Re: [Condor-users] Condor match-making problem

[Jaime Frey <jfrey@xxxxxxxxxxx>]

On Sep 2, 2005, at 12:15 PM, duane waktu wrote:

> My bad. Sorry about the silly problem. Anyway, I was able to run  
> the cmd without any problem. This is what I got displayed after that:
>
> ==========================================================
> Name                           OpSys          Arch   State       
> Activity   LoadAv Mem   ActvtyTime
>
> abc                               [?????????] [????] [????????]  
> [???]  [??]   [Unknown]
> condor_manager.ttk.com LINUX          INTEL  Unclaimed  Idle        
> 0.000  2048  0+00:00:03
> abc.ttk.com                   LINUX          INTEL  Unclaimed   
> Idle       0.000  2048[?????]
> def.ttk.com                  LINUX          INTEL  Unclaimed   
> Idle       0.120  2000[?????]
> ...
>                    (Omitted 1 malformed ads in computed attribute  
> totals)
> ==========================================================
>
> Basically 'abc' (the machine name that comes up because of running  
> condor_advertise) is the same machine that is already running in  
> the Condor pool 'abc.ttk.com'. My question is it is alright to have  
> all the fields of machine 'abc' filled with '???' ?

You can ignore the question marks. condor_status is looking for  
certain attributes that you're not placing in the ad. The grid  
universe doesn't need them, so leaving them out only affects with  
condor_status displays.

> Another question, when I used this submit file
> =============================================
>  universe       = grid
>  grid_type     = gt4
>  executable   = /bin/hostname
>  log              = test.log
>  output        = test.output_$(Cluster)
>  error           = test.error_$(Cluster)
>  globusscheduler = $$(gatekeeper_url)
>  requirements    = TARGET.gatekeeper_url =!= UNDEFINED
>  jobmanager_type = Fork
>  should_transfer_files = YES
>  when_to_transfer_output = ON_EXIT
>  queue
> ==========================================
>
> the job didn't get executed and failed with this error in  
> GridmanagerLog file:
>
> ---------------------------------------------------------------------- 
> -------------------------------------------------------------
> 9/2 13:02:14 [20664] DaemonCore: Command received via UDP from host  
> <host_ip:37272>
> 9/2 13:02:14 [20664] DaemonCore: received command 60000  
> (DC_RAISESIGNAL), calling handler (HandleSigCommand())
> 9/2 13:02:14 [20664] (99.0) doEvaluateState called: gmState  
> GM_DELEGATE_PROXY, globusState 32
> 9/2 13:02:19 [20664] No jobs left, shutting down
> 9/2 13:02:19 [20664] Got SIGTERM. Performing graceful shutdown.
> 9/2 13:02:19 [20664] **** condor_gridmanager (condor_GRIDMANAGER)  
> EXITING WITH STATUS 0
> ---------------------------------------------------------------------- 
> -------------------------------------------------------------

Do you get this same error if you replace $$(gatekeeper_url) with the  
URL for the machine?

+----------------------------------+---------------------------------+
|            Jaime Frey            |  Public Split on Whether        |
|        jfrey@xxxxxxxxxxx         |  Bush Is a Divider              |
|  http://www.cs.wisc.edu/~jfrey/  |         -- CNN Scrolling Banner |
+----------------------------------+---------------------------------+