Finding OSG Locations¶
In this section, we will learn how to quickly submit multiple jobs simultaneously using HTCondor and we will visualize where these jobs run so we can get an idea of where and jobs are distributed on the Open Science Pool.
Gathering network information from the OSG¶
Now to create a submit file that will run in the OSG!
- Use the tutorial command to download the job submission files:
- Change into the
Hostname fetching code¶
The following Python script finds the ClassAd of the machine it's running on and finds a network identity that can be used to perform lookups:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
This script (
wn-geoip.py) is contained in the zipped archive (
wn-geoip.tar.gz) that is transferred to the job and unpacked by the job wrapper script
location-wrapper.sh. You will be using
location-wrapper.sh as your executable and
wn-geoip.tar.gz as an input file.
The submit file for this job,
scalingup.submit, is setup to specify these files and
submit 100 jobs simultaneously. It also uses the job's
process value to create unique output, error and log files for each of the job.
$ cat scalingup.submit # The following requirments ensure we land on compute nodes # which have all the dependencies (modules, so we can # module load python2.7) and avoid some machines where # GeoIP does not work (such as Kubernetes containers) requirements = OSG_OS_STRING == "RHEL 7" && HAS_MODULES && GLIDEIN_Gatekeeper =!= UNDEFINED # We need the job to run our executable script, with the # input.txt filename as an argument, and to transfer the # relevant input and output files: executable = location-wrapper.sh transfer_input_files = wn-geoip.tar.gz # We can specify unique filenames for each job by using # the job's 'process' value. error = job.$(Process).error output = job.$(Process).output log = job.$(Process).log # The below are good base requirements for first testing jobs on OSG, # if you don't have a good idea of memory and disk usage. request_cpus = 1 request_memory = 1 GB request_disk = 1 GB # Queue 100 jobs with the above specifications. queue 100
Submit this job using the
$ condor_submit scalingup.submit
Wait for the results. Remember, you can use
watch condor_q to monitor the status of your jobs.
Collating your results¶
Now that you have your results, it's time to summarize them.
Rather than inspecting each output file individually, you can use the
to print the results from all of your output files at once. If all of your output
files have the format
job.10.output), your command will
look something like this:
$ cat job.*.output
* is a wildcard so the above
cat command runs on all files that start with
job- and end in
Additionally, you can use
cat in combination with the
uniq commands to print only the unique results:
$ cat job.*.output | sort | uniq
Mapping your results¶
To visualize the locations of the machines that your jobs ran on, you will be using http://www.mapcustomizer.com/. Copy and paste the collated results into the text box that pops up when clicking on the 'Bulk Entry' button on the right-hand side. Where did your jobs run?