Re: [condor-users] jobs don't start

[Erik Paulson <epaulson@xxxxxxxxxxx>]

On Thu, Jul 31, 2003 at 12:35:40PM +0100, Alexander Klyubin wrote:
> I wonder why the ShadowLog you sent earlier references some checkpoint 
> file which it allegedly fails to transfer: 
> C:\Condor\spool\cluster7.ickpt.subproc0. You aren't using any 
> checkpointing -- Vanilla universe does not support it...
> I hope trapezregel.exe was compiled *without* Condor checkpointing?
> 

It's OK. A long time ago, _everything_ in Condor had to be able to be
checkpointed. When you first submitted your job, Condor would copy the
executable to the spool directory (so it would always have a copy)
and called that the "initial checkpoint", hence the ickpt in the filename.
Condor knows that trapezregel is a vanilla job, and will remember that when
it starts the job up on the other side.


Thomas - the Shadow Exceptions seem to be because it's loosing contact
with the condor_starter on the other side. (The errno 10054 is a Winsock
"Connection reset by peer") - it would be helpful to see a StarterLog from
the machine 128.176.206.149.

You're also the first person that we know of who's tried to run on 
Windows 2003 Server, but I don't think that's a problem.

Something else to try - Condor 6.5.4 for Windows was put out on the 
Condor website but hasn't been fully announced. It does fix some random
crashes on Windows, which might be what's going on here (we'd really need to
see the StarterLog to figure that out)

Thanks,

-Erik


> 
> If trapezregel.exe is just a normal executable, which runs perfectly on 
> any of those machines if you install it there manually, then may be you 
> should try explicitly telling Condor which files to transfer and when. 
> Try adding following to the submission file:
> 
> transfer_input_files = trapezregel.exe
> should_transfer_files = YES
> when_to_transfer_output = ON_EXIT_OR_EVICT
> 
> The above assumes that you need to send in only the executable file. The 
> executable does not need anything else to run.
> 
> Regards,
> Alexander Klyubin
> 
> On 07/31/2003 12:26 PM, Thomas Bauer wrote:
> > Hi Alexander,
> > 
> > my submission file is:
> > --------------------------------------------------------------
> > universe=vanilla
> > executable=trapezregel.exe
> > output=trapez.out
> > error=trapez.err
> > log=trapez.log
> > requirements= OpSys=="WINNT50"
> > queue
> > ---------------------------------------------------------
> > 
> > Thomas
> > 
> > On Thu, 31 Jul 2003 11:29:48 +0100, Alexander Klyubin 
> > <A.Kljubin@xxxxxxxxxxx> wrote:
> > 
> >> Hello Thomas,
> >>
> >> can you send you job submission file (the one you submit using 
> >> condor_submit)?
> >>
> >>
> >> Alexander Klyubin
> >>
> >> On 07/31/2003 11:23 AM, Thomas Bauer wrote:
> >>
> >>> Hi again,
> >>> 
> >>> I checked the log files and found the following entrys:
> >>>
> >>> ------------------------ShadowLog on the 
> >>> Condor-Server-------------------
> >>>
> >>> --- --------------------
> >>> (...)
> >>> 7/31 11:31:05 ******************************************************
> >>> 7/31 11:31:05 ** condor_shadow (CONDOR_SHADOW) STARTING UP
> >>> 7/31 11:31:05 ** $CondorVersion: 6.5.3 Jul  3 2003 $
> >>> 7/31 11:31:05 ** $CondorPlatform: INTEL-WINNT40 $
> >>> 7/31 11:31:05 ** PID = 2368
> >>> 7/31 11:31:05 ******************************************************
> >>> 7/31 11:31:05 Using config file: C:\Condor\condor_config
> >>> 7/31 11:31:05 Using local config files: C:\Condor\condor_config.local
> >>> 7/31 11:31:06 DaemonCore: Command Socket at <128.176.206.141:2317>
> >>> 7/31 11:31:07 Initializing a VANILLA shadow
> >>> 7/31 11:31:11 (7.0) (2368): Request to run on <128.176.206.149:1047> 
> >>> was ACCEPTED
> >>> 7/31 11:31:17 (7.0) (2368): ReliSock: put_file: TransmitFile() 
> >>> failed, errno=10054
> >>> 7/31 11:31:17 (7.0) (2368): ERROR "DoUpload: Failed to send file 
> >>> C:\Condor\spool\cluster7.ickpt.subproc0, exiting at 1371
> >>> at line 1370 in file ..\src\condor_c++_util\file_transfer.C
> >>> (...)
> >>> ------------------------------------------------------------------------- 
> >>>
> >>>
> >>> --- ------
> >>>
> >>> There seems to be a problem to send the files to the Client.
> >>> The SchedLog says:
> >>>
> >>> ---------------------- SchedLog on Server 
> >>> -------------------------------
> >>>
> >>> --- ----
> >>> (...)
> >>> 7/31 11:31:01 Started shadow for job 7.0 on "<128.176.206.149:1047>", 
> >>> (shadow pid = 2368)
> >>> 7/31 11:31:02 Sent ad to central manager for tombauer@xxxxxxxxxxx 
> >>> muenster.de
> >>> 7/31 11:31:19 DaemonCore: Command received via UDP from host 
> >>> <128.176.206.141:2330>
> >>> 7/31 11:31:19 DaemonCore: received command 60001 (DC_PROCESSEXIT), 
> >>> calling handler (HandleProcessExitCommand())
> >>> 7/31 11:31:19 ERROR: Shadow exited with job exception code!
> >>> (...)
> >>> ------------------------------------------------------------------------- 
> >>>
> >>>
> >>> --- --
> >>>
> >>> Greetings,
> >>> Thomas Bauer
> >>> PS: My setup is an Intel Windows2003 Server (Terminal Server) and 3 
> >>> Intel Windows2000-Clients.
> >>> When I submit a job, I log on on the Terminalserver, copy my files to 
> >>> his local harddisk and start the job from there.
> >>> Maybe there is a problem with Condor and Win2003 Server?
> >>>
> >>>
> >>>
> >>>
> >>>
> >>> On Thu, 31 Jul 2003 09:36:36 +0100, Alexander Klyubin 
> >>> <A.Kljubin@xxxxxxxxxxx> wrote:
> >>>
> >>>> I encountered similar problems when the job actually failed to start 
> >>>> (due to a bug it contained). You may want to check the logs the job 
> >>>> produces on the remote machine. You should also definitely check 
> >>>> StartLog and StarterLog of the remote machine.
> >>>>
> >>>> Moreover, if you have a firewall between or on the submission 
> >>>> machine or on worker machine this may also contribute to the problem.
> >>>>
> >>>> Regards,
> >>>> Alexander Klyubin
> >>>>
> >>>> On 07/31/2003 09:31 AM, Thomas Bauer wrote:
> >>>>
> >>>>> Hello,
> >>>>>
> >>>>> I have a WINNT-Server and some WINNT-Clients. When I submit a job, 
> >>>>> the job don't start on any machine.
> >>>>> "condor_q - analyze" tells me, that "1 match, but prefer another 
> >>>>> specific job despite its worse user-prioriy"
> >>>>> The log-file of this job shows entry like these:
> >>>>> ----------------------------------------------------------------------- 
> >>>>>
> >>>>>
> >>>>> -
> >>>>> 007 (023.000.000) 07/08 17:52:34 Shadow exception! Can no longer 
> >>>>> communicate with condor_starter on execute machine
> >>>>> 0  -  Run Bytes Sent By Job
> >>>>> 528441  -  Run Bytes Received By Job
> >>>>> ----------------------------------------------------------------------- 
> >>>>>
> >>>>>
> >>>>> -
> >>>>>
> >>>>> Does anybody know, why these "Shadow exceptions" occur?
> >>>>>
> >>>>> Thomas
> >>>>
> >>>>
> >>>>
> >>>> 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>
> 
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>