LHCb software available on AFS at RAL

LHCb software available on the RAL afs system

In the RAL PPD systems, the LHCb software is mounted in the directory /opt/ppd/lhcb/lhcb/scripts. DaVinci, Panoramix and Ganga have been tested to be working on the heplnx machines. To use the LHCb software which is installed at RAL, you need one of the following :

Usage at RAL PPD : The instructions below also work at the RAL PPD ( tier 2 / 3) computing facility. It is not yet clear to me how to access the Tier 1 resources from here or to submit jobs to these machines (linux.pp.rl.ac.uk).
Usage anywhere else : You need afs mounted on your system. You will also need to get AFS tokens at the cell rl.ac.uk. You can get these by applying to the RAL CSF as above.

Usage of the software : You need to source the file

 /opt/ppd/lhcb/lhcb/scripts/lhcbsetup.csh  for (t)csh shells
 /opt/ppd/lhcb/lhcb/scripts/lhcbsetup.sh   for (z)sh shells

Next, depending on what package you want to run, you need to get the environment of that package. For example to use DaVinci, type

$ setenvDaVinci

and for Brunel, type

$ setenvBrunel

You will be asked for the version you want to be used. You can pick one out of the list provided. Then, you are up and running. Look at the guides of the respective packages to find out how to use them.

Note :

MYSITEROOT = /opt/ppd/lhcb/lhcb
This should have been setup by the script sourced above.
Remember the default user area is defined in $CMTPATH to be $HOME/cmtuser. This is set in the script lhcbsetup.(c)sh. You will need to change it if you want to work in someother area.
The "getpack" command works with ssh authentication. The way to set it up is as follows :

Set the following environment variables :


CVS_RSH = ssh

Use the command : getpack -f ssh < package > < version >
For those who want to get rid of the repeated annoying prompts for the password, the website at CERN on cvs access should help.

The queue system is the PBS batch system . The job submission and checking commands are qsub, qstat. For more information, see the linked page about the PBS batch system .

The jobs submitted using "qsub" end up in the queue "prod". This is a routing queue which shunts the jobs to sl3p or sl3m depending on the memory requirements.
Note that there is also a high priority exprese queue "express". Only one job at a time can be run for a given user in this queue.
All other queues are out of bounds for users who want to perform direct job submission on the lcgui machines using qsub. They are only available for jobs submitted using the grid, by the respective Virtual Organisations.
If your job crashes due to too little memory allotted you can ask for more at time of job submission, with the command (for example) "qsub -l pmem=1000mb". This problem is hinted by an abnormal termination of your job with something like the following lines in your output file clearly showing that the requested resources are much less than needed.

*    Resources Requested
*
*    cput=24:00:00,neednodes=fast,pcput=24:00:00,pmem=500mb,walltime=56:00:00
*
*    Resources Used
*
*    cput=06:35:19,mem=575556kb,vmem=756096kb,walltime=09:27:37

For those who have CSF (central simulation facility) accounts at RAL, you will also have access to a scratch disk (/scratch/lhcb/< username >) with some large area (220GB currently). Usage of this area is on a first come - first served basis. Clean up scripts delete all files older than 60 days and ensure that the file system is not much more than 80% full. This directory is also cross mounted on the farm of nodes for access by your jobs.
For information about other space available, the following web page may help

For those who want to install the software in their own computers / areas, the following pointers may come of use.

You need to set the environment variables SITEROOT (where you actually install the packages) and CMTCONFIG (the type of operating system it is : For example, rh73_gcc323, slc3_ia32_gcc323, win32_vc71). Then, follow the instructions given on the following web page to actually install the software.

If you are planning to install / support all the LHCb software, the following may be useful :

Copy the script /opt/ppd/lhcb/lhcb/scripts/lhcbsetup.(c)sh and edit the MYSITEROOT and CMTCONFIG (and whatever else you need). You can source this file to get the environment variables setup.
Remember you need also to source $MYSITEROOT/scripts/ExtCMT.(c)sh. This will set up most of the aliases for the LHCb package environments.
In case of Ganga, you will need to look at the directory structure to set up the GangaEnv.
In case you have HbookCNV problems (for DaVinci?), it helps to find out the location of HBookCNV and compile it before trying to compile DaVinci.
I have had complaints from cmt.exe about the following missing directories. Creating the necessary directories and symbolic links to (at RAL) $LHCBHOME/lhcb solved the problem. This seems to arise from the difference in the directory structure of LHCb software at CERN and the structure as installed using the installation package.

$LHCBHOME/software/releases
$LHCBHOME/software/NEW
$LHCBHOME/software/DEV

The aim is to provide the complete set of software comprising the various LHCb packages as will be needed by a generic user. The list of software available contains most of the versions (starting with the DC04 compatible ones) as in the LHCb Computing Webpages. If any version is missing, please do contact me at the email address given below. Of course, the setenv< package > command will give you the list of available versions for the given package.

For more information : contact Raja Nandakumar ( R dot Nandakumar at rl dot ac dot uk )