Page History

Turn Off History

It may be a challenge to get your users to read the Condor manual, so you can give them information via the MOTD and the fortune program.

If you do not know how to create a fortune file, a nice tutorial may be found here.

Below is the necessary data to put into your text file to generate the fortune. Another good idea is to pull this information from the wiki in a cron job. There are scripts attached to this wiki page that may help.

Condor Tip #1:

To see which machines are willing to run jobs now, run:

condor_status -available
%
Condor Tip #2:

To see which machines are currently running jobs, run:

condor_status -run
%
Condor Tip #3:

To list the machine ClassAds for all machines in the pool, run:

condor_status -l
%
Condor Tip #4:

You can allow your job to prefer machines with particular attributes
by adding a 'Rank' command and expression to your submit file.
%
Condor Tip #5:

If your standard universe job does not need remote I/O, you may turn
it off by adding the following to your submit file:

want_remote_io = False
%
Condor Tip #6:

If you'd like to check on your job while it is running, and you are
using vanilla, java, local, and/or parallel universes, run:

condor_ssh_to_job <job number>
%
Condor Tip #7:

If you'd like to use system environment variables in your submit file,
you may specify them using the following:

$ENV{EnvironmentVariableName}
%
Condor Tip #8:

If you'd like to use Machine Ad attributes in your submit file so that
you can change behavior based on the machine that will execute your
job, use the Machine Ad substitution macro:

$${MachineAdAttribute}
%
Condor Tip #9:

If you'd like a list of all users with jobs submitted, run:

condor_status -submitters
%
Condor Tip #10:

To see the list of jobs that you have currently in the submit queue,
run:

condor_q <YourUserName>
%
Condor Tip #11:

If all you want to do is have Condor run an executable, and do not
need any special features, you probably just need the 'vanilla'
universe.
%
Condor Tip #12:

Why is your job not running? To get more information, run:

condor_q -better-analyze <job number>

For Condor versions 7.5 and higher, use instead:

condor_q -analyze <job number>
%
Condor Tip #13:

Need to see a list of completed jobs? Run:

condor_history
%
Condor Tip #14:

Can't quite remember how a particular Condor command works? Use
the -help option with the command to get a quick reminder of the
command's usage.
%
Condor Tip #15:

If you are interested in running Java applications in the Java
Universe and want to see which machines in the Condor pool
support this, run:

condor_status -java
%
Condor Tip #16:

Do you have jobs that depend on other jobs to run first? If so,
check out the Condor tool "DAGMan" in the manual.
%
Condor Tip #17:

If you are concerned about how many jobs may be running at any
time with your DAGMan submission, you can limit the number of
simultaneous jobs with the -maxjobs option.
%
Condor Tip #18:

Do you need to visualize your DAG that you created with DAGMan?
DAGMan can output files for use with the 'dot' visualizer. See
the User Manual for specifics.
%
Condor Tip #19:

Which version of Condor is running on your system? Find out by running:

condor_version
%
Condor Tip #20:

All Condor commands have traditional UNIX man pages. To get help, run:

man <Condor command name>
%
Condor Tip #21:

Need to sync files back from your Vanilla or Parallel Universe jobs before they've actually finished? Use rsync and condor_ssh_to_job.

rsync -v -e condor_ssh_to_job <job number>:<remote filename> <local directory>

Attachments: