VMD can use
BioCoRE to
run external programs used by some plugins on different computers. This can be useful if your machine doesn't have that particular external program installed
or if you prefer to use a faster machine to get your results sooner.
Running jobs via BioCoRE does require some initial setup, such as to create
an account, join a project, and to create a job account, that tells BioCoRE how
to run jobs on the remote machine you want to use. Below are some instructions
for configuring BioCoRE to run jobs.
Logging in
To use the various BioCoRE enhancements, you must first login to BioCoRE. Further information on doing this is available on the
BioCoRE plugin help page.
Joining a Project
To use BioCoRE you must also be a member of a BioCoRE Project. Members of an
existing project might invite you to join, or you can create your own project.
Many public projects are available for you to join, however, users who
create public projects may not have enabled job management for those projects.
- To find out how to join an existing project, read about the BioCoRE utilities plugin.
- To find out how to create a new project, log in to BioCoRE via the
web and click on the "Create Project" link in the sidebar.
Telling BioCoRE about your remote computer account
To run a job via BioCoRE, you have to tell BioCoRE how to log in to the
machine where you want to run your job. You can do this by loging in to
BioCoRE via the web page, selecting the Job Management link in the sidebar,
and then clicking on the "Set up new account" button to tell BioCoRE about
your account.
BioCoRE will ask you to select an account type. Options include:
- Interactive login: Log in to any machine (thats accessible via SSH) and
run the job at the command line.
- Various supercomputer/cluster accounts: The other account types require you to have an account on the particular machine. If you do, BioCoRE will submit
your job to the queue on that machine
If you select "Interactive login", BioCoRE will ask for an account name and
description, the host name of the machine you want to log in to, and your
user name and password on that machine. BioCoRE does not store your user name
and password permanently. Instead, it sets up an SSH key so that it can log
in and run jobs on your behalf.
After clicking Continue, BioCoRE will attempt to connect to your machine and
set up the SSH login key. If its successful, it will test the connection,
and if that succeeds, you can immediately start using your account from VMD.
If you have a problem setting up your machine, the problem might be that due
to firewalls or other network configuration details, the BioCoRE server cannot
connect to your machine, or that there's a problem connecting via SSH.
For troubleshooting help, please contact the BioCoRE developers via the
BioCoRE Help (Public) project, or send email to biocore@ks.uiuc.edu.
Troubleshooting
Why do I never get any results back from my job?
BioCoRE jobs are submitted in the background, so the plugin cannot
immediately deterimine whether the job has started. BioCoRE will later
inform VMD when the job has finished successfully, or if it fails, but
in some cases, the job will fail silently and BioCoRE won't detect it.
A few of the reasons your job might fail include:
- The external program that the plugin is trying to run is not properly
installed on the remote machine. In this case, BioCoRE will usually detect
that the job has finished, but no results will be returned. Logging in to
the remote machine manually and running test jobs may help track down the
problem. You can find the script that BioCoRE tried to run by looking in
the "Remote work directory" specified when the job was submitted.
- BioCoRE expects the external program to be in your Unix path by default.
If it can't find the program, the job will terminate with no output. If the
program is not in the path, you can fill in an absolute path in the
Command field of the job form.
- Sometimes, the external program may run, but the output files will not be
generated correctly. In this case, BioCoRE will still bring back files with
the expected names, but the files will be empty. This may result from a
lack of disk space or a permission problem on the remote machine, or some
problem with running the program. Look in the work directory on the remote
machine to try to determine the source of the problem.
Other resources: