jacenkow/gpu-sentry

Flask-based package for monitoring utilisation of nVidia GPUs.

gpu-sentry's Issues

feature: add booking system

The idea is to have a calendar grid where a user will be able to reserve a particular GPU with 15-minutes slots. The reservation system will not block any physical resources within the GPU machines but rather inform other users that a resource might be needed.

The feature depends on

• persistent storage
• user management

web page not refresh

The server successfully received the data, but the web page did not refresh auto. the client only sent data once, not loop

Gpu-box not shown on webpage

Hey i followed your steps to install this on my computers
I started the client on one computer with gpu and started server with another computer.
I edited the config.py in client: added client computer details in permit_clients and changed server hostname to the ip of the server computer but client box is not seen on the webpage hosted on server computer.
Please help

error trying to run client

Here's the traceback:

Unhandled error in Deferred:

Traceback (most recent call last):
  File "cli.py", line 42, in cli
    run_client()
  File "/home/will/DevWorkspace/gpu-sentry/gpu_sentry/client.py", line 90, in run_client
    l.start(config.CLIENT_TIMEOUT)
  File "/home/will/anaconda3/lib/python3.6/site-packages/twisted/internet/task.py", line 194, in start
    self()
  File "/home/will/anaconda3/lib/python3.6/site-packages/twisted/internet/task.py", line 239, in __call__
    d = defer.maybeDeferred(self.f, *self.a, **self.kw)
--- <exception caught here> ---
  File "/home/will/anaconda3/lib/python3.6/site-packages/twisted/internet/defer.py", line 151, in maybeDeferred
    result = f(*args, **kw)
  File "/home/will/DevWorkspace/gpu-sentry/gpu_sentry/client.py", line 81, in send_statistics
    json={"codename": config.PERMIT_CLIENTS[host]['codename'],
builtins.KeyError: 'gpgpu'

Also, if I try to run outside of /gpu-sentry/gpu_sentry I get

python: can't open file 'cli.py': [Errno 2] No such file or directory

No module named gpu_sentry.client

I have installed the dependencies and still getting the following error:

python3 gpu_sentry/cli.py -m client
Traceback (most recent call last):
  File "gpu_sentry/cli.py", line 29, in <module>
    from gpu_sentry.client import run_client
ModuleNotFoundError: No module named 'gpu_sentry'

Missing hostname

What if my server can't resolve hostnames of GPU-hosts? How to configure client- and server-side config.py one having IPs only?

feature: add weekly usage statistics

At the end of usage period, e.x. 7 days, the system should create a report with weekly usage statistics. The report can be then dispatched using an e-mail.

The feature depends on

• persistent storage
• email (?)