LHC Data sources

LHC machine data is normally available through:

• the CERN Accelerator Logging Service that can be queried with the Timber application (provided Java 1.8 and jws are installed);
• the NFS file system available from the technical network (e.g. ssh cs-ccr-dev1) in the directory /user/slops/data/LHC_DATA/OP_DATA;
• in the Post Mortem System database accessible through its java API and other dedicated tools.

Here we describe alternative and/or complementary sources and methods to obtain data:

• PyTimber Python API for querying data of the CERN Accelerator Logging Service and integrated in the web Jupyter notebook server;
• EOS abp space tree in the EOS filesystem with copies of volatile data some of which can be queried with PageStore, a Python library similar to PyTimber using either the Jupyter notebook server or any linux system with a mounted EOS filesystem (e.g. lxplus-cb6).

Eos space

40 TB have been granted in the eosuser EOS instance on the space /eos/project/a/abpdata .

Data can be accessed from regular lxplus machines (e.g. ssh lxplus) in /eos/project/a/abpdata or

using the eos commands =ls, cp, ... = to operate on the files e.g. :

eos ls /eos/project/a/abpdata

or by mounting it as a filesystem in a local directory e.g:

mkdir eos; eos fuse mount ./eos .

Eos data can be also accessed with xrootd library, e.g.:

mkdir eos; xrootdfs ./eos -o rdr=root://eosuser.cern.ch//eos

Eos abpdata space can be also accessed via web by

http://cern.ch/abpdata

Castor

A subset of historical LHC data is stored in

/castor/cern.ch/user/r/rdemaria/data/
/castor/cern.ch/user/r/rdemaria/lhcfilldata/data

PyTimber and PageStore software

PyTimber is a Python library that allows to access the data stored in the CERN logging data.

PageStore is a Python library that allows to store named and indexed records on disk for fast bulk read.

The software is installed in the Jupyter notebook server and lxplus.

In order to use them one needs a JDK >= 1.8. For windows, be sure to install the 64bit version if running on a 64bit windows installation.

In case of a non-standard location for the java library set the path of libjvm.so in JAVA_JVM_LIB, e.g.:

export JAVA_JVM_LIB="/cvmfs/sft.cern.ch/lcg/releases/java/8u91-ae32f/x86_64-slc6-gcc49-dbg/jre/lib/amd64/server/libjvm.so"

or the MacOS installation

• install oracle java jdk from http://www.oracle.com/technetwork/java/javase/downloads/index.html
• check for installed java versions:

/usr/libexec/java_home -V

you should see sth. like:

    1.8.0_102, x86_64:    "Java SE 8"    /Library/Java/JavaVirtualMachines/jdk1.8.0_102.jdk/Contents/Home
    1.7.0_79, x86_64:    "Java SE 7"    /Library/Java/JavaVirtualMachines/jdk1.7.0_79.jdk/Contents/Home

choose the jdk you just installed, here:

export JAVA_HOME=`/usr/libexec/java_home -v 1.8.0_102`

running java -version should now display:

java version "1.8.0_102"
Java(TM) SE Runtime Environment (build 1.8.0_102-b14)
Java HotSpot(TM) 64-Bit Server VM (build 25.102-b14, mixed mode)

In alternative this variable (specific of cmmnbuild-dep-manager) can also be set:

export JAVA_JVM_LIB="/Library/Java/JavaVirtualMachines/jdk1.8.0_92.jdk/Contents/Home/jre/lib/server/libjvm.dylib"

Then install the following library on top of a recent python installation (e.g. anaconda):

In Linux:

pip install --upgrade  jpype1
pip install --upgrade  git+https://gitlab.cern.ch/scripting-tools/cmmnbuild-dep-manager.git
pip install --upgrade  git+https://github.com/rdemaria/pytimber.git
pip install --upgrade  git+https://github.com/rdemaria/pagestore.git

In Windows is more complicated since jpype cannot be installed smoothly with pip:

• install the latest anaconda distribution with Python 3.5.
• download and copy the file: JPype1-0.6.1-cp35-none-win32.whl or JPype1-0.6.1-cp35-none-win_amd64.whl depending on the architecture from

http://www.lfd.uci.edu/~gohlke/pythonlibs/#jpype

• open the anaconda prompt and type:

conda install git
conda install --upgrade numpy
conda install --upgrade matplotlib
pip install JPype1-0.6.1-cp35-none-win_amd64.whl
pip install --upgrade  git+https://gitlab.cern.ch/scripting-tools/cmmnbuild-dep-manager.git
pip install --upgrade  git+https://github.com/rdemaria/pytimber.git
pip install --upgrade  git+https://github.com/rdemaria/pagestore.git

.

A subset if historical LHC data is stored in EOS the location /eos/project/abpdata using PageStore and in Castor.

A centralized installation in lxplus is available by using

/afs/cern.ch/work/r/rdemaria/public/anaconda/bin/ipython

Accessing LHC database

Log in lxplus-cb6 to access the database in the EOS space or any lxplus to just use PyTimber and PageStore locally.

To start an terminal ipython prompt use

/afs/cern.ch/work/r/rdemaria/public/anaconda/bin/ipython

In order to use PageStore on EOS, type in the python prompt

!cp /eos/project/abpdata/lhc/lhc.db .

import pagestore
db=pagestore.PageStore('lhc.db','/eos/project/abpdata/lhc/datadb/')

Now it is possible to query the database similarly to PyTimber, e.g.:

#obtain old BBQ raw data
db.search('%BBQ%')
t1,t2=db.get_lim('LHC.BQBBQ.CONTINUOUS_HS.B1:ACQ_DATA_H')
t1,t2=db.get_lim('LHC.BQBBQ.CONTINUOUS.B1:ACQ_DATA_H')
print pagestore.dumpdate(t1)
print pagestore.dumpdate(t2)
data=db.get('LHC.BQBBQ.CONTINUOUS.B1:ACQ_DATA_H',t1,t1+100)

The library and the data can be also accessed using the SWAN service.

Swan: Centralized Jupyter (beta) server

Swan provide a Jupyter service accessible with CERN credentials.

The server can be accessed from swan.cern.ch. When accessing the website, click on "Start My server" and set /eos/project/abpdata/setup_swan.sh as setup script in the form.

From outside CERN network, one can set-up an SSH tunnel: ssh -D6789 lxplus.cern.ch and set a socket proxy in the browser, e.g. in Firefox:

Preferences -> Advanced -> Network -> Settings... -> Manual proxy configuration --> set SOCKS Host: to 127.0.0.1 and Port: to 6789.

Examples are in:

• /eos/project/abpdata/examples/LHC Page1.ipynb
• /eos/project/abpdata/examples/Test Pagestore.ipynb

Xrootd and eos compilation

Compilation in ubuntu needed for eos command line tool.

Compile XrootD

sudo apt-get install git cmake libzfslinux-dev uuid-dev libsparsehash-dev libreadline-dev xfslibs-dev libattr1-dev libcppunit-dev micro-httpd libmicrohttpd-dev libzmq3-dev zlib1g-dev libfuse-dev libfuse2 libkrb5-dev libcrypto++-dev
git clone https://github.com/xrootd/xrootd.git
cd xrootd
mkdir build
cd build
cmake .. -DCMAKE_INSTALL_PREFIX=/usr
sudo make install

Get special header for zmq (bug in ubuntu distribution or eos)

cd /tmp/
git clone https://github.com/zeromq/cppzmq
sudo cp cppzmq/zmq.hpp /usr/include/

Compile eos (not really successful)

git clone https://gitlab.cern.ch/dss/eos.git
cd eos
rm utils/zmq.hpp
mkdir build
cd build
cmake ../ -DCMAKE_INSTALL_PREFIX=/opt -DCLIENT=True

Meetings and Talks

1/6/2016 HSS meeting

General introduction to the tools: Python-based tool for LHC data mining

29/04/2016

Present: Gianni, Guido, Nicolò, Panagiotis, Xavier

Actions All

• test pagestore standalone and from eos. Report feedback
• collect list of interested variables to store systematically

-- Main.RiccardoDeMaria - 2016-04-30