[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] 7.8.3: "rejected by your job requirements" again



On 09/25/2012 08:44 AM, Tim St Clair wrote:
> can you please dump a condor_q -long on a stuck job into this email thread. 

It's too late now. I will if it happens again, but I've already changed
the submit file to "request_memory = 1K".

> Typically >> 7.8 it is recommended to use the Request* and not alter the requirements. 

What I actually want is "requrements = off" switch. I *know* that a)
condor gets memory requirements wrong for these jobs and b) all nodes in
my pool have enough resources to run these jobs.

As far as not clearly documented side-effects that are subject to change
without notice go, "Memory > 0" is a much cleaner one than
"request_memory = some-number".

What does "request_memory = 0" mean? -- nothing gets to run because
every node has > 0 memory?

What happens when I lie to condor and "request-memory = 1K" and then the
job actually asks for 2GB? And so on.

So it may be a much shinier knob for fine-tuning the requirements, but
unfortunately it's the exact opposite of what I need.

-- 
Dimitri Maziuk
Programmer/sysadmin
BioMagResBank, UW-Madison -- http://www.bmrb.wisc.edu

Attachment: signature.asc
Description: OpenPGP digital signature