def use_and_generate(host, port, bucket_name, frequency_width, volume, add_shutdown):
boto_data = get_aws_credentials('aws-chiles02')
if boto_data is not None:
connection = httplib.HTTPConnection(host, port)
connection.request('GET', '/api', None, {})
response = connection.getresponse()
if response.status != httplib.OK:
msg = 'Error while processing GET request for {0}:{1}/api (status {2}): {3}'.format(host, port, response.status, response.read())
raise Exception(msg)
json_data = response.read()
message_details = json.loads(json_data)
host_list = message_details['hosts']
nodes_running = get_nodes_running(host_list)
if len(nodes_running) > 0:
work_to_do = WorkToDo(
width=frequency_width,
bucket_name=bucket_name,
s3_split_name=get_s3_split_name(frequency_width))
work_to_do.calculate_work_to_do()
# Now build the graph
session_id = get_session_id()
graph = BuildGraphMsTransform(
work_to_do=work_to_do.work_to_do,
bucket_name=bucket_name,
volume=volume,
parallel_streams=7,
node_details=nodes_running,
shutdown=add_shutdown,
width=frequency_width,
session_id=session_id,
dim_ip=host,
)
graph.build_graph()
LOG.info('Connection to {0}:{1}'.format(host, port))
client = DataIslandManagerClient(host, port)
client.create_session(session_id)
client.append_graph(session_id, graph.drop_list)
client.deploy_session(session_id, get_roots(graph.drop_list))
else:
LOG.warning('No nodes are running')
def build_json(bucket, width, volume, nodes, parallel_streams, shutdown):
work_to_do = WorkToDo(width, bucket, get_s3_split_name(width))
work_to_do.calculate_work_to_do()
node_details = {
'i2.2xlarge': [{'ip_address': 'node_i2_{0}'.format(i)} for i in range(0, nodes)],
'i2.4xlarge': [{'ip_address': 'node_i4_{0}'.format(i)} for i in range(0, nodes)],
}
graph = BuildGraphMsTransform(
work_to_do.work_to_do,
bucket,
volume,
parallel_streams,
node_details,
shutdown,
width,
'json_test',
'1.2.3.4'
)
graph.build_graph()
json_dumps = json.dumps(graph.drop_list, indent=2)
LOG.info(json_dumps)
with open("/tmp/json_mstransform.txt", "w") as json_file:
json_file.write(json_dumps)
def create_and_generate(bucket_name, frequency_width, ami_id, spot_price1, spot_price2, volume, days_per_node, add_shutdown):
boto_data = get_aws_credentials('aws-chiles02')
if boto_data is not None:
work_to_do = WorkToDo(frequency_width, bucket_name, get_s3_split_name(frequency_width))
work_to_do.calculate_work_to_do()
days = work_to_do.work_to_do.keys()
nodes_required, node_count = get_nodes_required(days, days_per_node, spot_price1, spot_price2)
if len(nodes_required) > 0:
uuid = get_uuid()
ec2_data = EC2Controller(
ami_id,
nodes_required,
get_node_manager_user_data(boto_data, uuid),
AWS_REGION,
tags=[
{
'Key': 'Owner',
'Value': getpass.getuser(),
},
{
'Key': 'Name',
'Value': 'Daliuge NM - MsTransform',
},
{
'Key': 'uuid',
'Value': uuid,
}
]
)
ec2_data.start_instances()
reported_running = get_reported_running(
uuid,
node_count,
wait=600
)
hosts = build_hosts(reported_running)
# Create the Data Island Manager
data_island_manager = EC2Controller(
ami_id,
[
{
'number_instances': 1,
'instance_type': 'm4.large',
'spot_price': spot_price1
}
],
get_data_island_manager_user_data(boto_data, hosts, uuid),
AWS_REGION,
tags=[
{
'Key': 'Owner',
'Value': getpass.getuser(),
},
{
'Key': 'Name',
'Value': 'Daliuge DIM - MsTransform',
},
{
'Key': 'uuid',
'Value': uuid,
}
]
)
data_island_manager.start_instances()
data_island_manager_running = get_reported_running(
uuid,
1,
wait=600
)
if len(data_island_manager_running['m4.large']) == 1:
# Now build the graph
session_id = get_session_id()
instance_details = data_island_manager_running['m4.large'][0]
host = instance_details['ip_address']
graph = BuildGraphMsTransform(
work_to_do.work_to_do,
bucket_name,
volume,
7,
reported_running,
add_shutdown,
frequency_width,
session_id,
hosts,
)
graph.build_graph()
graph.tag_all_app_drops({
"session_id": session_id,
})
LOG.info('Connection to {0}:{1}'.format(host, DIM_PORT))
client = DataIslandManagerClient(host, DIM_PORT)
client.create_session(session_id)
client.append_graph(session_id, graph.drop_list)
client.deploy_session(session_id, graph.start_oids)
else:
LOG.error('Unable to find the AWS credentials')