CCP4i: Graphical User Interface
	Submitting jobs to batch queues

Introduction

Simple batch queues

More complicated batch queues

Specific batch queues

Condor

DQS

PBS

Sun GRID

Introduction

This document contains more detail on how to submit jobs to batch queues from ccp4i. The first sections give general information from a ccp4i point of view. Then there are more details on specific batch queues. The latter should be taken with a pinch of salt since our experience may be patchy. It is hoped that it will be helpful but you should of course check the queueing software documentation.

Simple batch queues

In the simplest cases, follow these steps:

1. In the Configure interface (find it under System Administration on right hand side of the main window) and in the "hardware resources" folder, add a batch queue.
2. Enter a name for your queue. This should be something that users will understand when they see it on a menu - it is not a computer command.
3. For versions of CCP4 after 5.0.2, there is a pull down menu to select the type of batch queue. This will also set a default command, see next point.
4. Enter the command appropriate to your system queueing software. This will typically be something like "qsub -cwd". ccp4i automatically adds the job command (something like "source /tmp/mdw/30_pdbset.def.com"). Do not add this yourself in the configure interface.
5. Higher up the Configure interface, find "Command to set up CCP4 (used by remote jobs)". By default, this is set to "setup_ccp4". Either create this script, or remove if from the Configure interface.
6. Save the Configure to the installation file if you want all users to see the option, or to your own home directory.
7. When you run a job, select "Run Remote/Batch/Later" and you will get a window which has "Remote" set as the default. Change this to "Batch". The content of the window should change so that you now see the name of the batch queue that you entered in the Configure window. Hit the Run button and hopefully everything runs. There is the possibility to add extra command line options which appear after the command entered in the Configure interface. In the example above, they would appear after "qsub -cwd".

ccp4i puts the files which are needed to run the job in the 'TEMPORARY' directory defined in the Directories and ProjectDir window. By default this is the same as CCP4_SCR. The most likely problem you will have is that the different machines (i.e. your local machine and the batch queue machine) do not look in the same place for 'TEMPORARY'.

More complicated batch queues

In some cases, it is not sufficient to send a CCP4 job directly to a batch queue, i.e. as an argument to a "qsub" command. Perhaps some administration of user options or directories is necessary. In this case, one can replace "qsub -cwd" defined in the Configure interface (see above) with "myqsub", where myqsub is a specially written script.

An example written for use with Condor is:

#!/bin/sh

#get intial dir in terms of yt tree
home=$YT_TREE

#create ccp4ish.sh
sh1="#!/bin/csh -f"
sh2="source /yt/usrloc/sfware/ccp4-4.2.1/include/ccp4.setup "$home"/ztmp"
sh3=${1+"$@"}

echo $sh1 >  ccp4ish.sh
echo $sh2 >> ccp4ish.sh
echo $sh3 >> ccp4ish.sh

chmod 755 ccp4ish.sh

## assign variables for condor script
uni="universe = vanilla"        
dir="initialdir = "$home
exe="executable = ccp4ish.sh"
out="output = out"
log="log    = log"
err="error = err"
que="queue"

## create condor script
echo $uni >  script.condor
echo $dir >> script.condor
echo $exe >> script.condor
echo $out >> script.condor
echo $log >> script.condor
echo $err >> script.condor
echo $que >> script.condor

## submit to condor
condor_submit script.condor

This handles, for example, the "yt-tree" directory structure used by Condor.

Specific batch queues

Condor

• Condor provides a job queueing mechanism, scheduling policy, priority scheme, resource monitoring, and resource management. Users submit their serial or parallel jobs to Condor, Condor places them into a queue, chooses when and where to run the jobs based upon a policy, carefully monitors their progress, and ultimately informs the user upon completion.
• Distributed under Condor Public License, a "very liberal license".
• Runs with heterogenous computing environment (so far, we've included Linux and Solaris, but more possible).
• Machines in the "pool" are identified as central manager (1 only), compute machine or submit-only machine.
• Applications can be compiled against Condor C library, but do not have to be.

DQS

• The Distributed Queueing System is designed as a management tool to aid in computational resource distribution across a network. DQS provides architecture transparency for both users and administration across a heterogeneous environment, allowing for seamless interaction for multiple architectures.
• At the time of writing (June 03) development of DQS appears to have stopped.

Open PBS

• The Portable Batch System (PBS) is a flexible batch queueing and workload management system. It operates on networked, multi-platform UNIX environments, including heterogeneous clusters of workstations, supercomputers, and massively parallel systems.
• There is a commercial version, PBS Pro, which is strongly pushed on the Open PBS web site.
• We have used PBS on an Alpha cluster without problems.

Sun Grid Engine

As far as I know, these two links give the same software. Presumably the second one includes support as well.

• Grid Engine is distributed resource management software for compute farms through to grid computing. It includes queue management.
• For versions of CCP4 after 5.0.2, select "Sun Grid Engine" from the pull down menu in Configure Interface (see above). If you don't, SGE will complain about "source" in the job command.
• We have SGE working on a computer cluster at Daresbury.
• At least one other group is using SGE for CCP4 jobs.