def test_partition_master_outgoing():
master_leader_ip = shakedown.master_leader_ip()
shakedown.partition_master(master_leader_ip, incoming=False, outgoing=True)
time.sleep(20)
shakedown.reconnect_master(master_leader_ip)
check_health()
def test_partition_master_both_ways():
master_leader_ip = shakedown.master_leader_ip()
shakedown.partition_master(master_leader_ip)
time.sleep(20)
shakedown.reconnect_master(master_leader_ip)
check_health()
def test_partition(install_framework):
host = get_node_host()
shakedown.partition_agent(host)
shakedown.reconnect_agent(host)
check_health()
def test_kill_all_task_in_node():
hosts = shakedown.get_service_ips(PACKAGE_NAME)
for host in hosts:
kill_task_with_pattern('CassandraDaemon', host)
recover_failed_agents(hosts)
check_health()
def test_config_update_then_master_killed():
master_leader_ip = shakedown.master_leader_ip()
run_planned_operation(
lambda: bump_cpu_count_config(-0.1),
lambda: kill_task_with_pattern('mesos-master', master_leader_ip))
verify_leader_changed(master_leader_ip)
check_health()
def test_cleanup_then_kill_all_task_in_node(install_framework):
hosts = shakedown.get_service_ips(PACKAGE_NAME)
run_planned_operation(
run_cleanup,
lambda: [kill_task_with_pattern('CassandraDaemon', h) for h in hosts])
check_health()
def test_config_update_then_scheduler_died(install_framework):
host = get_scheduler_host()
run_planned_operation(
bump_cpu_count_config,
lambda: kill_task_with_pattern('cassandra.scheduler.Main', host))
check_health()
def test_cleanup_then_scheduler_died(install_framework):
host = get_scheduler_host()
run_planned_operation(
run_cleanup,
lambda: kill_task_with_pattern('cassandra.scheduler.Main', host))
check_health()
def test_all_executors_killed():
hosts = shakedown.get_service_ips(PACKAGE_NAME)
for host in hosts:
kill_task_with_pattern('cassandra.executor.Main', host)
recover_failed_agents(hosts)
check_health()
def test_repair_then_scheduler_died():
host = get_scheduler_host()
run_planned_operation(
run_repair,
lambda: kill_task_with_pattern('cassandra.scheduler.Main', host))
check_health()
def test_config_update_then_kill_task_in_node(install_framework):
host = get_node_host()
run_planned_operation(
bump_cpu_count_config,
lambda: kill_task_with_pattern('CassandraDaemon', host))
check_health()
def test_nodes_decrease_by_one_should_fail():
completed_plan = infinity_commons.get_and_verify_plan(
lambda p: p['status'] == infinity_commons.PlanState.COMPLETE.value)
mc = dcos.marathon.create_client()
app = mc.get_app('/cassandra')
app = infinity_commons.strip_meta(app)
oe = app['env']
env_node_count = int(oe['NODES']) - 1
oe['NODES'] = str(env_node_count)
app['env'] = oe
print("Updated node count: {}".format(app['env']['NODES']))
print(mc.update_app(app_id='/cassandra', payload=app, force=True))
check_health()
plan = infinity_commons.get_and_verify_plan(
lambda p: (p['status'] == infinity_commons.PlanState.ERROR.value and
len(infinity_commons.filter_phase(p, "Deploy")['steps']) == 3))
print(plan)
assert plan['status'] == infinity_commons.PlanState.ERROR.value
# Revert
oe = app['env']
env_node_count = int(oe['NODES']) + 1
oe['NODES'] = str(env_node_count)
app['env'] = oe
print("Reverted node count: {}".format(app['env']['NODES']))
print(mc.update_app(app_id='/cassandra', payload=app, force=True))
check_health()
plan = infinity_commons.get_and_verify_plan(
lambda p: (p['status'] == infinity_commons.PlanState.COMPLETE.value) and
(len(infinity_commons.filter_phase(p, "Deploy")['steps']) == 3))
print(plan)
assert plan['status'] == infinity_commons.PlanState.COMPLETE.value
def test_change_disk_should_fail():
completed_plan = infinity_commons.get_and_verify_plan(lambda p: p['status'] == infinity_commons.PlanState.COMPLETE.value)
mc = dcos.marathon.create_client()
app = mc.get_app('/cassandra')
app = infinity_commons.strip_meta(app)
oe = app['env']
disk = int(oe['CASSANDRA_DISK_MB']) - 1
oe['CASSANDRA_DISK_MB'] = str(disk)
app['env'] = oe
print("Updated CASSANDRA_DISK_MB: {}".format(app['env']['CASSANDRA_DISK_MB']))
print(mc.update_app(app_id='/cassandra', payload=app, force=True))
check_health()
plan = infinity_commons.get_and_verify_plan(lambda p: p['status'] == infinity_commons.PlanState.ERROR.value)
print(plan)
assert plan['status'] == infinity_commons.PlanState.ERROR.value
# Revert
oe = app['env']
disk = int(oe['CASSANDRA_DISK_MB']) + 1
oe['CASSANDRA_DISK_MB'] = str(disk)
app['env'] = oe
print("Reverted CASSANDRA_DISK_MB: {}".format(app['env']['CASSANDRA_DISK_MB']))
print(mc.update_app(app_id='/cassandra', payload=app, force=True))
check_health()
plan = infinity_commons.get_and_verify_plan(lambda p: p['status'] == infinity_commons.PlanState.COMPLETE.value)
print(plan)
assert plan['status'] == infinity_commons.PlanState.COMPLETE.value
def test_config_update_then_executor_killed(install_framework):
host = get_node_host()
run_planned_operation(
bump_cpu_count_config,
lambda: kill_task_with_pattern('cassandra.executor.Main', host))
check_health()
def test_cleanup_then_executor_killed(install_framework):
host = get_node_host()
run_planned_operation(
run_cleanup,
lambda: kill_task_with_pattern('cassandra.executor.Main', host))
check_health()
def install_framework():
uninstall()
shakedown.install_package_and_wait(PACKAGE_NAME)
check_health()
yield
uninstall()
def test_partition():
host = get_node_host()
shakedown.partition_agent(host)
shakedown.reconnect_agent(host)
check_health()
def test_cleanup_then_all_executors_killed(install_framework):
hosts = shakedown.get_service_ips(PACKAGE_NAME)
run_planned_operation(
run_cleanup(), lambda:
[kill_task_with_pattern('cassandra.executor.Main', h) for h in hosts])
check_health()
def test_repair_then_master_killed():
run_planned_operation(
run_repair,
lambda: kill_task_with_pattern('mesos-master')
)
check_health()
def test_repair_then_zk_killed():
run_planned_operation(
run_repair,
lambda: kill_task_with_pattern('zookeeper')
)
check_health()
def test_partition(install_framework):
host = get_node_host()
shakedown.partition_agent(host)
shakedown.reconnect_agent(host)
check_health()
def install_framework():
shakedown.install_package_and_wait(PACKAGE_NAME)
check_health()
yield
uninstall()
def test_nodes_increase_by_one():
completed_plan = infinity_commons.get_and_verify_plan(
lambda p: p['status'] == infinity_commons.PlanState.COMPLETE.value)
mc = dcos.marathon.create_client()
app = mc.get_app('/cassandra')
app = infinity_commons.strip_meta(app)
oe = app['env']
env_node_count = int(oe['NODES']) + 1
oe['NODES'] = str(env_node_count)
app['env'] = oe
print("Updated node count: {}".format(app['env']['NODES']))
print(mc.update_app(app_id='/cassandra', payload=app, force=True))
check_health()
plan = infinity_commons.get_and_verify_plan(lambda p: (
p['status'] == infinity_commons.PlanState.COMPLETE.value and len(
infinity_commons.filter_phase(p, "Deploy")['steps']) == 4 and
(infinity_commons.filter_phase(p, "Deploy")['steps'][
env_node_count - 1]['status'] == infinity_commons.PlanState.
COMPLETE.value)))
print(plan)
assert plan['status'] == infinity_commons.PlanState.COMPLETE.value
# reinstall after increase:
uninstall()
install()
check_health()
def test_change_disk_should_fail():
completed_plan = infinity_commons.get_and_verify_plan(
lambda p: p['status'] == infinity_commons.PlanState.COMPLETE.value)
mc = dcos.marathon.create_client()
app = mc.get_app('/cassandra')
app = infinity_commons.strip_meta(app)
oe = app['env']
disk = int(oe['CASSANDRA_DISK_MB']) - 1
oe['CASSANDRA_DISK_MB'] = str(disk)
app['env'] = oe
print("Updated CASSANDRA_DISK_MB: {}".format(
app['env']['CASSANDRA_DISK_MB']))
print(mc.update_app(app_id='/cassandra', payload=app, force=True))
check_health()
plan = infinity_commons.get_and_verify_plan(
lambda p: p['status'] == infinity_commons.PlanState.ERROR.value)
print(plan)
assert plan['status'] == infinity_commons.PlanState.ERROR.value
# Revert
oe = app['env']
disk = int(oe['CASSANDRA_DISK_MB']) + 1
oe['CASSANDRA_DISK_MB'] = str(disk)
app['env'] = oe
print("Reverted CASSANDRA_DISK_MB: {}".format(
app['env']['CASSANDRA_DISK_MB']))
print(mc.update_app(app_id='/cassandra', payload=app, force=True))
check_health()
plan = infinity_commons.get_and_verify_plan(
lambda p: p['status'] == infinity_commons.PlanState.COMPLETE.value)
print(plan)
assert plan['status'] == infinity_commons.PlanState.COMPLETE.value
def test_upgrade_downgrade():
test_repo_name, test_repo_url = get_test_repo_info()
test_version = get_pkg_version()
print('Found test version: {}'.format(test_version))
remove_repo(test_repo_name, test_version)
master_version = get_pkg_version()
print('Found master version: {}'.format(master_version))
print('Installing master version')
install(package_version = master_version)
check_health()
plan = infinity_commons.get_and_verify_plan(lambda p: p['status'] == infinity_commons.PlanState.COMPLETE.value)
assert plan['status'] == infinity_commons.PlanState.COMPLETE.value
# TODO: write some data
print('Upgrading to test version')
destroy_service()
add_repo(test_repo_name, test_repo_url, master_version)
install(package_version = test_version)
check_post_version_change_health()
print('Downgrading to master version')
destroy_service()
install(package_version = master_version)
check_post_version_change_health()
def test_cleanup_then_kill_task_in_node(install_framework):
host = get_node_host()
run_planned_operation(
run_cleanup,
lambda: kill_task_with_pattern('CassandraDaemon', host)
)
check_health()
def test_repair_then_zk_killed():
master_leader_ip = shakedown.master_leader_ip()
run_planned_operation(
run_repair,
lambda: kill_task_with_pattern('zookeeper', master_leader_ip),
lambda: verify_leader_changed(master_leader_ip))
check_health()
def test_cleanup_then_scheduler_died(install_framework):
host = get_scheduler_host()
run_planned_operation(
run_cleanup,
lambda: kill_task_with_pattern('cassandra.scheduler.Main', host)
)
check_health()
def test_cleanup_then_master_killed():
master_leader_ip = shakedown.master_leader_ip()
run_planned_operation(
run_cleanup,
lambda: kill_task_with_pattern('mesos-master', master_leader_ip))
verify_leader_changed(master_leader_ip)
check_health()
def test_config_update_then_all_executors_killed():
hosts = shakedown.get_service_ips(PACKAGE_NAME)
run_planned_operation(
bump_cpu_count_config, lambda:
[kill_task_with_pattern('cassandra.executor.Main', h) for h in hosts],
lambda: recover_failed_agents(hosts))
check_health()
def test_cleanup_then_executor_killed(install_framework):
host = get_node_host()
run_planned_operation(
run_cleanup,
lambda: kill_task_with_pattern('cassandra.executor.Main', host)
)
check_health()
def test_cleanup_then_kill_all_task_in_node(install_framework):
hosts = shakedown.get_service_ips(PACKAGE_NAME)
run_planned_operation(
run_cleanup,
lambda: [kill_task_with_pattern('CassandraDaemon', h) for h in hosts]
)
check_health()
def test_config_update_then_zk_killed():
master_leader_ip = shakedown.master_leader_ip()
run_planned_operation(
bump_cpu_count_config,
lambda: kill_task_with_pattern('zookeeper', master_leader_ip),
lambda: verify_leader_changed(master_leader_ip))
check_health()
def test_config_update_then_executor_killed():
host = get_node_host()
run_planned_operation(
lambda: bump_cpu_count_config(-0.1),
lambda: kill_task_with_pattern('cassandra.executor.Main', host),
lambda: recover_failed_agents([host]))
check_health()
def test_repair_then_scheduler_died():
host = get_scheduler_host()
run_planned_operation(
run_repair,
lambda: kill_task_with_pattern('cassandra.scheduler.Main', host)
)
check_health()
def test_config_update_then_executor_killed():
host = get_node_host()
run_planned_operation(
bump_cpu_count_config,
lambda: kill_task_with_pattern('cassandra.executor.Main', host)
)
check_health()
def test_repair_then_executor_killed():
host = get_node_host()
run_planned_operation(
run_repair,
lambda: kill_task_with_pattern('cassandra.executor.Main', host)
)
check_health()
def test_config_update_then_kill_all_task_in_node():
hosts = shakedown.get_service_ips(PACKAGE_NAME)
run_planned_operation(
bump_cpu_count_config,
lambda: [kill_task_with_pattern('CassandraDaemon', h) for h in hosts]
)
check_health()
def test_config_update_then_scheduler_died():
host = get_scheduler_host()
run_planned_operation(
bump_cpu_count_config,
lambda: kill_task_with_pattern('cassandra.scheduler.Main', host)
)
check_health()
def test_config_update_then_kill_task_in_node():
host = get_node_host()
run_planned_operation(
bump_cpu_count_config,
lambda: kill_task_with_pattern('CassandraDaemon', host)
)
check_health()
def test_partition():
host = get_node_host()
_block_on_adminrouter()
shakedown.partition_agent(host)
shakedown.reconnect_agent(host)
check_health()
def test_repair_then_kill_all_task_in_node():
hosts = shakedown.get_service_ips(PACKAGE_NAME)
run_planned_operation(
run_repair,
lambda: [kill_task_with_pattern('CassandraDaemon', h) for h in hosts]
)
check_health()
def test_repair_then_kill_task_in_node():
host = get_node_host()
run_planned_operation(
run_repair,
lambda: kill_task_with_pattern('CassandraDaemon', host)
)
check_health()
def test_all_partition():
hosts = shakedown.get_service_ips(PACKAGE_NAME)
for host in hosts:
shakedown.partition_agent(host)
for host in hosts:
shakedown.reconnect_agent(host)
check_health()
def test_all_partition():
hosts = shakedown.get_service_ips(PACKAGE_NAME)
for host in hosts:
shakedown.partition_agent(host)
for host in hosts:
shakedown.reconnect_agent(host)
check_health()
def test_repair_then_kill_task_in_node():
hosts = shakedown.get_service_ips(PACKAGE_NAME)
host = get_node_host()
run_planned_operation(
run_repair, lambda: kill_task_with_pattern('CassandraDaemon', host),
lambda: recover_failed_agents(hosts))
check_health()
def test_cleanup_then_executor_killed():
host = get_node_host()
run_planned_operation(
run_cleanup,
lambda: kill_task_with_pattern('cassandra.executor.Main', host),
lambda: recover_failed_agents([host]))
check_health()
def test_cleanup_then_all_executors_killed(install_framework):
hosts = shakedown.get_service_ips(PACKAGE_NAME)
run_planned_operation(
run_cleanup(),
lambda: [
kill_task_with_pattern('cassandra.executor.Main', h) for h in hosts
]
)
check_health()
def test_repair_then_partition():
host = get_node_host()
def partition():
shakedown.partition_agent(host)
shakedown.reconnect_agent(host)
run_planned_operation(run_repair, partition)
check_health()
def test_repair_then_all_executors_killed():
hosts = shakedown.get_service_ips(PACKAGE_NAME)
run_planned_operation(
run_repair,
lambda: [
kill_task_with_pattern('cassandra.executor.Main', h) for h in hosts
]
)
check_health()
def test_config_update_then_all_executors_killed():
hosts = shakedown.get_service_ips(PACKAGE_NAME)
run_planned_operation(
bump_cpu_count_config,
lambda: [
kill_task_with_pattern('cassandra.executor.Main', h) for h in hosts
]
)
check_health()
def test_config_update_then_partition():
host = get_node_host()
def partition():
shakedown.partition_agent(host)
shakedown.reconnect_agent(host)
run_planned_operation(bump_cpu_count_config, partition)
check_health()
def test_cleanup_then_partition(install_framework):
host = get_node_host()
def partition():
shakedown.partition_agent(host)
shakedown.reconnect_agent(host)
run_planned_operation(run_cleanup, partition)
check_health()
def test_cleanup_then_all_partition(install_framework):
hosts = shakedown.get_service_ips(PACKAGE_NAME)
def partition():
for host in hosts:
shakedown.partition_agent(host)
for host in hosts:
shakedown.reconnect_agent(host)
run_planned_operation(run_cleanup, partition)
check_health()
def test_repair_then_all_partition():
hosts = shakedown.get_service_ips(PACKAGE_NAME)
def partition():
for host in hosts:
shakedown.partition_agent(host)
for host in hosts:
shakedown.reconnect_agent(host)
run_planned_operation(run_repair, partition)
check_health()
def test_config_update_then_all_partition():
hosts = shakedown.get_service_ips(PACKAGE_NAME)
def partition():
for host in hosts:
shakedown.partition_agent(host)
for host in hosts:
shakedown.reconnect_agent(host)
run_planned_operation(bump_cpu_count_config, partition)
check_health()