@<1544853688374267904:profile|ResponsiveAnt93> the autoscaler is not designed to spin docker images independently - it will always spin a machine and install an agent on it, than instruct the agent to pull tasks from a specified queue. The agent, in turn will pull a task and execute it. Executing a task within a docker container is a process in which an agent is spun inside the container and handles environment setup and task execution - it is not a process of just spinning a pre-packaged docker container (even in standalone mode, it assume an agent installed in the image).
What didn't work when cloning the repo and creating the environment?
Answered
I Am Trying To Run A Python File With The Clearml Using Aws. Because Cloning The Repo And Creating An Environment Kept Failing, I Have Decided To Use Docker.
I Created A Docker On My Local Computer. I Mounted 2 Volumes. One For My Directory With My Code
I am trying to run a python file with the clearML using AWS. Because cloning the repo and creating an environment kept failing, I have decided to use docker.
I created a docker on my local computer. I mounted 2 volumes. One for my directory with my code (this includes many subdirectories that I turn into python modules) and another where I store my training data. When I run this locally, I do the following:
- Start the docker - during startup as defined in the Docker file, this runs a few shell commands to create the python modules- run
clearml-init
and then put in my personal login info - run main.py with any relevant command line arguments
The above series of steps works locally and creates project on the clearML website.
Now I want to put this on AWS with the clearML agent and it isn't working
I clone the project on the clearML website and enqueue it. It seems to build the environment and then try to run the python code main.py
- The error is that there is "No module named [local module name]"- This is the module that was installed in the docker startup file- I tried to move the shell commands to the python file and call them using os.system() and that doesnt work
Can someone please help?
1K Views
1
Answer
one year ago
one year ago
Tags
Similar posts