--optGridNoSubmit asks for a parameter, but prun sees it as a flag?

[lhenkelm]

As the title says: despite the name, and indeed prun --helpGroup SUBMIT indicating it is a pure flag/switch (i.e. it only flips a bool from false to true, or vice versa), the xAH_run.py [...] prun --optGridNoSubmit currently requires an additional integer (?) parameter.

Is this intentional, or an oversight? If its intentional, what happens with integer in the PrunDriver?

Cheers,
Lars

[lhenkelm]

furthermore, it seems no matter what value I pass, the driver ignores it anyaways?
I tried -1, 0, and 1, and in every case jobs were sent to the grid as usual...

[kratsg]

Hi, this isn't an xAH problem. It's not supported by ELG_prun.py (and thus EventLoop).

[kratsg]

What is your use case for using this flag?

[lhenkelm]

Hi, to answer the use case question:
I was trying to play around with some of the other options, in particular --optGridOutputSampleName,
to get an idea of how they work, without having to send jobs and them kill them again.

[kratsg]

Dry-run isn't supported via EventLoopGrid. Giordon

…

On Mon, Feb 17, 2020 at 12:32 PM lhenkelm ***@***.***> wrote: Hi, to answer the use case question: I was trying to play around with some of the other options, in particular --optGridOutputSampleName, to get an idea of how they work, without having to send jobs and them kill them again. — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub <#1413?email_source=notifications&email_token=AAFZ5C53KY55NPTNZRRBZZDRDJYTXA5CNFSM4KVKQMAKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEL6CRLY#issuecomment-586950831>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAFZ5C3QZVOJND2QGT33LETRDJYTXANCNFSM4KVKQMAA> .

-- Dr. Giordon Stark (pronouns: he/him) ATLAS Post-doctoral scholar employee at SCIPP/UCSC https://giordonstark.com/ <https://giordonstark.com/[email protected]&utm_medium=email_signature>

[lhenkelm]

Hi Giordon, are you sure? I have never used the ELG_prun.py script you linked above, so I don't know, but it looks like they actually use the "--noSubmit" option internally to build a tarball before submitting: https://gitlab.cern.ch/atlas/athena/blob/21.2/PhysicsAnalysis/D3PDTools/EventLoopGrid/data/ELG_prun.py#L103. If this is not supported by the implementation, would it make sense to drop the option from the xAH CLI? Cheers, Lars

…

On 17/02/2020 22:13, Giordon Stark wrote: Dry-run isn't supported via EventLoopGrid. Giordon On Mon, Feb 17, 2020 at 12:32 PM lhenkelm ***@***.***> wrote: > Hi, to answer the use case question: > I was trying to play around with some of the other options, in particular > --optGridOutputSampleName, > to get an idea of how they work, without having to send jobs and them kill > them again. > > — > You are receiving this because you modified the open/close state. > Reply to this email directly, view it on GitHub > <#1413?email_source=notifications&email_token=AAFZ5C53KY55NPTNZRRBZZDRDJYTXA5CNFSM4KVKQMAKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEL6CRLY#issuecomment-586950831>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/AAFZ5C3QZVOJND2QGT33LETRDJYTXANCNFSM4KVKQMAA> > . > -- Dr. Giordon Stark (pronouns: he/him) ATLAS Post-doctoral scholar employee at SCIPP/UCSC https://giordonstark.com/ ***@***.***&utm_medium=email_signature> — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#1413?email_source=notifications&email_token=AI2ZUX4BN3YLL2XS77JOK7DRDMDZ5A5CNFSM4KVKQMAKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEL7ZJQQ#issuecomment-587175106>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AI2ZUX7GLHB36YJHP4BX5P3RDMDZ5ANCNFSM4KVKQMAA>.

[kratsg]

Hi @lhenkelm , if you use the EventLoopGrid drivers (e.g. PrunDriver) -- it calls/relies on ELG_prun.py for the submission. We can comment out that option given that it is not used. But note that configuration is still valid, just that the implementation that exists in atlas/[email protected] does not use it.