This page documents some SLAC specific examples of using the GRID for dataset production and retrieval.
GRID Certificates
The centrally maintained ATLAS page for starting on the GRID is the place to start
DQ2 Setup at SLAC
DQ2 is the ATLAS data management system. There is significant documentation on it's general prinicples, usage, and troubleshooting here
In order to being using the DQ2 tools at SLAC one simply needs to source the following script
| Code Block |
|---|
source /afs/slac.stanford.edu/g/atlas/etc/hepix/GridSetup.(c)sh
|
This script is automatically run for you if you are using the standard ATLAS setup and run "bash",
as described here: SLAC ATLAS Computing Environment
You then need to get a GRID ticket, which will last 12 hours or so:
| Code Block |
|---|
voms-proxy-init -voms atlas
|
Some simple dq2 commands
To add your files to a personal dataset (which can also be used by others):
| Code Block |
|---|
dq2-put --long-surls -p lcg -L SLACXRD_LOCALGROUPDISK -s mydirectory user09.AndrewHaas477621.SG_pythia_real_1000GeV.ESD.v1
|
where "mydirectory" is a directory containing the files you want to add to the dataset.
If you don't like answering "yes" to all the questions, include option "-a".
If you get "LFC exception [Could not create path with error Permission denied ... ", it's possible that the group membership of /grid/atlas/dq2/user09 is wrong. Complain to Non-Grid Jobs at SLAC!
If your directory contains ".pool.root" files, you need to setup a release first, so dq2-put can calculate the GUID for each file using "pool_extractFileIdentifier". Note: the GUID is created at the time the file is created, based upon the name of the file, machine, etc. To be safe,create the file with an original filename, by inserting some random string in it!
The dataset name you use should conform to the format "user09.DN.name.datatype.version", as above,
where DN is your identifier extracted from your certificate, and can be computed from:
| Code Block |
|---|
python /afs/cern.ch/atlas/offline/external/GRID/ddm/Operations/utils/dn.py
|
To list the files in a dataset (note, you can use wildcards...):
| Code Block |
|---|
dq2-ls -f user09.AndrewHaas477621.SG_pythia_real*.ESD.v1
|
To freeze the dataset:
| Code Block |
|---|
dq2-freeze-dataset user09.AndrewHaas477621.SG_pythia_real_1000GeV.ESD.v1
|
To get the dataset:
| Code Block |
|---|
cd /tmp
dq2-get user09.AndrewHaas477621.SG_pythia_real_1000GeV.ESD.v1
ls -lh user09.AndrewHaas477621.SG_pythia_real_1000GeV.ESD.v1/
|
Transfering large datasets
This is the old way:
To request an import of a large dataset to SLAC (it must be avilable first at BNL!):
| Code Block |
|---|
dq2-register-subscription --archive <dataSet> SLACXRD_LOCALGROUPDISK
|
(the --archive flag makes sure it doesn't automatically get deleted after a week)
There's similar code that works with DQ2 containers:
| Code Block |
|---|
dq2-register-subscription-container --archive data09_cos.00121416.physics_L1Calo.merge.DPD_CALOCOMM.r733_p37/ SLACXRD_LOCALGROUPDISK
|
The new way:
Go to:
| Panel |
|---|
http://panda.cern.ch:25980/server/pandamon/query?mode=ddm_req |
...
It will take some time for the data to appear. You can check with:
| Code Block |
|---|
dq2-ls -f <dataSet>
|
to see how many files are available locally.
And you can make a PoolFileCatalog.xml file directly:
| Code Block |
|---|
dq2-ls -P <dataSet>
sed s%srm://osgserv04.slac.stanford.edu:8443/srm/v2/server?SFN=/xrootd/atlas%root://atl-xrdr//atlas/xrootd%g PoolFileCatalog.xml >! PoolFileCatalog.xml
|
PATHENA Setup at SLAC
A commonly used set of tools for distributed analysis is PANDA
- The PanDA Production and Distributed Analysis System
- Performing distributed analsysis with PANDA: PATHENA
- Client tools for Panda analysis jobs
In order to being using these tools at SLAC, one simply needs to source 1 script and set 1 environment variable
| Code Block |
|---|
source /afs/slac/g/atlas/packages/panda-client/etc/panda/panda_setup.sh
export PATHENA_GRID_SETUP_SH="/afs/slac/package/vdt/wlcg-client/setup.sh"
|