Running Jobs > Condor > Condor on Public Site
Currently, Condor is available to public site computers, which means that if you do not have accounts on Altair and Blaze, you still can run your jobs on public site machines including roughly 500 computers on the campus for free.
Follow the instructions below to know more about how to compile, how to submit your jobs on public site machines.
- Check Condor Status
- Setup Condor Environments
- Compile Jobs
- Prepare Submit File
- Submit Jobs
- See Outputs
This tutorial only gives a minimal content that you should know before running your jobs. However, you need to read Condor official guide if you really want to do something serious.
Check Condor Status
There are two ways to check the current Condor status on public site machines. You can either directly check it in Windows operating systems or check it when you log onto LEAF.
In Windows
Click "Start" button and select "Run...". Type "cmd" to start a command window.
In the command window, type "condor_status".
You will get similar result like this:
On LEAF
Use certain SSH tools to log onto Leaf. If you do not know how to do that, please read more about it here. After you log onto LEAF, if you directly type "condor_status" in command line, you will get following error:
| -bash-3.2$ condor_status -bash: condor_status: command not found |
The reason is that you need to setup Condor environments first before you run any Condor commands, which is introduced in the following section(See Setup Condor Environments).
After successfully setting up your Condor environments, type "condor_status" and you will get similar result like this:
Setup Condor Environments
On LEAF, before you run any Condor commands, you need to set up your Condor environments first.
Type ". /Projects/condor/32/condor/condor.sh" in command line after you logging onto LEAF.
| -bash-3.2$ . /Projects/condor/32/condor/condor.sh |
Compile Jobs
Before you compile and run your jobs, you need to decide which Universe to choose, see Official Knowledge Base.
If you choose Standard Universe, you must re-compile your jobs using "condor_compile" command. Using condor_compile is pretty straight forward. Just precede the compilation command with condor_compile. e.g. in order to compile a program foo.cpp using g++ and to link with condor, do:
| condor_compile g++ helloworld.cpp -o helloworld |
See more detail about "condor_compile".
Prepare Submit File
Here is an example of submit file for Condor on Public sites. One advantage of Condor on Public sites is that you can specify programs or jobs to run on Windows operating systems.
universe = vanilla requirements = (OpSys=="WINNT51") && (Arch=="INTEL") should_transfer_files = YES when_to_transfer_output = ON_EXIT_OR_EVICT executable = condor_test.exe transfer_input_files = data.dat transfer_output_files = data.dat output = hello.out error = hello.err log = hello.log
More about submit files, please see here for more details.
Submit Jobs
Currently, you can only submit your Condor jobs through LEAF unless you have advanced accounts.
- On LEAF, please place your job executable file and submit file under "/tmp" directory.
- Submit your job in "/tmp" directory.
Remember, this is the only way to run your jobs successfully. If you directly submit jobs from your home directory like "/home/lih307/", your jobs will probably be held all the time.
Attention: Please check current hostname of LEAF where you submit your jobs. It is very important for you to remember this hostname for further usage like examining the log files and output files. (See next section)
You could type "hostname" command to see where you are.
| -bash-3.2$ hostname leaf29.cc.lehigh.edu |
See Outputs
Every time you log onto LEAF, the server cannot guarantee that you will get on the SAME server as last time due to node balancing. Therefore, since you would submit Condor jobs under "/tmp" directory, which is a local directory for a particular LEAF node, you will probably cannot find your log files and output files in current LEAF node.
For example, you firstly submit your jobs on LEAF node 31 and later you logout. Next time, you log onto LEAF and get on node 29. Thus, you cannot find your log files and output files on node 29. You must switch to node 31 to seek your files.
Use "lsh" command to switch between LEAF nodes. Say, you want to log onto LEAF node 31, just type "lsh leaf31".
| -bash-3.2$ lsh leaf31 lih307@leaf31's password: |
Last Modified: May 29, 2009