-
Notifications
You must be signed in to change notification settings - Fork 1.7k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
prefect integration support volcano job #17260
Comments
Thanks for the request @HeavenS221! I bet you could use an existing Kubernetes work pool to start volcano jobs by editing the Can you try updating the |
fwiw @kevingrismore wrote a volcano worker a while ago https://github.com/kevingrismore/volcano-worker/blob/main/prefect_volcano/worker.py |
@desertaxle Thanks for response. I've tried this option, but it doesn't work, there is a base job manifest validation in integrations/prefect-kubernetes/worker.py func _get_base_job_manifest() |
@zzstoatzz Thanks for info, is there any plan to accept this integration? |
@HeavenS221 I'd be open to relaxing that validation to enable alternative job implementations. Do you want to submit a PR for that? |
@desertaxle sounds good to me, any developer guilds? |
@HeavenS221 Yep! Here's our guide for contributing to our integrations: https://docs.prefect.io/contribute/contribute-integrations |
Describe the current behavior
current behavior:
work pool only support to launch kubernetes job, job template is hardcode written in src/integrations/prefect-kubernetes/prefect_kubernetes/.
Describe the proposed behavior
expect behavior:
support volcano job and kubernetes job, when work pool loading base job template['job_configuration']['job_manifest'] config, based on some key value like ['job_type'] = 'kubernetes'/ 'volcano' to decide which job will be launched.
Example Use
VolcanoJob, referred to as vcjob, is a CRD object for Volcano. Different from a Kubernetes job, it provides more advanced features such as specified scheduler, minimum number of members, task definition, lifecycle management, specific queue, and specific priority. VolcanoJob is ideal for high performance computing scenarios such as machine learning, big data applications, and scientific computing.
Our team hope to use prefect to scheduler several task flow, i.e. data processing, LLM model training, model eval, model compiling. There are several plugin functions like pytorch-plugin supported by volcano job.
Additional context
our team hope to use prefect to scheduler several task flow, i.e. data processing, LLM model training, model eval, model compiling. There are several plugin func supported by volcano job.
https://github.com/volcano-sh/volcano/blob/master/docs/user-guide/how_to_use_pytorch_plugin.md
The text was updated successfully, but these errors were encountered: