[Swift-user] Running first.swift remotely on NCSA

[Andriy Fedorov]

>> > By never, do you mean more than 1 minute or less?
>> >
>>
>> More than 5 minutes.
>
> That "never" enough.
>

You were EXACTLY right! It took 5 minutes 46 seconds, and it finished!
My fault, I guess, didn't wait long enough...

But it is just "echo"! Is this normal that it takes so long? When I do
the same test with WS GRAM and Fork, it completes in less than minute.


>> UC-GT4          echo_gt4_mpi    /bin/echo       INSTALLED
>> INTEL32::LINUX  GLOBUS::host_count="2:compute",jobType=mpi
>
> I'm a bit unsure whether that would work. I think that
> host_types="ia32-compute" has worked in the past on that site.
>

Well, it does work with "qsub":

[fedorov at tg-login1 GT_test]$ qsub -I -l nodes=1:compute
qsub: waiting for job 1724062.tg-master.uc.teragrid.org to start
qsub: job 1724062.tg-master.uc.teragrid.org ready

----------------------------------------
Begin PBS Prologue Fri Jun 20 16:50:24 CDT 2008
Job ID:         1724062.tg-master.uc.teragrid.org
Username:       fedorov
Group:          allocate
Nodes:          tg-c013
End PBS Prologue Fri Jun 20 16:50:24 CDT 2008
----------------------------------------
[fedorov at tg-c013 ~]$ logout

It also does work with pre-WS and WS GRAM on UC, when I use globusrun.
Why doesn't it work with Swift tc.data?

ia32-compute requests visualization nodes on UC, I need 64-bit compute
nodes (node names look like tg-c0*). I tried to use ia64-compute:

UC-GT4          echo_gt4_mpi    /bin/echo       INSTALLED
INTEL32::LINUX  GLOBUS::host_count="2:ia64-compute",jobType=mpi

but this doesn't finish at all (this time I waited for 10 minutes, to
be safe), and the job doesn't appear in the qsub.

Andrey