def test_integrity_on_data_node_failure():
shakedown.wait_for(
lambda: write_data_to_hdfs("data-0-node.hdfs.mesos", TEST_FILE_1_NAME),
HDFS_CMD_TIMEOUT_SEC)
# gives chance for write to succeed and replication to occur
time.sleep(5)
kill_task_with_pattern("DataNode", 'data-0-node.hdfs.mesos')
kill_task_with_pattern("DataNode", 'data-1-node.hdfs.mesos')
time.sleep(1) # give DataNode a chance to die
shakedown.wait_for(
lambda: read_data_from_hdfs("data-2-node.hdfs.mesos", TEST_FILE_1_NAME
), HDFS_CMD_TIMEOUT_SEC)
check_health()
def test_integrity_on_name_node_failure():
"""
The first name node (name-0-node) is the active name node by default when HDFS gets installed.
This test checks that it is possible to write and read data after the first name node fails.
"""
kill_task_with_pattern("NameNode", 'name-0-node.hdfs.mesos')
time.sleep(1) # give NameNode a chance to die
shakedown.wait_for(
lambda: write_data_to_hdfs("data-0-node.hdfs.mesos", TEST_FILE_2_NAME),
HDFS_CMD_TIMEOUT_SEC)
shakedown.wait_for(
lambda: read_data_from_hdfs("data-2-node.hdfs.mesos", TEST_FILE_2_NAME
), HDFS_CMD_TIMEOUT_SEC)
check_health()
def test_kill_namenode_executor():
kill_task_with_pattern('hdfs.executor.Main', 'name-0-node.hdfs.mesos')
check_health()
def test_kill_all_namenodes():
for host in shakedown.get_service_ips(PACKAGE_NAME):
kill_task_with_pattern('NameNode', host)
check_health()
def test_kill_journalnode_executor():
kill_task_with_pattern('hdfs.executor.Main', 'journal-0-node.hdfs.mesos')
check_health()
def test_kill_scheduler():
kill_task_with_pattern('hdfs.scheduler.Main', get_scheduler_host())
check_health()
def test_kill_journal_node():
kill_task_with_pattern('JournalNode', 'journal-0-node.hdfs.mesos')
check_health()
def test_kill_name_node():
kill_task_with_pattern('NameNode', 'name-0-node.hdfs.mesos')
time.sleep(1) # give NameNode a chance to die
check_health()
def test_kill_data_node():
kill_task_with_pattern('DataNode', 'data-0-node.hdfs.mesos')
check_health()
