YAO/YAO-agent
1
0
Fork 0
mirror of https://github.com/newnius/YAO-agent.git synced 2025-06-07 22:01:55 +00:00
Code Issues Releases Wiki Activity

update

Browse Source
This commit is contained in:
Newnius 2020-01-15 20:54:43 +08:00
parent 9c981a52bb
commit 702759aa18
2 changed files with 4 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
agent.py
View File

@ -36,7 +36,7 @@ def launch_tasks(stats):
entries_to_remove = [] entries_to_remove = []
lock.acquire() lock.acquire()
for task_id, task in pending_tasks.items(): for task_id, task in pending_tasks.items():
if int(utils[task['gpus'][0]]) < 75: if int(utils[task['gpus'][0]]) < 85:
entries_to_remove.append(task_id) entries_to_remove.append(task_id)
script = " ".join([ script = " ".join([
"docker exec", "docker exec",
@ -155,6 +155,7 @@ class MyHandler(BaseHTTPRequestHandler):
lock.release() lock.release()
if exit_code != 0: if exit_code != 0:
msg["code"] = 1 msg["code"] = 1
msg["error"] = output.decode('utf-8').rstrip('\n')
except Exception as e: except Exception as e:
msg = {"code": 1, "error": str(e)} msg = {"code": 1, "error": str(e)}
print(str(e)) print(str(e))
@ -214,7 +215,7 @@ class MyHandler(BaseHTTPRequestHandler):
def report(): def report():
interval = 5 interval = 1
while True: while True:
try: try:
status, msg_gpu = execute(['nvidia-smi', '-q', '-x', '-f', 'status.xml']) status, msg_gpu = execute(['nvidia-smi', '-q', '-x', '-f', 'status.xml'])
@ -322,4 +323,5 @@ if __name__ == '__main__':
t1.start() t1.start()
t2.start() t2.start()
while True: while True:
time.sleep(1)
pass pass

0
utils.py
View File