The Module and Queueing Systems¶
The Module System¶
There are hundreds of programs and software suites that people might want to use on the server. Whilst everyone is welcome to install each one they use for themselves, it’s more sensible to make the most common packages available for everyone. Further, different pieces of software have different dependencies, which may in some cases disagree with each other – for instance, whether to use Python 2.x or 3.x.
One way to resolve this is to use a module system, from which different software packages can be loaded and unloaded individually. This is not the same as installing the programs - they were there all along - it’s simply making them available in your path – a generic term for all of the programs and libraries your system is immediately aware of, without having to be shown where they are.
As an example, let’s load up the module for prodigal, a program for finding ORFs in prokaryotic genome sequences.
# Loading a module
module load prodigal
ml prodigal
Now if we issue the command prodigal the program loads straight from the command line. Note that module can be shortened to ml, and if you just put a module name it is assumed you want to load it.
It’s also possible to unload modules:
# Unloading a module
module unload prodigal
# Unload all modules
ml purge
Now if you try to run prodigal, it won’t recognise the command.
There are also commands to show which modules you have loaded, and which modules are available to load. If you want to run a particular piece of software and it isn’t available, let me know and I can see about making it available for you.
# What have I loaded?
module list
ml
# What can I load?
ml avail
Finally, if you want to search for a particular piece of software by name (or to find out the correct name, given that module names are case-sensitive), there is the command spider:
# Search for a module
ml spider blast
The SGE Queueing System¶
Many people have access to morgan and even more to euler. If everyone ran whatever program they liked, whenever they liked, the system would soon grind to a halt as it tried to manage the limited resources between all the users. To prevent this, and to ensure fair usage of the server, we run a queueing system that automatically manages which jobs are run when. Any program that will use either more than 1 core or thread, more than a few GB of RAM, or will run for longer than a few minutes, should be placed in the queue.
To correctly submit a job to the queue on morgan, it’s usually easiest to write a short shell script based on a template.
# Look at the template
less /science/teaching/submit.sh
#$ -cwd # run in current directory
#$ -S /bin/bash # interpreting shell for the job
#$ -N job1 # name of the job
#$ -V # .bashrc is read in all nodes
#$ -pe smp 10 # number of threads to be reserved
#$ -l h_vmem=16G # memory required
#$ -e error.log # error file
#$ -o out.log # output file
#$ -m bea # send an email at the beginning, end and if aborted
#$ -M yourmail@ethz.ch
# Insert your commands here
echo 'Hello World!'
The first few lines, beginning with #$, define the parameters for your job. The commands you want to run then appear below, and you can include as many as you like, one per line, which will run in succession.
When the script is ready, you will need the following commands:
# Submit the job to the queue
qsub submit.sh
# Check the status of your jobs
qstat
# Check the status of all jobs
qstat -u "*"
# Remove a job from the queue
qdel jobid
The LSF Queuing System¶
If you are working on euler then it uses a different queueing system and a submission script looks more like this:
#!/bin/bash
#BSUB -n 10 # number of threads
#BSUB -W 1440 # estimated time to run
#BSUB -R "rusage[mem=2000, scratch=2000]" # memory and disk space needed
#BSUB -e error.log # error file
#BSUB -o out.log # output file
#BSUB -u yourmail@ethz.ch # specify your email address
#BSUB -B # send email when job starts
#BSUB -N # send email when job ends
# Insert your commands here
echo 'Hello World!'
Then the equivalent commands:
# Submit the job to the queue
bsub < submit_lsf.sh
# Check the status of your jobs
bjobs
# Remove a job from the queue
bkill jobid
Exercises¶
Copy the submit.sh script to your home directory.
Load the ‘prodigal’ module and find out the program options
Change the ‘echo’ line to load the module for prodigal and then run the program on the E. coli genome.
You shouldn’t need more than 8 slots or 1GB of memory per slot.
When the job is finished, look at the output files for yourself!