start_load_app(self.test_context, ignite_config=load_config,
params=params, modules=modules)
- results.update(self._simulate_nodes_failure(servers,
node_fail_task(ignite_config, test_config), failed_nodes,
- survived_node))
+ results.update(self._simulate_nodes_failure(servers,
node_fail_task(ignite_config, test_config), failed_nodes))
return results
- def _simulate_nodes_failure(self, servers, kill_node_task, failed_nodes,
survived_node):
+ def _simulate_nodes_failure(self, servers, kill_node_task, failed_nodes):
"""
Perform node failure scenario
"""
@@ -206,20 +204,19 @@ class DiscoveryTest(IgniteTest):
logged_timestamps = []
data = {}
- for failed_id in ids_to_wait:
- logged_timestamps.append(
- get_event_time(servers, survived_node,
node_failed_event_pattern(failed_id)))
+ for survivor in [n for n in servers.nodes if n not in failed_nodes]:
+ for failed_id in ids_to_wait:
+ logged_timestamps.append(get_event_time(servers, survivor,
node_failed_event_pattern(failed_id)))
+
+ self._check_failed_number(failed_nodes, survivor)
- self._check_failed_number(failed_nodes, survived_node)
self._check_not_segmented(failed_nodes)
logged_timestamps.sort(reverse=True)
- first_kill_time = epoch_mills(first_terminated)
- detection_delay = epoch_mills(logged_timestamps[0]) - first_kill_time
-
- data['Detection of node(s) failure (ms)'] = detection_delay
- data['All detection delays (ms):'] = str([epoch_mills(ts) -
first_kill_time for ts in logged_timestamps])
+ data['Detection of node(s) failure (ms)'] =
epoch_mills(logged_timestamps[0]) - epoch_mills(first_terminated)
+ data['All detection delays (ms):'] = str(
+ [epoch_mills(ts) - epoch_mills(first_terminated) for ts in
logged_timestamps])
data['Nodes failed'] = len(failed_nodes)
return data
@@ -348,13 +345,9 @@ def choose_node_to_kill(servers, nodes_to_kill,
sequential):
idx = random.randint(0, len(to_kill) - nodes_to_kill)
to_kill = to_kill[idx:idx + nodes_to_kill]
- survive = random.choice([node for node in servers.nodes if node not in
to_kill])
-
assert len(to_kill) == nodes_to_kill, "Unable to pick up required number
of nodes to kill."
- assert survive, "Unable to select survived node to monitor the cluster on
it."
-
- return to_kill, survive
+ return to_kill
def order(node):
