Computing

Policies

LCLS users are responsible for complying with the data management and curation policies of their home institutions and funding agents and authorities. To enhance the scientific productivity of the LCLS user community, LCLS supplies on-site disk, tape and compute resources for prompt analysis of LCLS data, and software to access those resources with published durations for the retention of data. Compute resources are preferentially allocated to recent and running experiments.

Getting an Account

You will need a valid SLAC UNIX account in order to use the LCLS computing system. The instructions for getting a SLAC UNIX account are here.

Getting Access to the System

You can get into the LCLS photon computing system by ssh'ing to:

psexport.slac.stanford.edu

Note that, from within SLAC, you can directly connect to the bastion hosts (pslogin) without going through psexport. The SLAC wireless visitor network is not considered part of SLAC so you'll need to go through psexport when using your laptop on-site.

From psexport or pslogin you can then reach the analysis nodes (see below). From psexport you can also move data files in and out of the system.

Each control room has a number of nodes for local login. These nodes have access to the Internet and are named psusr<id>.

The controls and DAQ nodes used for operating an instrument work in kiosk mode so you don't need a personal account to run an experiment from the control room. Remote access to these nodes is not allowed for normal users.

Data Management

LCLS provides space for all your experiment's data at no cost for you. This includes the raw data from the detectors as well as the data derived from your analysis. Your raw data are available as XTC files or, on demand, as HDF5 files. The path to the experimental data is:

/reg/d/psdm/<instrument>/<experiment>

The LCLS data policy is described here. The tools for managing files are described here.

Data Export

You can use the psexport nodes for copying your data. The recommended tools for exporting the data offsite are bbcp and Globus Online. The former, bbcp, is slightly simpler to setup. On the other hand Globus Online is more efficient when transferring large amount of data because it babysits the overall process by, for example, automatically restarting a failed or stalled transfer. The performance of the two tools is very similar.

All control rooms and the overflow room in FEH have one or more taps on the Visitor Data Network. These taps can be used to transfer data to a laptop or a storage device. These devices will be automatically assigned an IP address through DHCP as soon as they are connected to the network tap.

There is a web interface to the experimental data accessible via

https://pswww.slac.stanford.edu/apps/explorer

The web interface also allows you to generate file lists that can be fed to the tool you use to export the data from SLAC to your home institution.

Running the Analysis

The analysis framework is documented in the Data Analysis page. This section describes the nodes which are available for running the analysis.

Interactive Pools

In order to get access to the interactive nodes, connect to the addresses psananeh or psanafeh. A load-balancing mechanism will connect you to the least loaded of the nodes in the pool:

ssh psananeh
ssh psanafeh

Each pool is currently made of six servers with the following general specifications:

• 8-cores, Opteron 2384, 16GB, diskless, 10Gb/s

Each node in the interactive pools has one single user Matlab license. You can find which nodes in the pool have a Matlab license available by running the following command on any of the psana nodes:

/reg/common/package/scripts/matlic

The current Matlab version is 2012a:

/reg/common/package/matlab/r2012a/bin/matlab

Batch Farm

There are batch farms located in the NEH and FEH. Depending on your data access you may need to submit jobs to a specific farm. This can be accomplished by submitting to the appropriate LSF batch queue. Refer to the table below. Multi-core OpenMPI jobs should be run in either the psnehmpiq or psfehmpiq batch queue, see the following section on "Submitting OpenMPI Batch Jobs". Simulation jobs should be submitted to the low priority queues psfehidle and psfehidle.

You can find more LCLS specific information about LSF in this PDF file. For a more detailed description and more LSF commands, please see:

http://www.slac.stanford.edu/comp/unix/unix-hpc.html

The batch farm is made of eighty servers with the following general specifications:

• 12 cores, Xeon X5675, 24GB memory, 500GB disk, QDR IB

Submitting Batch Jobs

Login first to pslogin (from SLAC) or to psexport (from anywhere). From there you can submit a job with the following command:

bsub -q psnehq -o <output file name> <job_script_command>

For example:

bsub -q psnehq -o ~/output/job.out my_program

This will submit a job (my_program) to the queue psnehq and write its output to a file named ~/output/job.out. You may check on the status of your jobs using the bjobs command.

Submitting OpenMPI Batch Jobs

The RedHat supplied OpenMPI packages are installed on pslogin, psexport and all of the psana batch servers.

The system default has been set to the current version as supplied by RedHat.

$ mpi-selector --query
default:openmpi-1.4-gcc-x86_64
level:system

Your environment should be set up to use this version (unless you have used RedHat's mpi-selector script, or your login scripts, to override the default). You can check to see if your PATH is correct by issuing the command which mpirun. Currently, this should return /usr/lib64/openmpi/1.4-gcc/bin/mpirun. Future updates to the MPI version may change the exact details of this path.

In addition, your LD_LIBRARY_PATH }}should include {{/usr/lib64/openmpi/1.4-gcc/lib (or something similar).

For notes on compiling examples, please see:

http://www.slac.stanford.edu/comp/unix/farm/mpi.html 

The following are examples of how to submit OpenMPI jobs to the PCDS psnehmpiq batch queue:

bsub -q psnehmpiq -a mympi -n 32 -o ~/output/%J.out ~/bin/hello

Will submit an OpenMPI job (-a mympi) requesting 32 processors (-n 32) to the psnehmpiq batch queue (-q psnehmpiq).

bsub -q psfehmpiq -a mympi -n 16 -R "span[ptile=1]" -o ~/output/%J.out ~/bin/hello

Will submit an OpenMPI job (-a mympi) requesting 16 processors (-n 16) spanned as one processor per host (-R "span[ptile=1]") to the psfehmpiq batch queue (-q psfehmpiq).

bsub -q psfehmpiq -a mympi -n 12 -R "span[hosts=1]" -o ~/output/%J.out ~/bin/hello

Will submit an OpenMPI job (-a mympi) requesting 12 processors (-n 12) spanned all on one host (-R "span[hosts=1]") to the psfehmpiq batch queue (-q psfehmpiq).