|
Neutrino Ettore Majorana Observatory | 
|
SuperNEMO on the GRID: supernemo.vo.eu-egee.org |
||
|
|
|
|
|
|
|
|
|
Neutrino Ettore Majorana Observatory |
|
SuperNEMO on the GRID: supernemo.vo.eu-egee.org |
||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The LHC Computing Grid (LCG) is being developed mainly to satisfy the data transfers, storage and computing needs of the four big LHC experiments. However, development and use of the worldwide structure is strongly encouraged by research communities both within or out of the High Energy Physics community.
Users of the LCG are organised in Virtual Organisations (VO). This is the computing equivalent of the physical "Collaboration" or "Experiment". ATLAS, CMS, ALICE, LHCb are the main example of VOs, and a large number of HEP collaborations did follow suite and setup their own VO (CDF, Dzero, ZEUS, etc...)
A VO meant for the physicists of the SuperNEMO collaboration has been setup as well. Its full name is: supernemo.vo.eu-egee.org. The VO is open to all physicists of the SuperNEMO collaboration, subject to its Acceptable Usage Policy, which was drafted as part of the procedure to establish the VO within the global EGEE (Enabling Grids for E-science) framework:
This Acceptable Use Policy applies to all members of supernemo.vo.eu-egee.org
Virtual Organisation, hereafter referred to as the VO, with reference to use of the
LCG/EGEE Grid infrastructure, hereafter referred to as the Grid. The SuperNEMO
collaboration owns and gives authority to this policy. The goal of the VO is to
enable the usage of Grid resources for users of SuperNEMO. Members and Managers
of the VO agree to be bound by the Grid Acceptable Usage Rules, VO Security Policy
and other relevant Grid Policies, and to use the Grid only in the furtherance of
the stated goal of the VO.
A SuperNEMO physicist willing to access the LCG resources will have to undergo a few preliminary steps involving firstly identification and certification to get the appropriate credentials needed to use the resources. If you want to join the LCG as a member of the supernemo VO, follow the steps outlined below:
Every Grid user is identified through a personal certificate issued centrally. In the UK, this is done by the UK e-Science Certification Authority. The details of the registration process are explained on their main page.
Visit their main page
with a browser that supports certificates (Firefox, Mozilla - At the time of writing,
Safari doesn't). You will need to click on Request a
Certificate and then ask for a User Certificate.
You will be
asked for a Registration Authority (RA) which will verify your
identity later.
From the drop down list, choose UCL EISD.
After you have sent the registration request, the RA will
contact you directly with details of the required proofs of identity.
You will have to present your proof of identity in person.
This whole process used to take several days or more, but it looks like now it's a speedy procedure and you can get your certificate in a matter of a day or two. Please note that you cannot do anything else until you have the certificate ready!
As soon as you get the certificate, install it in your browser. Several sites that have to do with Grid operations, support and monitoring require you to authenticate via your certificate (first example will be your next step: Register with the supernemo VO). This happens behind the scenes once your certificate is imported in your browser. You can import that certificate in any number of browsers on different machines.
Once your certificate is installed in your browser, you can register with the supernemo.vo.eu-egee.org Virtual Organisation (VO).
Follow the instructions on the registration website. An automated email will be sent to the supernemo.vo.eu-egee.org VO admins informing them of your request. You will hear from them directly once your request has been granted. Once your certificate is on the live VO membership list, you're ready to go.
To submit jobs to LCG, you need to be logged in to a computer with the necessary client software installed, i.e. a UI. At UCL HEP, the group Linux cluster desktop machines can now be used for this. You just have to run a command to set up your environment first. If you use the bash or Bourne shell, do:
source /grid/LCG-share/UI/glite/external/etc/profile.d/grid-env.sh
If you use C shell or tcsh do:
source /grid/LCG-share/UI/glite/external/etc/profile.d/grid-env.csh
If you are brave enough, you can also install a suitable UI on your machine or laptop. However, as long as you have ssh access to our cluster, you won't need your own UI (and the hassle deriving by installing, upgrading, maintaining and troubleshooting it!)
After importing the certificate into your browser, you'll need to
convert it into a form that the LCG command line tools can use. Back
up the certificate from the browser into a file, say
gridcert.p12. This is a file in PKCS12 format,
which must be converted into separate certificate and key files:
follow these steps in your home directory:
$ cd ~
$ mkdir -p .globus
$ openssl pkcs12 -nocerts -in gridcert.p12 -out .globus/userkey.pem
$ openssl pkcs12 -clcerts -nokeys -in gridcert.p12 -out .globus/usercert.pem
$ chmod 444 .globus/usercert.pem
$ chmod 400 .globus/userkey.pem
$ chmod 400 gridcert.p12
The key export command will ask for the password that the browser copy is protected with. Then you can set a new password for the user key. This is the password you will need EVERY TIME you wish to work on the grid.
To access LCG resource you have to authenticate first. After setting up the LCG environment by running the environment setup script as described above, run the following command:
voms-proxy-init --voms supernemo.vo.eu-egee.org
When prompted, supply the password you created when exporting the key from your certificate. This operation creates a proxy certificate (your LCG credentials), a copy of which will be sent along with every job that is submitted to remote hosts. The remote host will recongise your credentials and run your job. The default validity of the proxy is 12 hours. Please see the man pages for voms-proxi-init for the option to specify a different duration.
To run on LCG, a job needs to be specified in a JDL (Job
Description Language) file. This can then be submitted using the
command edg-job-submit --vo supernemo.vo.eu-egee.org myjob.jdl .
The best way to get started is following the
LCG2 User Scenario
while additional information can be found in chapter 6 of the
LCG2 User Guide.
There are not yet many sites supporting supernemo.vo.eu-egee.org, but this will change rapidly, as the UKI region (UK Ireland) Has accepted my request to support the VO and has passed it to all sites. The same request has been sent out by ther London Tier 2 management, asking London sites specifically to add support. In order to find out what Computing Element (CE) support the VO, use the following command after authenticating:
lcg-infosites --vo supernemo.vo.eu-egee.org ce
The CE is a batch farm (or an aggregate of farms) somewhere on the Grid. Directing the workloads to suitable computing resources is handled by the Resource Brokers (RB). These accept sob sumbissions from the end user and direct them to appropriate CEs. While the RB is (most of the times) transparent to the user, in some circumstances you might want to direct your job to a CE of your choice. In order to do so, you need to pick one of the CEs returned by the "lcg-infosites" commmand above and add that to the --resource option to the submission comamnd, e.g.:
glite-wms-job-submit -a --resource lcg-ce01.hep.ucl.ac.uk:2119/jobmanager-lcgpbs-lcgsupernemo --vo supernemo.vo.eu-egee.org myjob.jdl
In this example, lcg-ce01.hep.ucl.ac.uk is the gateway to our own farm for Grid workloads. You'll probably want to choose something better than that ;-) But it is a very useful way to debug your jobs (if needed), as I am at hand for digging into the farm, should your jobs be eaten up without leaving any trace, which is not uncommon still...