Re: [HTCondor-users] HTCondor 24.0.X: error/output transfer back to AP with held jobs

[Carles Acosta <cacosta@xxxxxx>]

Hi again,

Just to close this topic and clarify, I see that, using WhenToTransferOutput = "ON_EXIT_OR_EVICT" and SpoolOnEvict=false:

1) If the job is held due to a condition in the EP (startd), the standard output and error are not transferred back

2) If the job is held due to a condition in the AP (schedd), the standard output and error are transferred back

Is this the expected behavior?

Thank you very much.

Cheers,

Carles

On Mon, 31 Mar 2025 at 09:33, Carles Acosta <cacosta@xxxxxx> wrote:

Hi Todd,

Thanks for checking, and you're right, I was doing something wrong with my initial tests. Let me explain.Â

Running both versions, if I use the condor_hold command, the standard output and error are transferred back to the path or initialdir as expected.Â

However, the issue is when the job is held due to a condition in the EP, like Memory Exceded, something like this in the STARTD:

use POLICY : WANT_HOLD_IF(MEMORY_EXCEEDED, 102, $(MemoryHoldReason))

Then, when the job is held for exceeding the memory, the standard output and error are not transferred back. But this is not happening only for the 24.0.5 version, it is also happening for the 23.0.19 version. Last Friday, I was doing some tests, and I was pretty sure that with 23.0.19 version the output and error files were transferred, but I could not reproduce that today.Â

Cheers,

Carles

On Fri, 28 Mar 2025 at 20:58, Todd L Miller via HTCondor-users <htcondor-users@xxxxxxxxxxx> wrote:

> We are testing the 24.0.5 version now, and we see that the standard error
> and output are transferred back to the AP when the job is completed or
> removed but not when the job is held. Maybe we are missing something, but
> we are using the same configuration as in 23.0.X (and before).

    I can't reproduce the problem: in both versions, I see the job's
files in AP's SPOOL directory after running the condor_hold command. Do
you see the files when you run condor_hold but in some other situation(s)
when the job goes on hold? If so, could you be more specific about both
the job and the situation(s)?

-- ToddM
_______________________________________________
HTCondor-users mailing list
To unsubscribe, send a message to htcondor-users-request@xxxxxxxxxxx with a
subject: Unsubscribe

Join us in June at Throughput Computing 25: https://osg-htc.org/htc25

The archives can be found at: https://www-auth.cs.wisc.edu/lists/htcondor-users/

--

Carles Acosta i Silva

PIC (Port d'Informacià CientÃfica)

Campus UAB, Edifici D

E-08193 Bellaterra, Barcelona

Tel: +34 93 581 33 08

Fax: +34 93 581 41 10

http://www.pic.esÂ

AvÃs - Aviso - Legal Notice: Âhttp://legal.ifae.es

--

Carles Acosta i Silva

PIC (Port d'Informacià CientÃfica)

Campus UAB, Edifici D

E-08193 Bellaterra, Barcelona

Tel: +34 93 581 33 08

Fax: +34 93 581 41 10

http://www.pic.esÂ

AvÃs - Aviso - Legal Notice: Âhttp://legal.ifae.es