Re: [Condor-users] about condor_submit in condor-g


Date: Thu, 3 Feb 2005 15:22:15 +0800 (HKT)
From: "Carson Hung" <carson@xxxxxxxxxxxxxx>
Subject: Re: [Condor-users] about condor_submit in condor-g
Originally, I run the condor_master as root (it will bring up
condor_schedd also). But when I check with PS command, the owner become
condor.

By the way, the error msg is:
2/3 15:16:33 ******************************************************
2/3 15:16:33 ** condor_gridmanager (CONDOR_GRIDMANAGER) STARTING UP
2/3 15:16:33 ** /opt/condor/sbin/condor_gridmanager
2/3 15:16:33 ** $CondorVersion: 6.7.3 Dec 28 2004 $
2/3 15:16:33 ** $CondorPlatform: I386-LINUX_RH9 $
2/3 15:16:33 ** PID = 20301
2/3 15:16:33 ******************************************************
2/3 15:16:33 Using config file: /opt/condor/etc/condor_config
2/3 15:16:33 Using local config files:
/opt/condor/local.pc-0242/condor_config.local
2/3 15:16:33 DaemonCore: Command Socket at <192.168.1.1:57928>
2/3 15:16:36 [20301] DaemonCore: Command received via UDP from host
<192.168.1.1:32847>
2/3 15:16:36 [20301] DaemonCore: received command 60000 (DC_RAISESIGNAL),
calling handler (HandleSigCommand())
2/3 15:16:36 [20301] Found job 78.0 --- inserting
2/3 15:16:36 [20301] Failed to get expiration time of proxy
/opt/condor/local.pc-0242/spool/cluster79.proc0.subproc0/x509up_u500
2/3 15:16:36 [20301] (79.0) error acquiring proxy!

Thanks,
Carson

> Is your schedd running as root? What are the permissions on the proxy
> file? You can find the proxy file in the condor spool directory, in a
> directory named clusterXX.procXX.subproc, with the XX's replaced with
> the job id.
>
> On Thu, 3 Feb 2005, Carson Hung wrote:
>
>> Hi,
>>
>> First of all, thanks alot for your reply.
>>
>> I have setup the things according. However, I have met another
>> problem. The remote machine seems to transfer also its proxy
>> credential file to local machine spool. However, it seems that local
>> machine does not have the access right to open that file and the
>> prompt some security error.
>>
>> Thanks for any suggestion about this issue.
>>
>> Thanks,
>> Carson
>>
>> > On Wed, 2 Feb 2005, Carson Hung wrote:
>> >
>> >> I succeeded in queuing up the jobs in a local machine with
>> >> condor_schedd.
>> >>
>> >> However, May I ask a question on how to submit a job from a remote
>> machine to local machine?
>> >>
>> >> I have checked the condor_submit command. It has an option called
>> -r which takes in remote schedd name. However, How can I set the
>> schedd name in local machine? and how can remote machine knows this
>> schedd name?
>> >
>> > First, for remote submission to work, you need to have a collector
>> running. The collector is where the Condor daemons advertise
>> themeselves so that they can be found by others.
>> >
>> > Assuming everything is set up correctly, you should be able to run
>> 'condor_status -schedd' from the remote machine and see your schedd
>> listed in the output. The string under the Name heading (which may
>> be truncated) is what you use with condor_submit -r. By default,
>> schedds running as root or condor are named after their hostname.
>> Schedds running as a user are named <username>@<hostname>. You can
>> set a schedd's name with SCHEDD_NAME in the config file. This is
>> handy if you want to run multiple schedds on the same machine.
>> >
>> > Be aware that condor_submit -r doesn't transfer any of the job's
>> files from the remote to the local machine. If you want to do that,
>> you need to use condor_submit -s. However, condor_submit -s has
>> problems with schedds running as root, which we're currently working
>> to fix.
>
>
> +----------------------------------+---------------------------------+ |
>            Jaime Frey            |  Public Split on Whether        | |
>      jfrey@xxxxxxxxxxx         |  Bush Is a Divider              | |
> http://www.cs.wisc.edu/~jfrey/  |         -- CNN Scrolling Banner |
> +----------------------------------+---------------------------------+




[← Prev in Thread] Current Thread [Next in Thread→]