-
Notifications
You must be signed in to change notification settings - Fork 41
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
Match allocate_with_satisfiability success repsonse contains empty R
#679
Comments
Thanks. Once @grondo has ubuntu focal docker image, let's see if this still is the problem under that environment and if so debug... |
I pushed a |
Awesome! Thank you SO much @grondo! |
Turns out that this will happen when you use the This wasn't breaking in our tests with In terms of action items, here are the ones that come to mind:
Note: for the most part, I don't think this bug will affect a full production system since the job-ingest module will reject any non-V1 jobspecs masquerading as V1. |
This problem occurs both with hwloc v1 and v2, correct? Yes, Since jobspec will ultimately grow to be the full canonical specification, I think a future proof solution will be to add support to the traverser such that we can emit the full subtree under an exclusive allocation. Then, For example, currently, with jobspec= We talked about this before (I think the topic was shadowed resource or something) but decided to kick the can down the road. Is this a good time to revisit this?
For #1, please see Issue #618. We do need to firm up error propagation. If you want, you can add #1 to that ticket as well. |
Ah! That makes a lot of sense. I like the idea. Is there an open issue on that? I don't see one from a cursory search/glance. If not, maybe we split that out into a separate issue.
Good question. I'm not sure. Maybe something to synchronize on during the team meeting on Thursday?
Will do! I'll also open an issue about the versioning of jobspecs in the t/data directories. We can probably punt on that until we support jobspec v2 more thoroughly. |
Probably not on this particular issue. We should create a ticket on the semantics and handling for show resources. |
I was actually wrong about this. The current V1 validation is not robust enough to catch this type of invalid V1 jobspec. Opened an issue over in flux-core: flux-framework/flux-core#3039
Ok. I'll open a new issue. |
@SteVwonder: should we still keep this open with Issue #689 ? |
While interactively debugging an issue with #677, @dongahn and I encountered a situation where the
flux-ion-resource.py
script was printing a successful allocation with an emptyR
value. Need to investigate further under what circumstances the response is empty.The text was updated successfully, but these errors were encountered: