Re: [Condor-users] Flocking

[Todd Tannenbaum <tannenba@xxxxxxxxxxx>]

georgios skikos wrote:
> Hi all,
>
> Im trying to do flocking on VMware using condor 6.8.5. i followed the 
> instructions which is on condor web page. All the deamons in both pc are 
> running. I m using the command condor_submit -pool "name" -name "schedd" 
> submitfile.sub" and the command: condor_submit -remote "name" 
> -pool"name" submitfile.sub. The results are the same:
>
> Submitting job(s)
> ERROR: Failed to set Owner="Administrator" for job 42.0 (0)
>
> ERROR: Failed to queue job.
>
> Any ideas?

Without thinking hard about it, a quick idea:

Usually the above error happens when condor_submit tells the 
condor_schedd daemons "I am user X", but the condor_schedd authenticates 
the connection (via NTML/LanManager by default on Win32) and thinks that 
condor_submit is being run by user Y.  Thus it throws an error cuz 
something fishy is going on.

Try adding the following to your job submit file:
   +Owner = UNDEFINED
and then submit the job again.  This will tell the condor_schedd to set 
the owner to whoever it authenticates the connection as.

Another idea:  don't run condor_submit as user "administrator" or "local 
system".  Run it as a "normal user" (but one in the administrator group 
is ok).  This "normal user" should have an account on both machines with 
the same login/password.

Hope the above helps,
Todd