Skip to content
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

Enable pilot job submission for HTCondor service #105

Closed
3 tasks done
rafaelfsilva opened this issue May 16, 2019 · 5 comments
Closed
3 tasks done

Enable pilot job submission for HTCondor service #105

rafaelfsilva opened this issue May 16, 2019 · 5 comments
Assignees
Milestone

Comments

@rafaelfsilva
Copy link
Member

rafaelfsilva commented May 16, 2019

  • Implementation of pilot job support
  • Update documentation
  • Implement tests
@henricasanova
Copy link
Contributor

While trying a test for this, I become a bit confused. It looks like the HT service will simply start jobs on the next available sub-service (e.g., a BareMetalComputeService). But what happens is that a PilotJob would simply be sent to the next service. And if that service doesn't support pilot jobs, then there is an "doesn' t support pilot job". So I am not sure what it means for HTCondor to support pilot jobs actually.. The idea is that it simply relays pilot job requests to one of the subservices? But then it should filter out those services that do not support pilot jobs (and if none does, then too bad, failure). And then, when multiple do, how do we pick?

@rafaelfsilva
Copy link
Member Author

I see your confusion. Actually, HTCondor does not make difference between pilot jobs or regular jobs. However, since our implementation makes this clear difference (and callbacks are handled differently). HTCondor matches the first available service to submit the job (assuming all requirements are met), thus no smart decision is made here. If someone wants a very specific dataset, they need to pass it as part of a requirement. Does this make sense to you?

@henricasanova
Copy link
Contributor

Ok, so I'll just change the code in HTCondor so that if a pilot job is submitted, those sub-services that do not support pilot jobs are not considered (and if none does, then some error). Not sure what you mean by "dataset"?

@rafaelfsilva
Copy link
Member Author

yes, sounds good. "dataset" is most probably a typo ;) I believe I meant to type "behavior"

@henricasanova
Copy link
Contributor

henricasanova commented Nov 8, 2019

I have improved some of the implementation and error/catching based on the above. I am closing this issue and opening a new one.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

2 participants