[torqueusers] Re: pbs_iff and qsub wrapper problems

Doug Renfrew renfrew at email.unc.edu
Tue Mar 21 15:40:57 MST 2006


Okay, so I kind of figured out my first question, but not entirely.
When everything was unpacked on the client, pbs_iff and pbs_rcp were
not setuid root. I had assumed they would be cause that is the way
they were installed on the server. This was fixed by doing

#chmod u+s pbs_iff
#chmod u+s pbs_rcp

Now I can use qstat to see job status from the compute nodes with no
errors. However when I try to submit a job I get the following error.

qsub: Bad UID for job execution

The second question in the previous email still stands. Again thanks
in advance for any help.

Doug

On 3/21/06, Doug Renfrew <renfrew at email.unc.edu> wrote:
> Hi
>
> I am attempting to setup a small cluster (~20) of machines for my lab.
> I am having a couple of problems. Right now I have a machine setup as
> the server and a machine setup as a compute node. I am able to submit
> jobs from the machine running pbs_server; they run on the compute node
> and everything seems to work fine.
>
> Problem 1:
> I want to be able to submit jobs from the compute nodes. I added
>
> ALLOWCOMPUTEHOSTSUBMIT true
>
> to my /usr/spool/PBS/torque.cfg file and restarted pbs_server. However
> when I try to submit a job from one of the compute nodes I get the
> following error.
>
> pbs_iff: file not setuid root, likely misconfigured
> No Permission.
> qsub: cannot connect to server [correct server name] (errno=15007)
>
> not quite sure what is going on. I assume that the problem could be a
> number of things. The firewall on both machines is turned off so I
> don't think that it is it. The server name displayed is the correct
> one so I don't believe that is the problem.
>
> Do I need to restart the pbs_mom on the compute node after I restarted
> the pbs_server on the server? Did I misconfigure something (I did a
> make packages after make install and then unpacked the client and mom
> packages on the compute node)?
>
> Problem 2:
> My university has a cluster of machines setup using the LSF system.
> Most of my lab is familiar with the setup and I wanted to try to
> replicate it as much as possible. The way it is configured (I am not
> sure if it the default behavior) allows you to run a job with out a
> job script. For example if I wanted to run the following command line
>
> #a.out -x abc -y efg
>
> I would just type
>
> #bsub -q batch a.out -x abc -y efg
>
> Is there a way to replicate this behavior with Torque? I suspect that
> this is possible using the "qsub wrapper" scripts but I am not sure
> what it would look like?
>
> I am using torque 2.0.0p8 with the pbs_sched scheduler. As is probably
> evident I am a torque newbie, thanks in advance.
>
>
> Doug
> --
> ---------------------------------------------
> P. Douglas Renfrew
> Graduate Student
> Molecular and Cellular Biophysics Program
> Dept. Biochemistry and Biophysics
> Unv. of North Carolina at Chapel Hill
> ---------------------------------------------
>


--
---------------------------------------------
P. Douglas Renfrew
Graduate Student
Molecular and Cellular Biophysics Program
Dept. Biochemistry and Biophysics
Unv. of North Carolina at Chapel Hill
---------------------------------------------


More information about the torqueusers mailing list