Quickstart: Launch Interactive Build Workloads¶
Deep learning workloads can be divided into two generic types:
- Interactive "build" sessions. With these types of workloads, the data scientist opens an interactive session, via bash, Jupyter notebook, remote PyCharm, or similar and accesses GPU resources directly.
- Unattended "training" sessions. With these types of workloads, the data scientist prepares a self-running workload and sends it for execution. During the execution, the customer can examine the results.
With this Quickstart you will learn how to:
- Use the Run:AI command-line interface (CLI) to start a deep learning Build workload
- Open an ssh session to the Build workload
- Stop the Build workload
It is also possible to open ports to specific services within the container. See "Next Steps" at the end of this article.
To complete this Quickstart you must have:
- Run:AI software is installed on your Kubernetes cluster. See: Installing Run:AI on an on-premise Kubernetes Cluster
- Run:AI CLI installed on your machine. See: Installing the Run:AI Command-Line Interface
Step by Step Quickstart¶
- Login to the Projects area of the Run:AI Administration user interface at https://app.run.ai/projects
- Add a project named "team-a"
- Allocate 2 GPUs to the project
At the command-line run:
runai config project team-a runai submit build1 -i ubuntu -g 1 --interactive --command -- sleep infinity
The job is based on a sample docker image
- We named the job build1.
- Note the interactive flag which means the job will not have a start or end. It is the researcher's responsibility to close the job.
- The job is assigned to team-a with an allocation of a single GPU.
- The command provided is
sleep infinity. You must provide a command or the container will start and then exit immediately. Alternatively, replace these flags with
--attachto attach immediately to a session.
Follow up on the job's status by running:
runai list jobs
Typical statuses you may see:
- ContainerCreating - The docker container is being downloaded from the cloud repository
- Pending - the job is waiting to be scheduled
- Running - the job is running
A full list of Job statuses can be found here
To get additional status on your job run:
runai describe job build1
Get a Shell to the container¶
runai bash build1
This should provide a direct shell into the computer
View status on the Run:AI User Interface¶
- Go to https://app.run.ai/jobs
- Under "Jobs" you can view the new Workload:
Run the following:
runai delete build1
This would stop the training workload. You can verify this by running
runai list jobs again.
- Expose internal ports to your interactive build workload: Quickstart Launch an Interactive Build Workload with Connected Ports.
- Follow the Quickstart document: Launch Unattended Training Workloads.