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

Re: [HTCondor-users] how to do a stripped down Schedd as CondorCE submit node?



Hi Brian,

many thanks - yes, a plain 'condor_submit' works fine with my basal
schedd/collector - I probably was again a bit overthinking... :-/

Sorry for the noise,
  Thomas

On 10/08/2020 19.52, Brian Lin wrote:
> Hi Thomas,
> 
> This is another place where the new CE submission document would help
> out; I'm hoping to complete this by HTCondor week in September.
> `condor_ce_submit` is really the most useful when submitting jobs
> directly from the CE host since it uses the default CE configuration and
> so it communicates directly with the CE schedd.
> 
> If you have a regular submit host, you'll want to use `condor_submit` to
> submit the grid universe job into your local schedd, which will then
> turn around and submit that job to the remote CE schedd.
> 
> - Brian
> 
> On 8/10/20 10:27 AM, Thomas Hartmann wrote:
>> Hi all,
>>
>> I am trying to setup a simple submit node, that should just be able to
>> submit jobs to a CondorCE (probably later on contaienrized rather than a
>> dedicated node).
>>
>> Thing is, that my condor_ce_trace are successfully towards the CE (as
>> well as condor_ping on 9619) - however, I have not managed so far to
>> setup the local condor in a way, that I can submit a 'proper' job [2].
>>
>> I have tried to setup a minimal local daemon setup with
>>   DAEMON_LIST = MASTER, COLLECTOR, NEGOTIATOR, SCHEDD
>> on the intended submitter. The local collector is listening on 9618 as
>> well as the shared port on 9620. The SchedLog does not looks suspicious
>> and a condor_q against the local schedd returns no error.
>>
>> However, when I try to submit [2] with my CE's address as target pool,
>> condor_ce_submit raises [3] where the IP address is the submitter's own,
>> i.e., it is trying to connect to 9619 locally.
>> I would have expected, that the Schedd would submit the grid universe
>> job to CondorCE:9619 - I would have assuemd, that a simple submitter
>> should be fine with just the htcondor-ce-client [1], or?
>>
>> Cheers,
>>   Thomas
>>
>>
>> [1]
>> condor-8.9.7-1.el7.x86_64
>> condor-classads-8.9.7-1.el7.x86_64
>> condor-external-libs-8.9.7-1.el7.x86_64
>> condor-procd-8.9.7-1.el7.x86_64
>> htcondor-ce-client-4.4.0-1.el7.noarch
>> python2-condor-8.9.7-1.el7.x86_64
>> python3-condor-8.9.7-1.el7.x86_64
>>
>>
>> [2]
>>> cat HTCondorCE.submit
>> universe = grid
>> use_x509userproxy = true
>> grid_resource = condor grid-htcondorce0.desy.de
>> grid-htcondorce0.desy.de:9619
>> executable = mypayload.sh
>> output = stdout
>> error = stderr
>> log = logs
>> ShouldTransferFiles = YES
>> WhenToTransferOutput = ON_EXIT
>> queue
>>
>> [3]
>>> condor_ce_submit -debug HTCondorCE.submit
>> 08/10/20 17:18:55 attempt to connect to <131.169.223.130:9619> failed:
>> Connection refused (connect errno = 111).
>>
>> ERROR: Can't find address of local schedd
>>
>> This body part will be downloaded on demand.
> 

Attachment: 0x4C44535B5D7ADD74.asc
Description: application/pgp-keys

Attachment: smime.p7s
Description: S/MIME Cryptographic Signature