Skip to content

How to install software

General options

Local installation

This option means that the user installs the desired software by his/her own means anywhere into his/her home directory.

Variant 1: install from within interactive job (>= 5 min lasting installation)

# first start an interactive job with scratch directory
qsub -I -l select=1:ncpus=1:mem=4gb:scratch_local=10gb -l walltime=2:00:00

# after the job starts, redirect temporary files into scratch
mkdir $SCRATCHDIR/tmp
export TMPDIR=$SCRATCHDIR/tmp

# make an installation directory anywhere in your home
mkdir /storage/CITY_XY/home/user123/my_fancy_sw

# then install into this directory
cd /storage/CITY_XY/home/user123/my_fancy_sw 
install ...

Variant 2: install without interactive job (<= 5 min lasting installations ONLY)

# first redirect temporary files into any directory in your home
mkdir /storage/city_XY/home/user123/tmp
export /storage/city_XY/home/user123/tmp

# make an installation directory anywhere in your home
mkdir /storage/CITY_XY/home/user123/my_fancy_sw

# then install into this directory
cd /storage/CITY_XY/home/user123/my_fancy_sw 
install ...

why do I have to redirect to TMPDIR?

By default the variable TMPDIR points to system directory /tmp. This directory is writable to all, but users have a small quota of only 1 GB in /tmp. Therefore installations with default value of TMPDIR often fail due to insufficient space.

Features of local installation:

  • no need to contact user support, to wait or to ask for aything (unless help is needed),
  • if the CITY_XY storage is down, the software will not be available,
  • it is up to user to figure out how to install the software,
  • suitable for cases when the installation works out-of-the-box and the user wants to try the software ASAP.

We recommend this option in case:

  • you are not sure whether this software will be really useful and want to test it first,
  • the calculations run with this software will take hours rather than days and weeks,
  • as far as you know, there are no other users who would like to use this software,
  • the installation process seems to be manageable with your skills,
  • you do not want to interact with User support in any way.

Warning

With software installed in your home directory, the availability is limited to the availability of that particular /storage/city_XY. That is why, in case you have tested the software and intend to use it for long running jobs, we encourage users to ask us for system wide installation.

Tip

Often users try to install a minor and easy-looking software locally and fail. This is very common. If this happens (and you still want to keep the installation to your home directory), do not hesitate to contact User support at meta@cesnet.cz for help. They will reproduce the errors you get, find a solution and send it back to you in step-by-step guide. Any installation process, no matter how straightforward it looks at the beginning, can get tricky.

System-wide installation

This option means that the user sends request for software installation to meta@cesnet.cz. The User support team will install the software into system directories and make a modulefile through which the software can be loaded.

Features of system-wide installation:

  • you have to contact user support and wait for several days before the thing is done (hours if you are lucky, but most often several days),
  • no need to go into the installation process peculiarities,
  • availability of the software is independent on particular /storage availability.

We recommend this option in case:

  • it is a new version of already installed and/or commonly used software in MetaCentrum,
  • you are confident about the new software's usability,
  • you know the software will be useful to more users.

How to write software install request

There exists plethora of scientific software, sometimes with identical names. If you want us to install something, please give us as much of the following as possible:

  1. Software name (required),
  2. Version ("latest" or specific version) (required),
  3. Link to where the source can be downloaded from (required),
  4. Link to documentation,
  5. Link to installation guide (if there is any) ,
  6. Link to test suite or an example calculation (if there is any),
  7. Any dependencies, modules or optional parts you need to be compiled together with the main software (if there are any).

Common installation methods

When you download a software, it's usually compressed in some way and need to be extracted.

.tar

tar -x -f package.tar       # extract .tar archive

gz

gunzip package.gz           # unzip .gz file

.tgz, .tar.gz

tar -xvf package.tgz        # unzip tarred and gzipped file
tar -xvf package.tar.gz     # dtto

.zip

unzip package.zip           # unzip .zip file

Binaries

Some software is available as a ready-made binary file. In this case the only thing you need to do is to download and unpack the files.

user123@skirit:~$ mkdir binaries ; cd binaries # prepare directory for the software (noc necessarry)
user123@skirit:~/binaries$ wget https://sw_xy.tgz ; tar -xvf sw_xy.tgz # download and unpack 
user13@skirit:~/binaries$ chmod 755 exec_file # set the binary to be executable
user13@skirit:~/binaries$ ./exec_file # run the executable file

Example: Usage of software "satsuma" distributed as binary file.

user123@skirit:~$ mkdir satsuma ; cd satsuma # prepare dir for software "satsuma"
user123@skirit:~/satsuma$ wget https://github.com/bioinfologics/satsuma2/releases/download/untagged-2c08e401140c1ed03e0f/satsuma2-linux.tar.gz # download it
user13@skirit:~/satsuma$ tar -xvf satsuma2-linux.tar.gz # unpack it
user13@skirit:~/satsuma$ cd product/bin ; ls # list and run the executables

R packages

To install R packages locally, first create some directory to serve as your local R library:

mkdir /storage/CITY_N/home/USERNAME/Rpackages/

Then install the package as:

export R_LIBS_USER="/storage/CITY_N/home/USERNAME/Rpackages"
module add r/
R
>install.packages(PACKAGE_NAME,lib="/storage/CITY_N/home/USERNAME/Rpackages/")

and load the package

>library(PACKAGE_NAME)

Debian packages

Users can install debian packages (.deb) locally (in the user's home directory) without the root privileges.

Sometimes it is fastest solution to get one particular library required by a particular application.

Warnings

Users cannot use apt-get install, apt install directly. They can, however, use apt search, apt-get download and then install the package locally by dpkg command.

Obtain the package:

# some packages can be found by apt search directly
apt search libargtable2-0       
apt-get download libargtable2-0

# not all .deb packages may be available in repo
apt search libicu63      # returns nothing

# however the package is available in http://ftp.cz.debian.org/debian/pool/main/i/icu/
wget http://ftp.cz.debian.org/debian/pool/main/i/icu/libicu63_63.1-6+deb10u3_amd64.deb

Install the .deb package:

mkdir extract_dir
dpkg -x libargtable2-0_12-1.1_amd64.deb extract_dir

Add path to new bin/library to PATH/LD_LIBRARY_PATH: 

export LD_LIBRARY_PATH=storage/cityN/home/user123/extract_dir/lib/:$LD_LIBRARY_PATH
export PATH=/storage/cityN/home/user123/extract_dir/bin/:$PATH

Perl modules

First check if there is an already installed perl package in one of our modules:

module add bioperl

(BOOKWORM)user_123@skirit:~$ perldoc perllocal
# or
(BOOKWORM)user_123@skirit:~$ instmodsh

Install a new Perl package

cpanm (CPANMINUS tool) is the specialized tool for installing and uninstalling of Perl packages from CPAN. It is available via modules. Use it like this:

load module bioperl-1.6.9-gcc or bioperl-1.6.1

cpanm -L /specified/local/directory/ GD::Graph::bars – to install Perl library and all of it's dependencies including the libraries already present in system to specified directory

After that don't forget to set PATH and PERL5LIB to the specified directory bin and lib folders to be able to use the installed binaries and libraries.

Python packages

Use --user option rather that --root during pip install

A very convenient feature is to use the --user option of pip install. This will install modules, additional to the available system python install, in the location defined by the PYTHONUSERBASE environment variable. You will also need to add them to your path and pythonpath so that they are accessible from any folder you wish to execute your code. For this purpose, export PATH=$PYTHONUSERBASE/bin:$PATH and export PYTHONPATH=$PYTHONUSERBASE/bin:$PYTHONPATH will do the job.

Example: install software package "spektral".

First run an interactive job:

qsub -I -l select=1:ncpus=1:mem=4gb:scratch_local=10gb -l walltime=2:00:0

Then install the package: 

# store temporary files in SCRATCHDIR
export TMPDIR=$SCRATCHDIR

# load pip module
module add py-pip

# make a local directory to install to
mkdir /storage/cityN/home/user123/my_pip_libs

# ... and tell pip about it
export PYTHONUSERBASE=/storage/cityN/home/user123/my_pip_libs/     

# install spektral and its dependencies
pip3 install spektral --user        

exit

To run the package, first find out which python version was used to install the package with:

(BULLSEYE)user123@skirit:~$ ls /storage/cityN/home/user123/my_pip_libs/lib/
python3.7

Then choose any python module version 3.7.x:

module add python/3.7.7-intel-19.0.4-mgiwa7z

# setup system variables 
export PYTHONUSERBASE=/storage/cityN/home/user123/my_pip_libs
export PATH=$PYTHONUSERBASE/bin:$PATH
export PYTHONPATH=$PYTHONUSERBASE/lib/python3.7/site-packages:$PYTHONPATH

And use the package:

python
>>> import spektral
...

Conda/Mamba packages

Tip

Some of the most popular packages may be already installed - see module add mambaforge ; mamba env list.

Python package installers: Conda, Mamba, Micromamba

Conda, available through mambaforge and conda-modules/ module, is an open-source package management and environment management system that is commonly used in the Python programming ecosystem.

Mamba, available through mambaforge module, is a package manager and environment manager that is designed to be a faster and more efficient drop-in replacement for Conda. Both Conda and Mamba are closely related, as Mamba was developed as an open-source project to address some of the performance limitations of Conda.

Micromamba, available through micromamba module, is a lightweight and resource-efficient alternative to Mamba. Mamba, while faster than Conda, is still a relatively large application. In contrast, Micromamba is built to be more minimalistic and efficient, making it a suitable choice for resource-constrained environments, such as embedded systems or containers.

Tip

We recommend to use mamba from mambaforge module as a primary installer for new Conda environments.

Example: Install package segemehl.

module add mambaforge
# create new environment called segemehl-0.3.4 (with python 3.8)
mamba create --prefix /storage/city/home/user_name/segemehl-0.3.4 python=3.8 -y
# activate the environment
mamba activate /storage/city/home/user_name/segemehl-0.3.4
# install the package
mamba install -c bioconda segemehl -y
# leave the environment
mamba deactivate

If you need to install multiple programs at the same time, use the following syntax:

# with this syntax, mamba will resolve all environment dependencies only once
mamba create --prefix /storage/city/home/user_name/my_new_env -c channel_1 -c channel_2 <other channels> software_1 software_2 <other softwares>

Activate this environment (e.g. within batch jobs) as:

module add mambaforge
mamba activate /storage/city/home/user_name/segemehl-0.3.4
segemehl.x ... # run the job
mamba deactivate

Tip

If you want (esp. for long-running jobs) to make your calculations more robust, e.g. against failure of the network connection to the storage, you can clone your mamba environment directly to the scratch directory: cd $SCRATCHDIR
module add mambaforge
mamba create -p $SCRATCHDIR/my_env --clone /storage/city/home/user/ny_env
mamba activate $SCRATCHDIR/my_env

Containers

Apptainer (Singularity) images can be deployed and run directly on MetaCentrum machines.

If your software is released as Docker container, you have more option.

  • Apptainer (former Singularity) image can be run in MetaCentrum directly, see Singularity howto.
  • Docker images must be translated into an Apptainer image and run as described in this chapter.
  • Docker images can be run also via Kubernetes service. See Kubernetes for instruction.

Compile from source (C, C++, Fortran)

Choosing the compiler

You should choose what compiler you use for building an application. There are three main compilers:

  • GCC compiler - free compiler which is usually most compatible with the software being built. Better is to use the system version then some from module if that is possible.
  • Intel compiler - commercial compiler with excellent math libraries and optimization abilities.
  • PGI compiler - another commercial compiler with optimized math libraries for AMD clusters.

We suggest to first try the default Intel compiler and in case it fails, proceed to GCC.

Configuration tuning

Usually you need to configure the software prior to building. And there are usually three different "configurers". Some advices for each are here:

  • Libraries:
    • use MKL (math libraries from Intel CDK) when possible
    • use MPI when possible, preferably OpenMPI but MVAPICH is also available (MVAPICH is MPICH with support of InfiniBand)
    • use Thread (OpenMP) when possible
    • use CUDA in separate compilations
    • avoid all optimizations for machine (-xHost and -fast flags) where you are building your application
  • configure - use ./configure --help to get the available options. If no ./configure script is present, try first ./autogen.sh (for newer than system version use module autotools-2.26). Then look above to general advices.
  • cmake - first add one of the cmake modules. Then make a build directory (mkdir build && cd build) and run ccmake ../ to get and adjust configuration options. All options are available after pressing "t" key. Then look above to general advices. You can look in ccmake for options and then use them in command line with -D prefix. Like: cmake -DCMAKE_INSTALL_PREFIX=/software/prg/version.
  • Makefile - sometimes all configuration is done only in pre-generated Makefile. Edit it using your favourite editor. Don't forget to look above to general advices.

Environment variables and Flags

There are some usual environment variables in which you can put some "flags" that influence the compilation or linking. The standard make rule for compiling a C (or C++ or Fortran) program is:

%.o:%.file_type
   $(Compiler) $(PreprocessorFLAGS) $(CompilerFLAGS) -c -o $@ $<

The corresponding compilers and variables that influence it's behavior are described in following table.

Compiler Preprocessor flags Compiler flags
C CPPFLAGS CFLAGS
C++ CPPFLAGS CXXFLAGS
Fortran 77 FPPFLAGS F77FLAGS, FFLAGS
Fortran 90 FPPFLAGS F90FLAGS, FFLAGS

For example, if you use C compiler (gcc, icc, pgcc) and want to influence the compilation phase, you should set some flags in variable $CFLAGS. If you use C++ compiler, use $CXXFLAGS variable. If you use both and want to have some common flags, use $CPPFLAGS variable. So for C/C++ projects you will normally need to use only the CPPFLAGS for compilation.

Linker flags are always $LDFLAGS. You should always set capital "L" paths prior to linked libraries.

Type of flags Purpose Example
Preprocessor flags Compiler inspecific optimization and include paths. CPPFLAGS="-I/software/prg1/include -I/software/prg2/include -O2 -msse -fPIC"
Compiler flags Compiler specific optimization and include paths. CXXFLAGS="-I/software/prg1/include -I/software/prg2/include -O2 -msse -fPIC"
Linker flags (LDFLAGS) Linker directives and library paths. LDFLAGS="-L/software/prg1/version/lib -L/software/prg2/version/lib -lcrypt -lmkl_blas95_lp64 -lpthread /software/prg1/lib/libprg.a"

If the programs you are dealing with support pkgconfig mechanism, it is a good idea to set the $PKG_CONFIG_PATH, usually /software/prg/version/lib/pkgconfig.

In module meta/0-utils are available scripts set-* (ls /software/meta-utils/internal/set-*) for setting of certain compilation environments. Use it at least for inspiration.

Math libraries introduction

Let's describe some relationships among the linear algebra libraries BLAS, LAPACK, BLACS and ScaLAPACK. For the quick overview see picture in Intel compiler.

For example,

  • BLAS is a dependency of LAPACK and you can not link LAPACK without BLAS,
  • LAPACK, BLAS and BLACS (pBLAS) are dependencies of ScaLAPACK and you should link it all if you are using ScaLapack
  • BLACS (pBLAS) are dependent also on MPI implementation. You should choose the right library depending on MPI you are using (OpenMPI or M(VA)PICH). Math libraries linking examples are described on INTEL CDK page.