Re: [condor-users] Missing condor_halt [should be condor_hold]

[axs@xxxxxxxxxx]

Hi and thanks for all the replies.

I was directed to /usr/local/condor/bin for condor_hold, I was looking in
the wrong directory.

> On Tue, Sep 23, 2003 at 12:47:36PM +0100, Sandy Spence wrote:
>> Hi,
>>
>> I don't want to stop the current job/s I just want to prevent any more
>> being submitted.
>>
>
> That's not quite what condor_hold does.

Should have made myself clearer also, I should have said executed and not
submittted.  I ran condor_hold 708.06 and sure enough the jobs state went
from 'I' to 'H', ran condor_release 708.06 back to 'I'.


Thanks again,

Sandy 'the newbee'.
>
> In condor, "submit" means to add more jobs to the queue of the Condor
> schedd. Jobs in the queue may be Idle, Running, or on Hold (there are also
> two more states, Completed and removed (indicated by 'X') that are
> temporary states and disappear from condor_q output in short order)
>
> Jobs that are Idle can move to Running, and we use the terms "running" and
> "executing" interchangeably. To prevent Idle jobs from running, you can
> put them on Hold. (And putting a Running job on Hold will cause it to
> stop Running and go to Hold. When it's released, it will return to Idle)
>
> There's currently no way to force the schedd to put every new job on
> Hold when it is submitted.
>
> To answer your original question, if you're missing condor_hold, you can
> copy condor_rm to condor_hold and run that - condor_rm, hold, and release
> are actually all the same binaries, and look at argv[0] to decide what
> they should do. (I'm nearly 100% sure that the condor_rm in 6.4.1 knows
> how to put a job on hold, but you might want to submit a test job first,
> just to make sure :)
>
> -Erik
>
>> Cheers,
>>
>> Sandy
>>
>> On Tue, 2003-09-23 at 12:38, Alexander Klyubin wrote:
>> > Hello!
>> >
>> > One of the options I see is to stop Condor on the machine(s) from
>> where
>> > the jobs were submitted. If it's just one machine, then it should be
>> > quite an easy task.
>> >
>> > Regards,
>> > Alexander Klyubin
>> >
>> > Sandy Spence wrote:
>> > > Hi,
>> > >
>> > > I am the administrator for our three Beowulf clusters we are running
>> > > Condor version 6.4.1 on our latest.  I wanted to carry out some
>> > > maintenance work and tried to put on hold the idle jobs in the
>> current
>> > > queue (all of the jobs take in excess of two days to complete).  It
>> > > appears that condor_hold is not present; can anybody tell me how I
>> can
>> > > work around this?
>> > >
>> > >
>> > > Cheers
>> > >
>> >
>> > Condor Support Information:
>> > http://www.cs.wisc.edu/condor/condor-support/
>> > To Unsubscribe, send mail to majordomo@xxxxxxxxxxx with
>> > unsubscribe condor-users <your_email_address>
>> --
>> Sandy Spence
>> Computer Officer
>> Department of Computer Science
>> University of Wales, Aberystwyth
>> Penglais Campus
>> Aberystwyth
>> SY23 3DB
>>
>> Tel:  01970-622433
>> Fax:  01970-628536
>> http://users.aber.ac.uk/axs
>>
>> Condor Support Information:
>> http://www.cs.wisc.edu/condor/condor-support/
>> To Unsubscribe, send mail to majordomo@xxxxxxxxxxx with
>> unsubscribe condor-users <your_email_address>
>>
> Condor Support Information:
> http://www.cs.wisc.edu/condor/condor-support/
> To Unsubscribe, send mail to majordomo@xxxxxxxxxxx with
> unsubscribe condor-users <your_email_address>
>
>

Condor Support Information:
http://www.cs.wisc.edu/condor/condor-support/
To Unsubscribe, send mail to majordomo@xxxxxxxxxxx with
unsubscribe condor-users <your_email_address>