class RebalanceinJython(BaseTestCase):
def setUp(self):
super(RebalanceinJython, self).setUp()
self.task_manager = TaskManager()
self.load_gen_tasks = []
def tearDown(self):
self.task_manager.shutdown_task_manager()
super(RebalanceinJython, self).tearDown()
def start_load_gen(self, docs, bucket):
cluster = CouchbaseCluster.create(self.servers[0].ip)
cluster.authenticate("Administrator", "password")
bucket = cluster.openBucket(bucket)
k = "rebalance"
v = {"value": "asd"}
docloaders = []
num_executors = 5
total_num_executors = 5
num_docs = docs / total_num_executors
load_gen_task_name = "Loadgen"
for i in xrange(total_num_executors):
task_name = "{0}_{1}".format(load_gen_task_name, i)
self.load_gen_tasks.append(
DocloaderTask(bucket, num_docs, i * num_docs, k, v, task_name))
for task in self.load_gen_tasks:
self.task_manager.add_new_task(task)
def finish_load_gen(self):
for task in self.load_gen_tasks:
print self.task_manager.get_task_result(task)
def start_rebalance_nodes(self):
servs_in = [
self.servers[i + self.nodes_init] for i in range(self.nodes_in)
]
self.rebalance_task = rebalanceTask(self.servers[:self.nodes_init],
to_add=servs_in,
to_remove=[])
self.task_manager.add_new_task(self.rebalance_task)
def get_rebalance_result(self):
return self.task_manager.get_task_result(self.rebalance_task)
def test_rebalance_in(self):
self.start_load_gen(100000, self.buckets[0].name)
self.sleep(5)
#test_Task = TestTask(30)
#self.task_manager.add_new_task(test_Task)
self.start_rebalance_nodes()
print self.get_rebalance_result()
#print self.task_manager.get_task_result(test_Task)
self.finish_load_gen()
class CBASClusterOperations(CBASBaseTest):
def setUp(self):
self.input = TestInputSingleton.input
self.input.test_params.update({"default_bucket": False})
self.rebalanceServers = None
self.nodeType = "KV"
self.wait_for_rebalance = True
super(CBASClusterOperations, self).setUp()
self.num_items = self.input.param("items", 1000)
self.bucket_util.create_default_bucket()
# self.cbas_util.createConn("default")
if 'nodeType' in self.input.test_params:
self.nodeType = self.input.test_params['nodeType']
self.rebalance_both = self.input.param("rebalance_cbas_and_kv", False)
if not self.rebalance_both:
if self.nodeType == "KV":
self.rebalanceServers = self.kv_servers
self.wait_for_rebalance = False
elif self.nodeType == "CBAS":
self.rebalanceServers = [self.cbas_node] + self.cbas_servers
else:
self.rebalanceServers = self.kv_servers + [self.cbas_node
] + self.cbas_servers
self.nodeType = "KV" + "-" + "CBAS"
self.assertTrue(
len(self.rebalanceServers) > 1,
"Not enough %s servers to run tests." % self.rebalanceServers)
self.log.info("This test will be running in %s context." %
self.nodeType)
self.task_manager = TaskManager()
self.load_gen_tasks = []
def setup_for_test(self, skip_data_loading=False):
if not skip_data_loading:
# Load Couchbase bucket first.
self.perform_doc_ops_in_all_cb_buckets(self.num_items, "create", 0,
self.num_items)
self.cbas_util.createConn(self.cb_bucket_name)
# Create bucket on CBAS
self.assertTrue(
self.cbas_util.create_bucket_on_cbas(
cbas_bucket_name=self.cbas_bucket_name,
cb_bucket_name=self.cb_bucket_name,
cb_server_ip=self.cb_server_ip),
"bucket creation failed on cbas")
# Create dataset on the CBAS bucket
self.cbas_util.create_dataset_on_bucket(
cbas_bucket_name=self.cb_bucket_name,
cbas_dataset_name=self.cbas_dataset_name)
# Create indexes on the CBAS bucket
self.create_secondary_indexes = self.input.param(
"create_secondary_indexes", False)
if self.create_secondary_indexes:
self.index_fields = "profession:string,number:bigint"
create_idx_statement = "create index {0} on {1}({2});".format(
self.index_name, self.cbas_dataset_name, self.index_fields)
status, metrics, errors, results, _ = self.cbas_util.execute_statement_on_cbas_util(
create_idx_statement)
self.assertTrue(status == "success", "Create Index query failed")
self.assertTrue(
self.cbas_util.verify_index_created(
self.index_name, self.index_fields.split(","),
self.cbas_dataset_name)[0])
# Connect to Bucket
self.cbas_util.connect_to_bucket(
cbas_bucket_name=self.cbas_bucket_name,
cb_bucket_password=self.cb_bucket_password)
if not skip_data_loading:
# Validate no. of items in CBAS dataset
if not self.cbas_util.validate_cbas_dataset_items_count(
self.cbas_dataset_name, self.num_items):
self.fail(
"No. of items in CBAS dataset do not match that in the CB bucket"
)
def test_rebalance_in(self):
'''
Description: This will test the rebalance in feature i.e. one node coming in to the cluster.
Then Rebalance. Verify that is has no effect on the data ingested to cbas.
Steps:
1. Setup cbas. bucket, datasets/shadows, connect.
2. Add a node and rebalance. Don't wait for rebalance completion.
3. During rebalance, do mutations and execute queries on cbas.
Author: Ritesh Agarwal/Mihir Kamdar
Date Created: 18/07/2017
'''
query = "select count(*) from {0};".format(self.cbas_dataset_name)
self.setup_for_test()
self.add_node(node=self.rebalanceServers[1],
rebalance=True,
wait_for_rebalance_completion=self.wait_for_rebalance)
self.log.info("Rebalance state:%s" %
self.rest._rebalance_progress_status())
self.perform_doc_ops_in_all_cb_buckets(self.num_items, "create",
self.num_items,
self.num_items * 2)
self.log.info("Rebalance state:%s" %
self.rest._rebalance_progress_status())
self.cbas_util._run_concurrent_queries(
query, None, 2000, batch_size=self.concurrent_batch_size)
self.log.info("Rebalance state:%s" %
self.rest._rebalance_progress_status())
if not self.cbas_util.validate_cbas_dataset_items_count(
self.cbas_dataset_name, self.num_items * 2, 0):
self.fail(
"No. of items in CBAS dataset do not match that in the CB bucket"
)
def test_rebalance_out(self):
'''
Description: This will test the rebalance out feature i.e. one node going out of cluster.
Then Rebalance.
Steps:
1. Add a node, Rebalance.
2. Setup cbas. bucket, datasets/shadows, connect.
3. Remove a node and rebalance. Don't wait for rebalance completion.
4. During rebalance, do mutations and execute queries on cbas.
Author: Ritesh Agarwal/Mihir Kamdar
Date Created: 18/07/2017
'''
self.add_node(node=self.rebalanceServers[1])
query = "select count(*) from {0};".format(self.cbas_dataset_name)
self.setup_for_test()
otpnodes = []
nodes = self.rest.node_statuses()
for node in nodes:
if node.ip == self.rebalanceServers[1].ip:
otpnodes.append(node)
self.remove_node(otpnodes, wait_for_rebalance=self.wait_for_rebalance)
self.log.info("Rebalance state:%s" %
self.rest._rebalance_progress_status())
self.perform_doc_ops_in_all_cb_buckets(self.num_items, "create",
self.num_items,
self.num_items * 2)
self.cbas_util._run_concurrent_queries(
query, "immediate", 2000, batch_size=self.concurrent_batch_size)
if not self.cbas_util.validate_cbas_dataset_items_count(
self.cbas_dataset_name, self.num_items * 2, 0):
self.fail(
"No. of items in CBAS dataset do not match that in the CB bucket"
)
def test_swap_rebalance(self):
'''
Description: This will test the swap rebalance feature i.e. one node going out and one node coming in cluster.
Then Rebalance. Verify that is has no effect on the data ingested to cbas.
Steps:
1. Setup cbas. bucket, datasets/shadows, connect.
2. Add a node that is to be swapped against the leaving node. Do not rebalance.
3. Remove a node and rebalance.
4. During rebalance, do mutations and execute queries on cbas.
Author: Ritesh Agarwal/Mihir Kamdar
Date Created: 20/07/2017
'''
query = "select count(*) from {0};".format(self.cbas_dataset_name)
self.setup_for_test()
otpnodes = []
nodes = self.rest.node_statuses()
if self.nodeType == "KV":
service = ["kv"]
else:
service = ["cbas"]
otpnodes.append(self.add_node(node=self.servers[1], services=service))
self.add_node(node=self.servers[3], services=service, rebalance=False)
self.remove_node(otpnodes, wait_for_rebalance=self.wait_for_rebalance)
self.perform_doc_ops_in_all_cb_buckets(self.num_items, "create",
self.num_items,
self.num_items * 2)
self.cbas_util._run_concurrent_queries(
query, "immediate", 2000, batch_size=self.concurrent_batch_size)
if not self.cbas_util.validate_cbas_dataset_items_count(
self.cbas_dataset_name, self.num_items * 2, 0):
self.fail(
"No. of items in CBAS dataset do not match that in the CB bucket"
)
def test_failover(self):
'''
Description: This will test the node failover both graceful and hard failover based on
graceful_failover param in testcase conf file.
Steps:
1. Add node to the cluster which will be failed over.
2. Create docs, setup cbas.
3. Mark the node for fail over.
4. Do rebalance asynchronously. During rebalance perform mutations.
5. Run some CBAS queries.
6. Check for correct number of items in CBAS datasets.
Author: Ritesh Agarwal/Mihir Kamdar
Date Created: 20/07/2017
'''
#Add node which will be failed over later.
self.add_node(node=self.rebalanceServers[1])
query = "select count(*) from {0};".format(self.cbas_dataset_name)
graceful_failover = self.input.param("graceful_failover", False)
self.setup_for_test()
failover_task = self._cb_cluster.async_failover(
self.input.servers, [self.rebalanceServers[1]], graceful_failover)
failover_task.get_result()
self.rebalance()
self.perform_doc_ops_in_all_cb_buckets(self.num_items, "create",
self.num_items,
self.num_items * 3 / 2)
self.cbas_util._run_concurrent_queries(
query, "immediate", 2000, batch_size=self.concurrent_batch_size)
if not self.cbas_util.validate_cbas_dataset_items_count(
self.cbas_dataset_name, self.num_items * 3 / 2, 0):
self.fail(
"No. of items in CBAS dataset do not match that in the CB bucket"
)
'''
-i b/resources/4-nodes-template.ini -t cbas.cbas_cluster_operations.CBASClusterOperations.test_rebalance_in_cb_cbas_together,cb_bucket_name=default,cbas_bucket_name=default_bucket,cbas_dataset_name=default_ds,items=10,nodeType=KV,rebalance_cbas_and_kv=True,wait_for_rebalace=False
'''
def test_rebalance_in_cb_cbas_together(self):
self.log.info("Creates cbas buckets and dataset")
dataset_count_query = "select count(*) from {0};".format(
self.cbas_dataset_name)
self.setup_for_test()
self.log.info("Rebalance in KV node")
wait_for_rebalace_complete = self.input.param("wait_for_rebalace",
False)
self.add_node(node=self.rebalanceServers[1],
rebalance=False,
wait_for_rebalance_completion=wait_for_rebalace_complete)
self.log.info("Rebalance in CBAS node")
self.add_node(node=self.rebalanceServers[3],
rebalance=True,
wait_for_rebalance_completion=wait_for_rebalace_complete)
self.log.info(
"Perform document create as rebalance is in progress : Rebalance state:%s"
% self.rest._rebalance_progress_status())
self.perform_doc_ops_in_all_cb_buckets(self.num_items, "create",
self.num_items,
self.num_items * 2)
self.log.info(
"Run queries as rebalance is in progress : Rebalance state:%s" %
self.rest._rebalance_progress_status())
handles = self.cbas_util._run_concurrent_queries(
dataset_count_query,
None,
2000,
batch_size=self.concurrent_batch_size)
self.log.info("Log concurrent query status")
self.cbas_util.log_concurrent_query_outcome(self.master, handles)
if not self.cbas_util.validate_cbas_dataset_items_count(
self.cbas_dataset_name, self.num_items * 2, 0):
self.fail(
"No. of items in CBAS dataset do not match that in the CB bucket"
)
'''
-i b/resources/4-nodes-template.ini -t cbas.cbas_cluster_operations.CBASClusterOperations.test_rebalance_out_cb_cbas_together,cb_bucket_name=default,cbas_bucket_name=default_bucket,cbas_dataset_name=default_ds,items=10,nodeType=KV,rebalance_cbas_and_kv=True,wait_for_rebalace=False
'''
def test_rebalance_out_cb_cbas_together(self):
self.log.info(
"Rebalance in KV node and wait for rebalance to complete")
self.add_node(node=self.rebalanceServers[1])
self.log.info(
"Rebalance in CBAS node and wait for rebalance to complete")
self.add_node(node=self.rebalanceServers[3])
self.log.info("Creates cbas buckets and dataset")
dataset_count_query = "select count(*) from {0};".format(
self.cbas_dataset_name)
self.setup_for_test()
self.log.info("Fetch and remove nodes to rebalance out")
wait_for_rebalace_complete = self.input.param("wait_for_rebalace",
False)
otpnodes = []
nodes = self.rest.node_statuses()
for node in nodes:
if node.ip == self.rebalanceServers[
1].ip or node.ip == self.rebalanceServers[3].ip:
otpnodes.append(node)
for every_node in otpnodes:
self.remove_node(otpnodes,
wait_for_rebalance=wait_for_rebalace_complete)
self.sleep(30,
message="Sleep for 30 seconds for remove node to complete")
self.log.info(
"Perform document create as rebalance is in progress : Rebalance state:%s"
% self.rest._rebalance_progress_status())
self.perform_doc_ops_in_all_cb_buckets(self.num_items, "create",
self.num_items,
self.num_items * 2)
self.log.info(
"Run queries as rebalance is in progress : Rebalance state:%s" %
self.rest._rebalance_progress_status())
handles = self.cbas_util._run_concurrent_queries(
dataset_count_query,
"immediate",
2000,
batch_size=self.concurrent_batch_size)
self.log.info("Log concurrent query status")
self.cbas_util.log_concurrent_query_outcome(self.master, handles)
if not self.cbas_util.validate_cbas_dataset_items_count(
self.cbas_dataset_name, self.num_items * 2, 0):
self.fail(
"No. of items in CBAS dataset do not match that in the CB bucket"
)
'''
-i b/resources/4-nodes-template.ini -t cbas.cbas_cluster_operations.CBASClusterOperations.test_swap_rebalance_cb_cbas_together,cb_bucket_name=default,cbas_bucket_name=default_bucket,cbas_dataset_name=default_ds,items=10,rebalance_cbas_and_kv=True,wait_for_rebalance=True
'''
def test_swap_rebalance_cb_cbas_together(self):
self.log.info("Creates cbas buckets and dataset")
wait_for_rebalance = self.input.param("wait_for_rebalance", True)
dataset_count_query = "select count(*) from {0};".format(
self.cbas_dataset_name)
self.setup_for_test()
self.log.info("Add KV node and don't rebalance")
self.add_node(node=self.rebalanceServers[1], rebalance=False)
self.log.info("Add cbas node and don't rebalance")
self.add_node(node=self.rebalanceServers[3], rebalance=False)
otpnodes = []
nodes = self.rest.node_statuses()
for node in nodes:
if node.ip == self.rebalanceServers[
0].ip or node.ip == self.rebalanceServers[2].ip:
otpnodes.append(node)
self.log.info("Remove master node")
self.remove_node(otpnode=otpnodes,
wait_for_rebalance=wait_for_rebalance)
self.log.info("Create instances pointing to new master nodes")
c_utils = cbas_utils(self.rebalanceServers[1],
self.rebalanceServers[3])
c_utils.createConn(self.cb_bucket_name)
self.log.info("Create reference to SDK client")
client = SDKClient(scheme="couchbase",
hosts=[self.rebalanceServers[1].ip],
bucket=self.cb_bucket_name,
password=self.rebalanceServers[1].rest_password)
self.log.info("Add more document to default bucket")
documents = ['{"name":"value"}'] * (self.num_items // 10)
document_id_prefix = "custom-id-"
client.insert_custom_json_documents(document_id_prefix, documents)
self.log.info(
"Run queries as rebalance is in progress : Rebalance state:%s" %
self.rest._rebalance_progress_status())
handles = c_utils._run_concurrent_queries(
dataset_count_query,
"immediate",
2000,
batch_size=self.concurrent_batch_size)
self.log.info("Log concurrent query status")
self.cbas_util.log_concurrent_query_outcome(self.master, handles)
if not c_utils.validate_cbas_dataset_items_count(
self.cbas_dataset_name, self.num_items +
(self.num_items // 10), 0):
self.fail(
"No. of items in CBAS dataset do not match that in the CB bucket"
)
def test_rebalance_in_multiple_cbas_on_a_busy_system(self):
node_services = []
node_services.append(self.input.param('service', "cbas"))
self.log.info("Setup CBAS")
self.setup_for_test(skip_data_loading=True)
self.log.info("Run KV ops in async while rebalance is in progress")
json_generator = JsonGenerator()
generators = json_generator.generate_docs_simple(
docs_per_day=self.num_items, start=0)
tasks = self.bucket_util._async_load_all_buckets(
self.cluster.master, generators, "create", 0)
self.log.info("Run concurrent queries to simulate busy system")
statement = "select sleep(count(*),50000) from {0} where mutated=0;".format(
self.cbas_dataset_name)
handles = self.cbas_util._run_concurrent_queries(
statement, self.mode, self.num_concurrent_queries)
self.log.info("Rebalance in CBAS nodes")
self.cluster_util.add_node(node=self.rebalanceServers[1],
services=node_services,
rebalance=False,
wait_for_rebalance_completion=False)
self.cluster_util.add_node(node=self.rebalanceServers[3],
services=node_services,
rebalance=True,
wait_for_rebalance_completion=True)
self.log.info("Get KV ops result")
for task in tasks:
task.get_result()
self.log.info("Log concurrent query status")
self.cbas_util.log_concurrent_query_outcome(self.cluster.master,
handles)
if not self.cbas_util.validate_cbas_dataset_items_count(
self.cbas_dataset_name, self.num_items, 0):
self.fail(
"No. of items in CBAS dataset do not match that in the CB bucket"
)
def _start_load_gen(self, docs, bucket, num_executors):
cluster = CouchbaseCluster.create(self.servers[0].ip)
cluster.authenticate("Administrator", "password")
bucket = cluster.openBucket(bucket)
k = "rebalance"
v = {"value": "asd"}
docloaders = []
total_num_executors = num_executors
num_docs = docs / total_num_executors
load_gen_task_name = "Loadgen"
for i in xrange(total_num_executors):
task_name = "{0}_{1}".format(load_gen_task_name, i)
self.load_gen_tasks.append(
DocloaderTask(bucket, num_docs, i * num_docs, k, v, task_name))
for task in self.load_gen_tasks:
self.task_manager.add_new_task(task)
def _finish_load_gen(self):
for task in self.load_gen_tasks:
self.log.info(self.task_manager.get_task_result(task))
def test_rebalance_in_multiple_cbas_on_a_busy_system_jython(self):
services = []
services.append(self.input.param('service', "cbas"))
services.append(self.input.param('service', 'cbas'))
self.log.info("Setup CBAS")
self.setup_for_test(skip_data_loading=True)
self.log.info("Run KV ops in background")
num_executors = 9 // self.buckets.__len__()
for bucket in self.buckets:
self._start_load_gen(self.num_items, bucket.name, num_executors)
self.log.info("Run concurrent queries to simulate busy system")
statement = "select sleep(count(*),50000) from {0} where mutated=0;".format(
self.cbas_dataset_name)
handles = self.cbas_util._run_concurrent_queries(
statement, self.mode, self.num_concurrent_queries)
self.log.info("Rebalance in CBAS nodes")
nodes_in = [self.rebalanceServers[1], self.rebalanceServers[3]]
rebalance_task = rebalanceTask(self.servers[:2],
to_add=nodes_in,
to_remove=[],
services=services)
self.task_manager.add_new_task(rebalance_task)
rebalance_result = self.task_manager.get_task_result(rebalance_task)
self.log.info(rebalance_result)
self.log.info("Get KV ops result")
self._finish_load_gen()
self.log.info("Log concurrent query status")
self.cbas_util.log_concurrent_query_outcome(self.master, handles)
if not self.cbas_util.validate_cbas_dataset_items_count(
self.cbas_dataset_name, self.num_items, 0):
self.fail(
"No. of items in CBAS dataset do not match that in the CB bucket"
)
def test_rebalance_out_multiple_cbas_on_a_busy_system(self):
node_services = []
node_services.append(self.input.param('service', "cbas"))
self.log.info("Rebalance in CBAS nodes")
self.add_node(node=self.rebalanceServers[1], services=node_services)
self.add_node(node=self.rebalanceServers[3], services=node_services)
self.log.info("Setup CBAS")
self.setup_for_test(skip_data_loading=True)
self.log.info("Run KV ops in async while rebalance is in progress")
json_generator = JsonGenerator()
generators = json_generator.generate_docs_simple(
docs_per_day=self.num_items, start=0)
tasks = self._async_load_all_buckets(self.master, generators, "create",
0)
self.log.info("Run concurrent queries to simulate busy system")
statement = "select sleep(count(*),50000) from {0} where mutated=0;".format(
self.cbas_dataset_name)
handles = self.cbas_util._run_concurrent_queries(
statement, self.mode, self.num_concurrent_queries)
self.log.info("Fetch and remove nodes to rebalance out")
self.rebalance_cc = self.input.param("rebalance_cc", False)
out_nodes = []
nodes = self.rest.node_statuses()
if self.rebalance_cc:
for node in nodes:
if node.ip == self.cbas_node.ip or node.ip == self.servers[
1].ip:
out_nodes.append(node)
self.cbas_util.closeConn()
self.log.info(
"Reinitialize CBAS utils with ip %s, since CC node is rebalanced out"
% self.servers[3].ip)
self.cbas_util = cbas_utils(self.master, self.servers[3])
self.cbas_util.createConn("default")
else:
for node in nodes:
if node.ip == self.servers[3].ip or node.ip == self.servers[
1].ip:
out_nodes.append(node)
self.log.info("Rebalance out CBAS nodes %s %s" %
(out_nodes[0].ip, out_nodes[1].ip))
self.remove_all_nodes_then_rebalance([out_nodes[0], out_nodes[1]])
self.log.info("Get KV ops result")
for task in tasks:
task.get_result()
self.log.info("Log concurrent query status")
self.cbas_util.log_concurrent_query_outcome(self.master, handles)
if not self.cbas_util.validate_cbas_dataset_items_count(
self.cbas_dataset_name, self.num_items, 0):
self.fail(
"No. of items in CBAS dataset do not match that in the CB bucket"
)
'''
cbas.cbas_cluster_operations.CBASClusterOperations.test_rebalance_swap_multiple_cbas_on_a_busy_system,cb_bucket_name=default,cbas_bucket_name=default_bucket,cbas_dataset_name=default_ds,items=10,rebalance_cbas_and_kv=True,service=cbas,rebalance_cc=False
cbas.cbas_cluster_operations.CBASClusterOperations.test_rebalance_swap_multiple_cbas_on_a_busy_system,cb_bucket_name=default,cbas_bucket_name=default_bucket,cbas_dataset_name=default_ds,items=10,rebalance_cbas_and_kv=True,service=cbas,rebalance_cc=True
'''
def test_rebalance_swap_multiple_cbas_on_a_busy_system(self):
'''
1. We have 4 node cluster with 1 KV and 3 CBAS. Assume the IPS end with 101(KV), 102(CBAS), 103(CBAS), 104(CBAS)
2, Post initial setup - 101 running KV and 102 running CBAS as CC node
3. As part of test test add an extra NC node that we will swap rebalance later - Adding 103 and rebalance
4. If swap rebalance NC - then select the node added in #3 for remove and 104 to add during swap
5. If swap rebalance CC - then select the CC node added for remove and 104 to add during swap
'''
self.log.info('Read service input param')
node_services = []
node_services.append(self.input.param('service', "cbas"))
self.log.info(
"Rebalance in CBAS nodes, this node will be removed during swap")
self.add_node(node=self.rebalanceServers[1], services=node_services)
self.log.info("Setup CBAS")
self.setup_for_test(skip_data_loading=True)
self.log.info("Run KV ops in async while rebalance is in progress")
json_generator = JsonGenerator()
generators = json_generator.generate_docs_simple(
docs_per_day=self.num_items, start=0)
tasks = self._async_load_all_buckets(self.master, generators, "create",
0)
self.log.info("Run concurrent queries to simulate busy system")
statement = "select sleep(count(*),50000) from {0} where mutated=0;".format(
self.cbas_dataset_name)
handles = self.cbas_util._run_concurrent_queries(
statement, self.mode, self.num_concurrent_queries)
self.log.info("Fetch node to remove during rebalance")
self.rebalance_cc = self.input.param("rebalance_cc", False)
out_nodes = []
nodes = self.rest.node_statuses()
reinitialize_cbas_util = False
for node in nodes:
if self.rebalance_cc and (node.ip == self.cbas_node.ip):
out_nodes.append(node)
reinitialize_cbas_util = True
elif not self.rebalance_cc and node.ip == self.rebalanceServers[
1].ip:
out_nodes.append(node)
self.log.info("Swap rebalance CBAS nodes")
self.add_node(node=self.rebalanceServers[3],
services=node_services,
rebalance=False)
self.remove_node([out_nodes[0]], wait_for_rebalance=True)
self.log.info("Get KV ops result")
for task in tasks:
task.get_result()
if reinitialize_cbas_util is True:
self.cbas_util = cbas_utils(self.master, self.rebalanceServers[3])
self.cbas_util.createConn("default")
self.log.info("Log concurrent query status")
self.cbas_util.log_concurrent_query_outcome(self.master, handles)
if not self.cbas_util.validate_cbas_dataset_items_count(
self.cbas_dataset_name, self.num_items, 0):
self.fail(
"No. of items in CBAS dataset do not match that in the CB bucket"
)
'''
test_fail_over_node_followed_by_rebalance_out_or_add_back,cb_bucket_name=default,graceful_failover=True,cbas_bucket_name=default_cbas,cbas_dataset_name=default_ds,items=10000,nodeType=KV,rebalance_out=True,concurrent_batch_size=500
test_fail_over_node_followed_by_rebalance_out_or_add_back,cb_bucket_name=default,graceful_failover=True,cbas_bucket_name=default_cbas,cbas_dataset_name=default_ds,items=10000,nodeType=KV,rebalance_out=False,recovery_strategy=full,concurrent_batch_size=500
test_fail_over_node_followed_by_rebalance_out_or_add_back,cb_bucket_name=default,graceful_failover=True,cbas_bucket_name=default_cbas,cbas_dataset_name=default_ds,items=10000,nodeType=KV,rebalance_out=False,recovery_strategy=delta,concurrent_batch_size=500
test_fail_over_node_followed_by_rebalance_out_or_add_back,cb_bucket_name=default,graceful_failover=False,cbas_bucket_name=default_cbas,cbas_dataset_name=default_ds,items=10000,nodeType=KV,rebalance_out=True,concurrent_batch_size=500
test_fail_over_node_followed_by_rebalance_out_or_add_back,cb_bucket_name=default,graceful_failover=False,cbas_bucket_name=default_cbas,cbas_dataset_name=default_ds,items=10000,nodeType=KV,rebalance_out=False,recovery_strategy=full,concurrent_batch_size=500
test_fail_over_node_followed_by_rebalance_out_or_add_back,cb_bucket_name=default,graceful_failover=False,cbas_bucket_name=default_cbas,cbas_dataset_name=default_ds,items=10000,nodeType=KV,rebalance_out=False,recovery_strategy=delta,concurrent_batch_size=500
test_fail_over_node_followed_by_rebalance_out_or_add_back,cb_bucket_name=default,graceful_failover=False,cbas_bucket_name=default_cbas,cbas_dataset_name=default_ds,items=10000,nodeType=CBAS,rebalance_out=True,concurrent_batch_size=500
test_fail_over_node_followed_by_rebalance_out_or_add_back,cb_bucket_name=default,graceful_failover=False,cbas_bucket_name=default_cbas,cbas_dataset_name=default_ds,items=10000,nodeType=CBAS,rebalance_out=False,recovery_strategy=full,concurrent_batch_size=500
'''
def test_fail_over_node_followed_by_rebalance_out_or_add_back(self):
"""
1. Start with an initial setup, having 1 KV and 1 CBAS
2. Add a node that will be failed over - KV/CBAS
3. Create CBAS buckets and dataset
4. Fail over the KV node based in graceful_failover parameter specified
5. Rebalance out/add back based on input param specified in conf file
6. Perform doc operations
7. run concurrent queries
8. Verify document count on dataset post failover
"""
self.log.info("Add an extra node to fail-over")
self.add_node(node=self.rebalanceServers[1])
self.log.info("Read the failure out type to be performed")
graceful_failover = self.input.param("graceful_failover", True)
self.log.info("Set up test - Create cbas buckets and data-sets")
self.setup_for_test()
self.log.info("Perform Async doc operations on KV")
json_generator = JsonGenerator()
generators = json_generator.generate_docs_simple(
docs_per_day=self.num_items * 3 / 2, start=self.num_items)
kv_task = self._async_load_all_buckets(self.master, generators,
"create", 0)
self.log.info("Run concurrent queries on CBAS")
query = "select count(*) from {0};".format(self.cbas_dataset_name)
handles = self.cbas_util._run_concurrent_queries(
query,
"async",
self.num_concurrent_queries,
batch_size=self.concurrent_batch_size)
self.log.info("fail-over the node")
fail_task = self._cb_cluster.async_failover(self.input.servers,
[self.rebalanceServers[1]],
graceful_failover)
fail_task.get_result()
self.log.info(
"Read input param to decide on add back or rebalance out")
self.rebalance_out = self.input.param("rebalance_out", False)
if self.rebalance_out:
self.log.info("Rebalance out the fail-over node")
self.rebalance()
else:
self.recovery_strategy = self.input.param("recovery_strategy",
"full")
self.log.info("Performing %s recovery" % self.recovery_strategy)
success = False
end_time = datetime.datetime.now() + datetime.timedelta(
minutes=int(1))
while datetime.datetime.now() < end_time or not success:
try:
self.sleep(10, message="Wait for fail over complete")
self.rest.set_recovery_type(
'ns_1@' + self.rebalanceServers[1].ip,
self.recovery_strategy)
success = True
except Exception:
self.log.info(
"Fail over in progress. Re-try after 10 seconds.")
pass
if not success:
self.fail("Recovery %s failed." % self.recovery_strategy)
self.rest.add_back_node('ns_1@' + self.rebalanceServers[1].ip)
self.rebalance()
self.log.info("Get KV ops result")
for task in kv_task:
task.get_result()
self.log.info("Log concurrent query status")
self.cbas_util.log_concurrent_query_outcome(self.master, handles)
self.log.info("Validate dataset count on CBAS")
count_n1ql = self.rest.query_tool(
'select count(*) from `%s`' %
self.cb_bucket_name)['results'][0]['$1']
if not self.cbas_util.validate_cbas_dataset_items_count(
self.cbas_dataset_name,
count_n1ql,
0,
timeout=400,
analytics_timeout=400):
self.fail(
"No. of items in CBAS dataset do not match that in the CB bucket"
)
'''
test_to_fail_initial_rebalance_and_verify_subsequent_rebalance_succeeds,cb_bucket_name=default,cbas_bucket_name=default_bucket,cbas_dataset_name=default_ds,items=10,nodeType=CBAS,num_queries=10,restart_couchbase_on_incoming_or_outgoing_node=True,rebalance_type=in
test_to_fail_initial_rebalance_and_verify_subsequent_rebalance_succeeds,cb_bucket_name=default,cbas_bucket_name=default_bucket,cbas_dataset_name=default_ds,items=10,nodeType=CBAS,num_queries=10,restart_couchbase_on_incoming_or_outgoing_node=True,rebalance_type=out
test_to_fail_initial_rebalance_and_verify_subsequent_rebalance_succeeds,cb_bucket_name=default,cbas_bucket_name=default_bucket,cbas_dataset_name=default_ds,items=10,nodeType=CBAS,num_queries=10,restart_couchbase_on_incoming_or_outgoing_node=True,rebalance_type=swap
'''
def test_to_fail_initial_rebalance_and_verify_subsequent_rebalance_succeeds(
self):
self.log.info("Pick the incoming and outgoing nodes during rebalance")
self.rebalance_type = self.input.param("rebalance_type", "in")
nodes_to_add = [self.rebalanceServers[1]]
nodes_to_remove = []
reinitialize_cbas_util = False
if self.rebalance_type == 'out':
nodes_to_remove.append(self.rebalanceServers[1])
self.add_node(self.rebalanceServers[1])
nodes_to_add = []
elif self.rebalance_type == 'swap':
self.add_node(nodes_to_add[0], rebalance=False)
nodes_to_remove.append(self.cbas_node)
reinitialize_cbas_util = True
self.log.info(
"Incoming nodes - %s, outgoing nodes - %s. For rebalance type %s "
% (nodes_to_add, nodes_to_remove, self.rebalance_type))
self.log.info("Creates cbas buckets and dataset")
dataset_count_query = "select count(*) from {0};".format(
self.cbas_dataset_name)
self.setup_for_test()
self.log.info("Perform async doc operations on KV")
json_generator = JsonGenerator()
generators = json_generator.generate_docs_simple(
docs_per_day=self.num_items * 3 / 2, start=self.num_items)
kv_task = self._async_load_all_buckets(self.master,
generators,
"create",
0,
batch_size=5000)
self.log.info("Run concurrent queries on CBAS")
handles = self.cbas_util._run_concurrent_queries(
dataset_count_query, "async", self.num_concurrent_queries)
self.log.info("Fetch the server to restart couchbase on")
restart_couchbase_on_incoming_or_outgoing_node = self.input.param(
"restart_couchbase_on_incoming_or_outgoing_node", True)
if not restart_couchbase_on_incoming_or_outgoing_node:
node = self.cbas_node
else:
node = self.rebalanceServers[1]
shell = RemoteMachineShellConnection(node)
self.log.info("Rebalance nodes")
self.cluster.async_rebalance(self.servers, nodes_to_add,
nodes_to_remove)
self.log.info("Restart Couchbase on node %s" % node.ip)
shell.restart_couchbase()
self.sleep(30, message="Waiting for service to be back again...")
self.log.info("Verify subsequent rebalance is successful")
nodes_to_add = [
] # Node is already added to cluster in previous rebalance, adding it again will throw exception
self.assertTrue(
self.cluster.rebalance(self.servers, nodes_to_add,
nodes_to_remove))
self.log.info("Get KV ops result")
for task in kv_task:
task.get_result()
self.log.info("Log concurrent query status")
self.cbas_util.log_concurrent_query_outcome(self.master, handles)
if reinitialize_cbas_util is True:
self.cbas_util = cbas_utils(self.master, self.rebalanceServers[1])
self.cbas_util.createConn("default")
self.log.info("Validate dataset count on CBAS")
if not self.cbas_util.validate_cbas_dataset_items_count(
self.cbas_dataset_name, self.num_items * 3 / 2, 0):
self.fail(
"No. of items in CBAS dataset do not match that in the CB bucket"
)
'''
test_rebalance_on_nodes_running_multiple_services,cb_bucket_name=default,cbas_bucket_name=default_bucket,cbas_dataset_name=default_ds,items=10,nodeType=KV,num_queries=10,rebalance_type=in
test_rebalance_on_nodes_running_multiple_services,cb_bucket_name=default,cbas_bucket_name=default_bucket,cbas_dataset_name=default_ds,items=10,nodeType=KV,num_queries=10,rebalance_type=out
test_rebalance_on_nodes_running_multiple_services,cb_bucket_name=default,cbas_bucket_name=default_bucket,cbas_dataset_name=default_ds,items=10,num_queries=10,rebalance_type=swap,rebalance_cbas_and_kv=True
'''
def test_rebalance_on_nodes_running_multiple_services(self):
self.log.info("Pick the incoming and outgoing nodes during rebalance")
active_services = ['cbas,fts,kv']
self.rebalance_type = self.input.param("rebalance_type", "in")
nodes_to_add = [self.rebalanceServers[1]]
nodes_to_remove = []
if self.rebalance_type == 'out':
# This node will be rebalanced out
nodes_to_remove.append(self.rebalanceServers[1])
# Will be running services as specified in the list - active_services
self.add_node(nodes_to_add[0], services=active_services)
# No nodes to remove so making the add notes empty
nodes_to_add = []
elif self.rebalance_type == 'swap':
# Below node will be swapped with the incoming node specified in nodes_to_add
self.add_node(nodes_to_add[0], services=active_services)
nodes_to_add = []
nodes_to_add.append(self.rebalanceServers[3])
# Below node will be removed and swapped with node that was added earlier
nodes_to_remove.append(self.rebalanceServers[1])
self.log.info(
"Incoming nodes - %s, outgoing nodes - %s. For rebalance type %s "
% (nodes_to_add, nodes_to_remove, self.rebalance_type))
self.log.info("Creates cbas buckets and dataset")
dataset_count_query = "select count(*) from {0};".format(
self.cbas_dataset_name)
self.setup_for_test()
self.log.info("Perform async doc operations on KV")
json_generator = JsonGenerator()
generators = json_generator.generate_docs_simple(
docs_per_day=self.num_items * 3 / 2, start=self.num_items)
kv_task = self._async_load_all_buckets(self.master,
generators,
"create",
0,
batch_size=5000)
self.log.info("Run concurrent queries on CBAS")
handles = self.cbas_util._run_concurrent_queries(
dataset_count_query, "async", self.num_concurrent_queries)
self.log.info("Rebalance nodes")
# Do not add node to nodes_to_add if already added as add_node earlier
self.cluster.rebalance(self.servers,
nodes_to_add,
nodes_to_remove,
services=active_services)
self.log.info("Get KV ops result")
for task in kv_task:
task.get_result()
self.log.info("Log concurrent query status")
self.cbas_util.log_concurrent_query_outcome(self.master, handles)
self.log.info("Validate dataset count on CBAS")
if not self.cbas_util.validate_cbas_dataset_items_count(
self.cbas_dataset_name, self.num_items * 3 / 2, 0):
self.fail(
"No. of items in CBAS dataset do not match that in the CB bucket"
)
def tearDown(self):
super(CBASClusterOperations, self).tearDown()
class BaseTestCase(unittest.TestCase):
def setUp(self):
self.input = TestInputSingleton.input
# Framework specific parameters
self.log_level = self.input.param("log_level", "info").upper()
self.infra_log_level = self.input.param("infra_log_level",
"error").upper()
self.skip_setup_cleanup = self.input.param("skip_setup_cleanup", False)
self.tear_down_while_setup = self.input.param("tear_down_while_setup",
True)
self.test_timeout = self.input.param("test_timeout", 3600)
self.thread_to_use = self.input.param("threads_to_use", 30)
self.case_number = self.input.param("case_number", 0)
# End of framework parameters
# Cluster level info settings
self.log_info = self.input.param("log_info", None)
self.log_location = self.input.param("log_location", None)
self.stat_info = self.input.param("stat_info", None)
self.port = self.input.param("port", None)
self.port_info = self.input.param("port_info", None)
self.servers = self.input.servers
self.cb_clusters = OrderedDict()
self.num_servers = self.input.param("servers", len(self.servers))
self.primary_index_created = False
self.index_quota_percent = self.input.param("index_quota_percent",
None)
self.gsi_type = self.input.param("gsi_type", 'plasma')
# CBAS setting
self.jre_path = self.input.param("jre_path", None)
self.enable_dp = self.input.param("enable_dp", False)
# End of cluster info parameters
# Bucket specific params
self.bucket_type = self.input.param("bucket_type", Bucket.Type.MEMBASE)
self.bucket_ttl = self.input.param("bucket_ttl", 0)
self.bucket_size = self.input.param("bucket_size", None)
self.bucket_conflict_resolution_type = \
self.input.param("bucket_conflict_resolution",
Bucket.ConflictResolution.SEQ_NO)
self.bucket_replica_index = self.input.param("bucket_replica_index", 1)
self.bucket_eviction_policy = \
self.input.param("bucket_eviction_policy",
Bucket.EvictionPolicy.VALUE_ONLY)
self.flush_enabled = self.input.param("flushEnabled",
Bucket.FlushBucket.DISABLED)
self.bucket_time_sync = self.input.param("bucket_time_sync", False)
self.standard_buckets = self.input.param("standard_buckets", 1)
self.num_replicas = self.input.param("replicas", Bucket.ReplicaNum.ONE)
self.active_resident_threshold = \
int(self.input.param("active_resident_threshold", 100))
self.compression_mode = \
self.input.param("compression_mode",
Bucket.CompressionMode.PASSIVE)
self.bucket_storage = \
self.input.param("bucket_storage",
Bucket.StorageBackend.couchstore)
if self.bucket_storage == Bucket.StorageBackend.magma:
self.bucket_eviction_policy = Bucket.EvictionPolicy.FULL_EVICTION
self.scope_name = self.input.param("scope", CbServer.default_scope)
self.collection_name = self.input.param("collection",
CbServer.default_collection)
self.bucket_durability_level = self.input.param(
"bucket_durability", Bucket.DurabilityLevel.NONE).upper()
self.bucket_purge_interval = self.input.param("bucket_purge_interval",
1)
self.bucket_durability_level = \
BucketDurability[self.bucket_durability_level]
# End of bucket parameters
# Doc specific params
self.key = self.input.param("key", "test_docs")
self.key_size = self.input.param("key_size", 8)
self.doc_size = self.input.param("doc_size", 256)
self.sub_doc_size = self.input.param("sub_doc_size", 10)
self.doc_type = self.input.param("doc_type", "json")
self.num_items = self.input.param("num_items", 100000)
self.target_vbucket = self.input.param("target_vbucket", None)
self.maxttl = self.input.param("maxttl", 0)
self.random_exp = self.input.param("random_exp", False)
self.randomize_doc_size = self.input.param("randomize_doc_size", False)
self.randomize_value = self.input.param("randomize_value", False)
self.rev_write = self.input.param("rev_write", False)
self.rev_read = self.input.param("rev_read", False)
self.rev_update = self.input.param("rev_update", False)
self.rev_del = self.input.param("rev_del", False)
self.random_key = self.input.param("random_key", False)
self.mix_key_size = self.input.param("mix_key_size", False)
# End of doc specific parameters
# Transactions parameters
self.transaction_timeout = self.input.param("transaction_timeout", 100)
self.transaction_commit = self.input.param("transaction_commit", True)
self.update_count = self.input.param("update_count", 1)
self.sync = self.input.param("sync", True)
self.default_bucket = self.input.param("default_bucket", True)
self.num_buckets = self.input.param("num_buckets", 0)
self.atomicity = self.input.param("atomicity", False)
self.defer = self.input.param("defer", False)
# end of transaction parameters
# Client specific params
self.sdk_client_type = self.input.param("sdk_client_type", "java")
self.replicate_to = self.input.param("replicate_to", 0)
self.persist_to = self.input.param("persist_to", 0)
self.sdk_retries = self.input.param("sdk_retries", 5)
self.sdk_timeout = self.input.param("sdk_timeout", 5)
self.time_unit = self.input.param("time_unit", "seconds")
self.durability_level = self.input.param("durability", "").upper()
self.sdk_client_pool = self.input.param("sdk_client_pool", None)
self.sdk_pool_capacity = self.input.param("sdk_pool_capacity", 1)
# Client compression settings
self.sdk_compression = self.input.param("sdk_compression", None)
compression_min_ratio = self.input.param("min_ratio", None)
compression_min_size = self.input.param("min_size", None)
if type(self.sdk_compression) is bool:
self.sdk_compression = {"enabled": self.sdk_compression}
if compression_min_size:
self.sdk_compression["minSize"] = compression_min_size
if compression_min_ratio:
self.sdk_compression["minRatio"] = compression_min_ratio
# Doc Loader Params
self.process_concurrency = self.input.param("process_concurrency", 20)
self.batch_size = self.input.param("batch_size", 2000)
self.dgm_batch = self.input.param("dgm_batch", 5000)
self.ryow = self.input.param("ryow", False)
self.check_persistence = self.input.param("check_persistence", False)
# End of client specific parameters
# initial number of items in the cluster
self.services_init = self.input.param("services_init", None)
self.nodes_init = self.input.param("nodes_init", 1)
self.nodes_in = self.input.param("nodes_in", 1)
self.nodes_out = self.input.param("nodes_out", 1)
self.services_in = self.input.param("services_in", None)
self.forceEject = self.input.param("forceEject", False)
self.wait_timeout = self.input.param("wait_timeout", 120)
self.verify_unacked_bytes = \
self.input.param("verify_unacked_bytes", False)
self.disabled_consistent_view = \
self.input.param("disabled_consistent_view", None)
self.rebalanceIndexWaitingDisabled = \
self.input.param("rebalanceIndexWaitingDisabled", None)
self.rebalanceIndexPausingDisabled = \
self.input.param("rebalanceIndexPausingDisabled", None)
self.maxParallelIndexers = \
self.input.param("maxParallelIndexers", None)
self.maxParallelReplicaIndexers = \
self.input.param("maxParallelReplicaIndexers", None)
self.quota_percent = self.input.param("quota_percent", 90)
self.skip_buckets_handle = self.input.param("skip_buckets_handle",
False)
# SDKClientPool object for creating generic clients across tasks
if self.sdk_client_pool is True:
self.init_sdk_pool_object()
# Initiate logging variables
self.log = logger.get("test")
self.infra_log = logger.get("infra")
self.cleanup_pcaps()
self.collect_pcaps = self.input.param("collect_pcaps", False)
if self.collect_pcaps:
self.start_collect_pcaps()
# variable for log collection using cbCollect
self.get_cbcollect_info = self.input.param("get-cbcollect-info", False)
# Variable for initializing the current (start of test) timestamp
self.start_timestamp = datetime.now()
'''
Be careful while using this flag.
This is only and only for stand-alone tests.
During bugs reproductions, when a crash is seen
stop_server_on_crash will stop the server
so that we can collect data/logs/dumps at the right time
'''
self.stop_server_on_crash = self.input.param("stop_server_on_crash",
False)
self.collect_data = self.input.param("collect_data", False)
# Configure loggers
self.log.setLevel(self.log_level)
self.infra_log.setLevel(self.infra_log_level)
# Support lib objects for testcase execution
self.task_manager = TaskManager(self.thread_to_use)
self.task = ServerTasks(self.task_manager)
# End of library object creation
self.sleep = sleep
self.cleanup = False
self.nonroot = False
self.test_failure = None
self.crash_warning = self.input.param("crash_warning", False)
self.summary = TestSummary(self.log)
# Populate memcached_port in case of cluster_run
cluster_run_base_port = ClusterRun.port
if int(self.input.servers[0].port) == ClusterRun.port:
for server in self.input.servers:
server.port = cluster_run_base_port
cluster_run_base_port += 1
# If not defined in node.ini under 'memcached_port' section
if server.memcached_port is CbServer.memcached_port:
server.memcached_port = \
ClusterRun.memcached_port \
+ (2 * (int(server.port) - ClusterRun.port))
self.log_setup_status(self.__class__.__name__, "started")
cluster_name_format = "C%s"
default_cluster_index = counter_index = 1
if len(self.input.clusters) > 1:
# Multi cluster setup
for _, nodes in self.input.clusters.iteritems():
cluster_name = cluster_name_format % counter_index
tem_cluster = CBCluster(name=cluster_name, servers=nodes)
self.cb_clusters[cluster_name] = tem_cluster
counter_index += 1
else:
# Single cluster
cluster_name = cluster_name_format % counter_index
self.cb_clusters[cluster_name] = CBCluster(name=cluster_name,
servers=self.servers)
# Initialize self.cluster with first available cluster as default
self.cluster = self.cb_clusters[cluster_name_format %
default_cluster_index]
self.cluster_util = ClusterUtils(self.cluster, self.task_manager)
self.bucket_util = BucketUtils(self.cluster_util, self.task)
if self.standard_buckets > 10:
self.bucket_util.change_max_buckets(self.cluster.master,
self.standard_buckets)
for cluster_name, cluster in self.cb_clusters.items():
shell = RemoteMachineShellConnection(cluster.master)
self.os_info = shell.extract_remote_info().type.lower()
if self.os_info != 'windows':
if cluster.master.ssh_username != "root":
self.nonroot = True
shell.disconnect()
break
shell.disconnect()
""" some tests need to bypass checking cb server at set up
to run installation """
self.skip_init_check_cbserver = \
self.input.param("skip_init_check_cbserver", False)
try:
if self.skip_setup_cleanup:
self.cluster.buckets = self.bucket_util.get_all_buckets(
self.cluster)
return
self.services_map = None
self.log_setup_status("BaseTestCase", "started")
for cluster_name, cluster in self.cb_clusters.items():
if not self.skip_buckets_handle \
and not self.skip_init_check_cbserver:
self.log.debug("Cleaning up cluster")
cluster_util = ClusterUtils(cluster, self.task_manager)
bucket_util = BucketUtils(cluster_util, self.task)
cluster_util.cluster_cleanup(bucket_util)
# Avoid cluster operations in setup for new upgrade / upgradeXDCR
if str(self.__class__).find('newupgradetests') != -1 or \
str(self.__class__).find('upgradeXDCR') != -1 or \
str(self.__class__).find('Upgrade_EpTests') != -1 or \
self.skip_buckets_handle:
self.log.warning("Cluster operation in setup will be skipped")
self.primary_index_created = True
self.log_setup_status("BaseTestCase", "finished")
return
# avoid clean up if the previous test has been tear down
if self.case_number == 1 or self.case_number > 1000:
if self.case_number > 1000:
self.log.warn("TearDown for prev test failed. Will retry")
self.case_number -= 1000
self.cleanup = True
if not self.skip_init_check_cbserver:
self.tearDownEverything()
self.tear_down_while_setup = False
if not self.skip_init_check_cbserver:
for cluster_name, cluster in self.cb_clusters.items():
self.log.info("Initializing cluster")
cluster_util = ClusterUtils(cluster, self.task_manager)
cluster_util.reset_cluster()
master_services = cluster_util.get_services(
cluster.servers[:1], self.services_init, start_node=0)
if master_services is not None:
master_services = master_services[0].split(",")
self.quota = self._initialize_nodes(
self.task,
cluster,
self.disabled_consistent_view,
self.rebalanceIndexWaitingDisabled,
self.rebalanceIndexPausingDisabled,
self.maxParallelIndexers,
self.maxParallelReplicaIndexers,
self.port,
self.quota_percent,
services=master_services)
cluster_util.change_env_variables()
cluster_util.change_checkpoint_params()
self.log.info("{0} initialized".format(cluster))
else:
self.quota = ""
# Enable dp_version since we need collections enabled
if self.enable_dp:
for server in self.cluster.servers:
shell_conn = RemoteMachineShellConnection(server)
cb_cli = CbCli(shell_conn)
cb_cli.enable_dp()
shell_conn.disconnect()
for cluster_name, cluster in self.cb_clusters.items():
cluster_util = ClusterUtils(cluster, self.task_manager)
if self.log_info:
cluster_util.change_log_info()
if self.log_location:
cluster_util.change_log_location()
if self.stat_info:
cluster_util.change_stat_info()
if self.port_info:
cluster_util.change_port_info()
if self.port:
self.port = str(self.port)
self.log_setup_status("BaseTestCase", "finished")
if not self.skip_init_check_cbserver:
self.__log("started")
except Exception as e:
traceback.print_exc()
self.task.shutdown(force=True)
self.fail(e)
def cleanup_pcaps(self):
for server in self.servers:
shell = RemoteMachineShellConnection(server)
# Stop old instances of tcpdump if still running
stop_tcp_cmd = "if [[ \"$(pgrep tcpdump)\" ]]; " \
"then kill -s TERM $(pgrep tcpdump); fi"
_, _ = shell.execute_command(stop_tcp_cmd)
shell.execute_command("rm -rf pcaps")
shell.execute_command("rm -rf " + server.ip + "_pcaps.zip")
shell.disconnect()
def start_collect_pcaps(self):
for server in self.servers:
shell = RemoteMachineShellConnection(server)
# Create path for storing pcaps
create_path = "mkdir -p pcaps"
o, e = shell.execute_command(create_path)
shell.log_command_output(o, e)
# Install tcpdump command if it doesn't exist
o, e = shell.execute_command("yum install -y tcpdump")
shell.log_command_output(o, e)
# Install screen command if it doesn't exist
o, e = shell.execute_command("yum install -y screen")
shell.log_command_output(o, e)
# Execute the tcpdump command
tcp_cmd = "screen -dmS test bash -c \"tcpdump -C 500 -W 10 " \
"-w pcaps/pack-dump-file.pcap -i eth0 -s 0 tcp\""
o, e = shell.execute_command(tcp_cmd)
shell.log_command_output(o, e)
shell.disconnect()
def start_fetch_pcaps(self):
log_path = TestInputSingleton.input.param("logs_folder", "/tmp")
for server in self.servers:
remote_client = RemoteMachineShellConnection(server)
# stop tcdump
stop_tcp_cmd = "if [[ \"$(pgrep tcpdump)\" ]]; " \
"then kill -s TERM $(pgrep tcpdump); fi"
o, e = remote_client.execute_command(stop_tcp_cmd)
remote_client.log_command_output(o, e)
if self.is_test_failed():
# install zip unzip
o, e = remote_client.execute_command(
"yum install -y zip unzip")
remote_client.log_command_output(o, e)
# zip the pcaps folder
zip_cmd = "zip -r " + server.ip + "_pcaps.zip pcaps"
o, e = remote_client.execute_command(zip_cmd)
remote_client.log_command_output(o, e)
# transfer the zip file
zip_file_copied = remote_client.get_file(
"/root", os.path.basename(server.ip + "_pcaps.zip"),
log_path)
self.log.info("%s node pcap zip copied on client : %s" %
(server.ip, zip_file_copied))
if zip_file_copied:
# Remove the zips
remote_client.execute_command("rm -rf " + server.ip +
"_pcaps.zip")
# Remove pcaps
remote_client.execute_command("rm -rf pcaps")
remote_client.disconnect()
def tearDown(self):
self.task_manager.shutdown_task_manager()
self.task.shutdown(force=True)
self.task_manager.abort_all_tasks()
if self.sdk_client_pool:
self.sdk_client_pool.shutdown()
if self.collect_pcaps:
self.log.info("Starting Pcaps collection!!")
self.start_fetch_pcaps()
result = self.check_coredump_exist(self.servers, force_collect=True)
self.tearDownEverything()
if not self.crash_warning:
self.assertFalse(result, msg="Cb_log file validation failed")
if self.crash_warning and result:
self.log.warn("CRASH | CRITICAL | WARN messages found in cb_logs")
def tearDownEverything(self):
if self.skip_setup_cleanup:
return
for _, cluster in self.cb_clusters.items():
cluster_util = ClusterUtils(cluster, self.task_manager)
bucket_util = BucketUtils(cluster_util, self.task)
try:
if self.skip_buckets_handle:
return
test_failed = self.is_test_failed()
if test_failed \
and TestInputSingleton.input.param("stop-on-failure",
False) \
or self.input.param("skip_cleanup", False):
self.log.warn("CLEANUP WAS SKIPPED")
else:
if test_failed:
# Collect logs because we have not shut things down
if self.get_cbcollect_info:
self.fetch_cb_collect_logs()
get_trace = \
TestInputSingleton.input.param("get_trace", None)
if get_trace:
for server in cluster.servers:
shell = \
RemoteMachineShellConnection(server)
output, _ = shell.execute_command(
"ps -aef|grep %s" % get_trace)
output = shell.execute_command(
"pstack %s" % output[0].split()[1].strip())
self.infra_log.debug(output[0])
shell.disconnect()
else:
self.log.critical("Skipping get_trace !!")
rest = RestConnection(cluster.master)
alerts = rest.get_alerts()
if alerts is not None and len(alerts) != 0:
self.infra_log.warn("Alerts found: {0}".format(alerts))
self.log.debug("Cleaning up cluster")
cluster_util.cluster_cleanup(bucket_util)
except BaseException as e:
# kill memcached
traceback.print_exc()
self.log.warning("Killing memcached due to {0}".format(e))
cluster_util.kill_memcached()
# Increase case_number to retry tearDown in setup for next test
self.case_number += 1000
finally:
# stop all existing task manager threads
if self.cleanup:
self.cleanup = False
else:
cluster_util.reset_env_variables()
self.infra_log.info("========== tasks in thread pool ==========")
self.task_manager.print_tasks_in_pool()
self.infra_log.info("==========================================")
if not self.tear_down_while_setup:
self.task_manager.shutdown_task_manager()
self.task.shutdown(force=True)
def is_test_failed(self):
return (hasattr(self, '_resultForDoCleanups')
and len(self._resultForDoCleanups.failures
or self._resultForDoCleanups.errors)) \
or (hasattr(self, '_exc_info')
and self._exc_info()[1] is not None)
def handle_setup_exception(self, exception_obj):
# Shutdown client pool in case of any error before failing
if self.sdk_client_pool is not None:
self.sdk_client_pool.shutdown()
# print the tracback of the failure
traceback.print_exc()
# Throw the exception so that the test will fail at setUp
raise exception_obj
def __log(self, status):
try:
msg = "{0}: {1} {2}" \
.format(datetime.now(), self._testMethodName, status)
RestConnection(self.servers[0]).log_client_error(msg)
except Exception as e:
self.log.warning("Exception during REST log_client_error: %s" % e)
def log_setup_status(self, class_name, status, stage="setup"):
self.log.info("========= %s %s %s for test #%d %s =========" %
(class_name, stage, status, self.case_number,
self._testMethodName))
def _initialize_nodes(self,
task,
cluster,
disabled_consistent_view=None,
rebalance_index_waiting_disabled=None,
rebalance_index_pausing_disabled=None,
max_parallel_indexers=None,
max_parallel_replica_indexers=None,
port=None,
quota_percent=None,
services=None):
quota = 0
init_tasks = []
ssh_sessions = dict()
# Open ssh_connections for command execution
for server in cluster.servers:
ssh_sessions[server.ip] = RemoteMachineShellConnection(server)
for server in cluster.servers:
# Make sure that data_and index_path are writable by couchbase user
if not server.index_path:
server.index_path = server.data_path
for path in set(
[_f for _f in [server.data_path, server.index_path] if _f]):
for cmd in ("rm -rf {0}/*".format(path),
"chown -R couchbase:couchbase {0}".format(path)):
ssh_sessions[server.ip].execute_command(cmd)
rest = RestConnection(server)
rest.set_data_path(data_path=server.data_path,
index_path=server.index_path,
cbas_path=server.cbas_path)
init_port = port or server.port or '8091'
assigned_services = services
if cluster.master != server:
assigned_services = None
init_tasks.append(
task.async_init_node(
server,
disabled_consistent_view,
rebalance_index_waiting_disabled,
rebalance_index_pausing_disabled,
max_parallel_indexers,
max_parallel_replica_indexers,
init_port,
quota_percent,
services=assigned_services,
index_quota_percent=self.index_quota_percent,
gsi_type=self.gsi_type))
for _task in init_tasks:
node_quota = self.task_manager.get_task_result(_task)
if node_quota < quota or quota == 0:
quota = node_quota
if quota < 100 and not len(set([server.ip
for server in self.servers])) == 1:
self.log.warn("RAM quota was defined less than 100 MB:")
for server in cluster.servers:
ram = ssh_sessions[server.ip].extract_remote_info().ram
self.log.debug("Node: {0}: RAM: {1}".format(server.ip, ram))
# Close all ssh_connections
for server in cluster.servers:
ssh_sessions[server.ip].disconnect()
if self.jre_path:
for server in cluster.servers:
rest = RestConnection(server)
rest.set_jre_path(self.jre_path)
return quota
def fetch_cb_collect_logs(self):
log_path = TestInputSingleton.input.param("logs_folder", "/tmp")
is_single_node_server = len(self.servers) == 1
for _, cluster in self.cb_clusters.items():
rest = RestConnection(cluster.master)
nodes = rest.get_nodes()
# Creating cluster_util object to handle multi_cluster scenario
cluster_util = ClusterUtils(cluster, self.task_manager)
status = cluster_util.trigger_cb_collect_on_cluster(
rest, nodes, is_single_node_server)
if status is True:
cluster_util.wait_for_cb_collect_to_complete(rest)
cluster_util.copy_cb_collect_logs(rest, nodes, cluster,
log_path)
else:
self.log.error("API perform_cb_collect returned False")
def log_failure(self, message):
self.log.error(message)
self.summary.set_status("FAILED")
if self.test_failure is None:
self.test_failure = message
def validate_test_failure(self):
if self.test_failure is not None:
self.fail(self.test_failure)
def get_clusters(self):
return [self.cb_clusters[name] for name in self.cb_clusters.keys()]
def get_task(self):
return self.task
def get_task_mgr(self):
return self.task_manager
def init_sdk_pool_object(self):
self.sdk_client_pool = SDKClientPool()
DocLoaderUtils.sdk_client_pool = self.sdk_client_pool
def check_coredump_exist(self, servers, force_collect=False):
bin_cb = "/opt/couchbase/bin/"
lib_cb = "/opt/couchbase/var/lib/couchbase/"
# crash_dir = "/opt/couchbase/var/lib/couchbase/"
crash_dir_win = "c://CrashDumps"
result = False
self.data_sets = dict()
def find_index_of(str_list, sub_string):
for i in range(len(str_list)):
if sub_string in str_list[i]:
return i
return -1
def get_gdb(gdb_shell, dmp_path, dmp_name):
dmp_file = dmp_path + dmp_name
core_file = dmp_path + dmp_name.strip(".dmp") + ".core"
gdb_shell.execute_command("rm -rf " + core_file)
gdb_shell.execute_command("/" + bin_cb + "minidump-2-core " +
dmp_file + " > " + core_file)
cmd = "gdb --batch {} -c {} -ex \"bt full\" -ex quit"\
.format(os.path.join(bin_cb, "memcached"), core_file)
self.log.info("running %s" % cmd)
gdb_out = gdb_shell.execute_command(cmd)[0]
t_index = find_index_of(gdb_out, "Core was generated by")
gdb_out = gdb_out[t_index:]
gdb_out = " ".join(gdb_out)
return gdb_out
def get_full_thread_dump(gdb_shell):
cmd = 'gdb -p `(pidof memcached)` -ex "thread apply all bt" -ex detach -ex quit'
self.log.info("running %s" % cmd)
thread_dump = gdb_shell.execute_command(cmd)[0]
index = find_index_of(
thread_dump, "Thread debugging using libthread_db enabled")
print(" ".join(thread_dump[index:]))
def check_if_new_messages(grep_output_list):
"""
Check the grep's last line for the latest timestamp.
If this timestamp < start_timestamp of the test,
then return False (as the grep's output is from previous tests)
Note: This method works only if slave's time(timezone) matches
that of VM's. Else it won't be possible to compare timestamps
"""
last_line = grep_output_list[-1]
if not re.match(r"[0-9]{4}-[0-9]{2}-[0-9]{2}T", last_line):
# To check if line doesn't begin with any yyyy-mm-ddT
self.log.critical("%s does not match any timestamp" %
last_line)
return True
timestamp = last_line.split()[0]
timestamp = timestamp.split(".")[0]
timestamp = datetime.strptime(timestamp, "%Y-%m-%dT%H:%M:%S")
self.log.info("Comparing timestamps: Log's latest timestamp: %s, "
"Test's start timestamp is %s" %
(timestamp, self.start_timestamp))
if timestamp > self.start_timestamp:
return True
else:
return False
for idx, server in enumerate(servers):
shell = RemoteMachineShellConnection(server)
shell.extract_remote_info()
crash_dir = lib_cb + "crash/"
if shell.info.type.lower() == "windows":
crash_dir = crash_dir_win
if int(server.port) in range(ClusterRun.port,
ClusterRun.port + 10):
crash_dir = os.path.join(
TestInputSingleton.input.servers[0].cli_path, "ns_server",
"data", "n_%s" % str(idx), "crash")
dmp_files = shell.execute_command("ls -lt " + crash_dir)[0]
dmp_files = [f for f in dmp_files if ".core" not in f]
dmp_files = [f for f in dmp_files if "total" not in f]
dmp_files = [f.split()[-1] for f in dmp_files if ".core" not in f]
dmp_files = [f.strip("\n") for f in dmp_files]
if dmp_files:
msg = "%s: %d core dump seen" % (server.ip, len(dmp_files))
self.log.critical(msg)
self.log.critical("%s: Stack Trace of first crash - %s\n%s" %
(server.ip, dmp_files[-1],
get_gdb(shell, crash_dir, dmp_files[-1])))
get_full_thread_dump(shell)
if self.stop_server_on_crash:
shell.stop_couchbase()
result = True
else:
self.log.debug(server.ip + ": No crash files found")
logs_dir = lib_cb + "logs/"
if int(server.port) in range(ClusterRun.port,
ClusterRun.port + 10):
logs_dir = os.path.join(
TestInputSingleton.input.servers[0].cli_path, "ns_server",
"logs", "n_%s" % str(idx))
# Perform log file searching based on the input yaml config
yaml = YAML()
with open("lib/couchbase_helper/error_log_config.yaml", "r") as fp:
y_data = yaml.load(fp.read())
for file_data in y_data["file_name_patterns"]:
log_files = shell.execute_command(
"ls " + os.path.join(logs_dir, file_data['file']))[0]
if len(log_files) == 0:
self.log.debug("%s: No '%s' files found" %
(server.ip, file_data['file']))
continue
if 'target_file_index' in file_data:
log_files = [
log_files[int(file_data['target_file_index'])]
]
for log_file in log_files:
log_file = log_file.strip("\n")
for grep_pattern in file_data['grep_for']:
grep_for_str = grep_pattern['string']
err_pattern = exclude_pattern = None
if 'error_patterns' in grep_pattern:
err_pattern = grep_pattern['error_patterns']
if 'exclude_patterns' in grep_pattern:
exclude_pattern = grep_pattern['exclude_patterns']
cmd_to_run = "grep -r '%s' %s" \
% (grep_for_str, log_file)
if exclude_pattern is not None:
for pattern in exclude_pattern:
cmd_to_run += " | grep -v '%s'" % pattern
grep_output = shell.execute_command(cmd_to_run)[0]
if grep_output and check_if_new_messages(grep_output):
regex = r"(\bkvstore-\d+)"
grep_str = "".join(grep_output)
kvstores = list(set(re.findall(regex, grep_str)))
self.data_sets[server] = kvstores
grep_str = None
if err_pattern is not None:
for pattern in err_pattern:
index = find_index_of(grep_output, pattern)
grep_output = grep_output[:index]
if grep_output:
self.log.info("unwanted messages in %s" %
log_file)
if check_if_new_messages(grep_output):
self.log.critical(
"%s: Found '%s' logs - %s" %
(server.ip, grep_for_str,
"".join(grep_output)))
result = True
break
else:
if grep_output \
and check_if_new_messages(grep_output):
self.log.info("unwanted messages in %s" %
log_file)
self.log.critical(
"%s: Found '%s' logs - %s" %
(server.ip, grep_for_str, grep_output))
result = True
break
if result is True:
if self.stop_server_on_crash:
shell.stop_couchbase()
break
shell.disconnect()
if result and force_collect and not self.stop_server_on_crash:
self.fetch_cb_collect_logs()
self.get_cbcollect_info = False
if (self.is_test_failed() or result) and self.collect_data:
self.copy_data_on_slave()
return result
def copy_data_on_slave(self, servers=None):
log_path = TestInputSingleton.input.param("logs_folder", "/tmp")
if servers is None:
servers = self.cluster.nodes_in_cluster
for node in servers:
if "kv" not in node.services.lower():
servers.remove(node)
if type(servers) is not list:
servers = [servers]
remote_path = RestConnection(servers[0]).get_data_path()
file_path = os.path.join(remote_path, self.cluster.buckets[0].name)
file_name = self.cluster.buckets[0].name + ".tar.gz"
def get_tar(remotepath, filepath, filename, servers, todir="."):
if type(servers) is not list:
servers = [servers]
for server in servers:
shell = RemoteMachineShellConnection(server)
_ = shell.execute_command("tar -zcvf %s.tar.gz %s" %
(filepath, filepath))
file_check = shell.file_exists(remotepath, filename)
if not file_check:
self.log.error(
"Tar File {} doesn't exist".format(filename))
tar_file_copied = shell.get_file(remotepath, filename, todir)
if not tar_file_copied:
self.log.error("Failed to copy Tar file")
_ = shell.execute_command("rm -rf %s.tar.gz" % filepath)
copy_path_msg_format = "Copying data, Server :: %s, Path :: %s"
'''
Temporarily enabling data copy
of all nodes irrespective of nodes in
data_sets
'''
if False and self.data_sets and self.bucket_storage == "magma":
self.log.critical("data_sets ==> {}".format(self.data_sets))
wal_tar = "wal.tar.gz"
config_json_tar = "config.json.tar.gz"
for server, kvstores in self.data_sets.items():
shell = RemoteMachineShellConnection(server)
if not kvstores:
copy_to_path = os.path.join(log_path,
server.ip.replace(".", "_"))
if not os.path.isdir(copy_to_path):
os.makedirs(copy_to_path, 0o777)
self.log.info(copy_path_msg_format %
(server.ip, copy_to_path))
get_tar(remote_path,
file_path,
file_name,
server,
todir=copy_to_path)
else:
for kvstore in kvstores:
if int(kvstore.split("-")[1]) >= self.vbuckets:
continue
kvstore_path = shell.execute_command(
"find %s -type d -name '%s'" %
(remote_path, kvstore))[0][0]
magma_dir = kvstore_path.split(kvstore)[0]
wal_path = kvstore_path.split(kvstore)[0] + "wal"
config_json_path = kvstore_path.split(kvstore)[0]
+ "config.json"
kvstore_path = kvstore_path.split(kvstore)[0] + kvstore
kvstore_tar = kvstore + ".tar.gz"
copy_to_path = os.path.join(log_path, kvstore)
if not os.path.isdir(copy_to_path):
os.makedirs(copy_to_path, 0o777)
self.log.info(copy_path_msg_format %
(server.ip, copy_to_path))
get_tar(magma_dir,
kvstore_path,
kvstore_tar,
server,
todir=copy_to_path)
get_tar(magma_dir,
wal_path,
wal_tar,
server,
todir=copy_to_path)
get_tar(magma_dir,
config_json_path,
config_json_tar,
server,
todir=copy_to_path)
else:
for server in servers:
copy_to_path = os.path.join(log_path,
server.ip.replace(".", "_"))
if not os.path.isdir(copy_to_path):
os.makedirs(copy_to_path, 0o777)
self.log.info(copy_path_msg_format % (server.ip, copy_to_path))
get_tar(remote_path,
file_path,
file_name,
server,
todir=copy_to_path)