feat: communication about docker image information
This commit is contained in:
parent
3184669630
commit
55211c3856
@ -1,4 +1,6 @@
|
||||
from os.path import isdir
|
||||
import socket
|
||||
import os
|
||||
import json
|
||||
import threading
|
||||
import docker
|
||||
@ -113,10 +115,14 @@ class ClusterCommunicationModule():
|
||||
self.client_sock.send('[START_LISTEN_TASK_CHECK] {}'.encode())
|
||||
print("The master has started listening for new task from Sepolia testnet...")
|
||||
elif command == '[RUN_CONTAINER]':
|
||||
image = json.loads(args)['image']
|
||||
print(f"[RUN_CONTAINER] {image}")
|
||||
time.sleep(3)
|
||||
args = json.loads(args)
|
||||
image = args['image']
|
||||
train = args['train']
|
||||
|
||||
print(f"[RUN_CONTAINER] {image}")
|
||||
self.run_container(image, train, args)
|
||||
|
||||
print(f"[RUN_CONTAINER] {image}")
|
||||
self.client_sock.send('[RUN_CONTAINER_SUCCESS] {}'.encode())
|
||||
|
||||
return True
|
||||
@ -185,18 +191,75 @@ class ClusterCommunicationModule():
|
||||
for conn in self.worker_conns:
|
||||
conn.close()
|
||||
|
||||
def run_container(self, image_name):
|
||||
def run_container(self, image_name, train, train_args={})
|
||||
'''
|
||||
train_args
|
||||
- index
|
||||
- node_num
|
||||
'''
|
||||
if not os.path.isdir('./dataset_dir'):
|
||||
os.mkdir('./dataset_dir')
|
||||
print("Create ./dataset_dir dir.")
|
||||
if not os.path.isdir('./output'):
|
||||
os.mkdir('./output')
|
||||
print("Create ./output dir.")
|
||||
|
||||
if not train:
|
||||
container = self.node_manager.docker_client.containers.run(
|
||||
image_name,
|
||||
volumes={'dataset_dir': {'bind': '/dataset', 'mode': 'rw'}},
|
||||
detach=True
|
||||
)
|
||||
else:
|
||||
container = self.node_manager.docker_client.containers.run(
|
||||
image_name,
|
||||
volumes={
|
||||
'dataset_dir': {'bind': '/dataset', 'mode': 'rw'},
|
||||
'output': {'bind': '/output', 'mode': 'rw'},
|
||||
},
|
||||
network='train-net',
|
||||
runtime='nvidia',
|
||||
device_requests=[
|
||||
docker.types.DeviceRequest(count=-1, capabilities=[['gpu']])
|
||||
],
|
||||
name=f'train-{train_args["index"]}',
|
||||
env={
|
||||
'GPU_NUM': self.node_manager.GPU_num,
|
||||
'NODE_NUM': train_args['node_num'],
|
||||
'NODE_RANK': train_args['index'],
|
||||
'MASTER_IP': 'train-0',
|
||||
'MASTER_PORT': 21046,
|
||||
},
|
||||
detach=True
|
||||
)
|
||||
|
||||
print(container.short_id)
|
||||
for line in container.logs(stream=True):
|
||||
print(line.strip().decode())
|
||||
|
||||
result = container.wait()
|
||||
status_code = result['StatusCode']
|
||||
print(status_code, type(status_code))
|
||||
|
||||
def scatter_container(self, image_name, train=False):
|
||||
def master_run(image_name):
|
||||
print("[Master] run")
|
||||
for i in range(5):
|
||||
print(i)
|
||||
time.sleep(1)
|
||||
train_args = {
|
||||
'index': 0,
|
||||
'node_num': len(self.worker_conns)+1
|
||||
}
|
||||
self.run_container(image_name, train, train_args)
|
||||
print("[Master] finished")
|
||||
|
||||
def send_and_wait(conn_index, conn, image_name):
|
||||
try:
|
||||
# build command
|
||||
data = {'image': image_name}
|
||||
data = {
|
||||
'image': image_name,
|
||||
'train': train,
|
||||
'index': conn_index+1,
|
||||
'node_num': len(self.worker_conns)+1
|
||||
}
|
||||
command = '[RUN_CONTAINER] {}'.format(json.dumps(data))
|
||||
|
||||
# send
|
||||
@ -226,7 +289,7 @@ class ClusterCommunicationModule():
|
||||
thread.join()
|
||||
|
||||
# all finished
|
||||
print("[INFO] All workers finished.")
|
||||
print("\n[INFO] All workers finished.")
|
||||
|
||||
|
||||
'''
|
||||
|
||||
@ -119,10 +119,10 @@ class NodeManager():
|
||||
print(f" - Training Image: {train_image}")
|
||||
'''
|
||||
|
||||
data_image = "test/test"
|
||||
data_image = "snsd0805/cifar10-dataset:v1"
|
||||
|
||||
# Start Downloading
|
||||
self.cluster_communication_module.run_container(data_image)
|
||||
self.cluster_communication_module.scatter_container(data_image)
|
||||
|
||||
|
||||
else:
|
||||
|
||||
Loading…
Reference in New Issue
Block a user