Intel® Cluster Checker User Guide

User Guide

  • 2021.1
  • 01/08/2021
  • Public Content
Contents

Installation

Intel® Cluster Checker is a robust, dynamic tool useful for diagnosing many cluster-related issues. There exists a myriad of options and configurations available for clusters to enhance their performance and functionality. This guide will only address the bare minimum of requirements to facilitate the installation of Intel® Cluster Checker. For specific network fabrics, compilers, messaging protocols, etc., please refer to the installation requirements and procedures for the definitive guide and use of these options. In many cases, Intel® Cluster Checker has been enhanced to utilize the tools provided with a wide range of popular network fabrics, compilers and messaging protocols, including Mellanox*, Intel® Parallel Studio Cluster Edition, and Intel® MPI Library.
Note
 There are currently two major releases of Intel® Cluster Checker available. A 2019 version and a 2021 version. The 2021 version of Cluster Checker is also part of Intel® oneAPI HPC Toolkit. Both versions are covered in this documentation. In areas where there is different ways of installation or pathing, we cover both options. The commands and options used in the two versions are the same. Future updates will occur to the 2021 version.
Installation Guidance
We suggest the use of the latest version of Intel® Cluster Checker. Your fist choice is to select a version. As guidance, we would suggest using the 2021 version that is included with Intel® oneAPI Base & HPC Toolkits. If you are looking to install Cluster Checker on an existing system on its own, use the YUM repository install options if you have Internet connectivity from your cluster headnode. Otherwise the standalone installer is a good option.
If you plan to install Intel® Cluster Checker as a non-privileged user only, you will need to make use of the standalone installer tools. Either through Intel® oneAPI Base & HPC Toolkits or through the standalone Intel® Cluster Checker 2019 installer.
Options for installing
Intel® Cluster Checker 2021:
Install option
Notes
YUM Repository (recommended)
Intel® oneAPI and oneAPI HPC Toolkit (recommended if using oneAPI tools).
Standalone
  • Download the ‘Local’ or ‘online’ install for Intel® Cluster Checker component (select ‘online’ version if you have web-access from the installation server as this will just download what you select to install).
  • Run the install script
    l_clck_p_2021.1.1.68_offline.sh
     and follow the on screen text user interface to complete the installation.
Standalone with the Intel® oneAPI HPC Toolkit
  • Download the ‘Local’ install for HPC Toolkit (select ‘online’ version if you have web-access from the installation server as this will just download what you select to install).
  • Please note the download page for HPC Toolkit warns that your require the oneAPI basekit in addition to the HPC toolkit. If you are just looking for Intel® Cluster Checker 2021 and already have other components installed, this warning can be ignored.
  • When running the install script
    l_HPCKit_p_2021.1.0.2684_offline.sh
     customize the install selection for just Cluster Checker
Options for installing
Intel® Cluster Checker 2019:
Install option
Notes
Through YUM Repository
Installing Intel® Cluster Checker 2021
System Requirements: The following system requirements are necessary in order to use Intel® Cluster Checker 2021.
Hardware
  • Intel® Xeon® processor
  • 4 GB of RAM recommended
  • 1 GB of free hard disk space required for installation
Software
  • One of the following operating systems:
    • CentOS 7 and 8
    • Red Hat* Enterprise Linux* 7 and 8
    • SUSE* Linux* Enterprise Server 12 or 15
    • Ubuntu* 16.04 or 18.04 LTS (There are known limitations to Ubuntu support, see release notes section 7 for additional details).
  • Additional Software Packages
    • Intel® oneAPI HPC toolkit and basekit. To provide Intel® MPI Library, Intel® Math Kernel Library and Intel® C++ Compiler.
    • Optional
      (advanced)
      : OSU Micro-Benchmarks to run OSU tests (see Installation instructions). After successful build and install of OSU benchmarks, the binary path for these micro-benchmarks need to be added to the users $PATH environment variable by either adding to .bashrc or as part of the slurm launch script. export PATH=$PATH:/location/of/osu/binaries.
    • Optional
      (advanced):
       MPICH to run Cluster Checker (see Installers Guide). If you are planning to use MPICH rather than Intel® MPI Library to launch Cluster Checker on compute systems, insure the PATH and LD_LIBRARY_PATH are setup to take priority over any other MPI in the users PATH. I.e.
      export
      PATH=/location/of/mpich/bin:$PATH
       and
      export
      LD_LIBRARY_PATH=/location/of/mpich/libraries:$LD_LIBRARY_PATH
By default Intel® Cluster Checker is configured to use PDSH to launch data collection on compute nodes in the cluster. This can be changed to use MPI and is covered in more details in
Data Collection
 configuration section. Today Cluster Checker only supports Intel® MPI Library and MPICH for data collection via MPI. Other MPI varieties are not expected to work correctly.
YUM installation for Intel® Cluster Checker 2021
Intel® oneAPI provides a YUM repository for installing Intel® Cluster Checker. You can also find details directly on the oneAPI installation guidance page, which also covers how to install the entire oneAPI HPC Toolkit. https://software.intel.com/content/www/us/en/develop/articles/installation-guide-for-intel-oneapi-toolkits.html [Note: These steps require privileged access]
  1. If you had previously installed early Intel® oneAPI beta packges, remove them 
    
sudo -E yum remove intel-hpckit intel-basekit
  2. Create a oneAPI repo file as a regular user 
    
tee > /tmp/oneAPI.repo << EOF
[oneAPI]
name=Intel(R) oneAPI repository
baseurl=https://yum.repos.intel.com/oneapi
enabled=1
gpgcheck=1
repo_gpgcheck=1
gpgkey=https://yum.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS-2023.PUB
EOF
  3. Add the Intel® oneAPI YUM (DNF) repository to
    /etc/yum.repos.d
    
sudo mv /tmp/oneAPI.repo /etc/yum.repos.d
  4. Check available Intel® Cluster Checker 2021 versions 
    
repoquery --repofrompath=reponame,https://yum.repos.intel.com/oneapi --repoid=reponame -a | grep clck
    or
    
sudo -E yum --disablerepo="\*" --enablerepo="oneAPI" list available | grep intel-oneapi | grep -v intel-oneapi-runtime | grep clck
  5. Install the desired version of Intel® Cluster Checker 2021 using the YUM package manager. The following variables are used in the installation command: <COMPONENT>,<VERSION>,<UPDATE>,<BUILD_NUM>:

sudo yum install <COMPONENT>-<VERSION>.<UPDATE>-<BUILD_NUM>
  • Example:

sudo yum install intel-oneapi-clck-2021.1.1-68
The following table lists the available versions of Intel® Cluster Checker 2021:
<COMPONENT>
<VERSION>
<UPDATE>
<BUILD_NUM>
YUM Repository Path
intel-oneapi-clck
2021
1.1
68
Installing Intel® Cluster Checker 2019
System Requirements - The following system requirements are necessary in order to use Intel® Cluster Checker.
Hardware
  • Intel® Xeon® processor
  • 4 GB of RAM recommended
  • 1 GB of free hard disk space required for installation
Software
  • One of the following operating systems:
    • CentOS 7 and 8
    • Red Hat* Enterprise Linux* 7 and 8
    • SUSE* Linux* Enterprise Server 12 or 15
    • Ubuntu* 16.04 or 18.04 LTS (There are known limitations to Ubuntu support, see release notes section 7 for additional details).
  • Additional Software Packages
    • Cluster Checker uses Intel® MPI Library, Intel® Math Kernel Library via available runtimes or full software packages;
    • Optional
      (advanced)
      : OSU Micro-Benchmarks to run OSU tests (see Installation instructions). After successful build and install of OSU benchmarks, the binary path for these micro-benchmarks need to be added to the users $PATH environment variable by either adding to .bashrc or as part of the slurm launch script. export PATH=$PATH:/location/of/osu/binaries.
    • Optional
      (advanced):
       MPICH to run Cluster Checker (see Installers Guide). If you are planning to use MPICH rather than Intel® MPI Library to launch Cluster Checker on compute systems, insure the PATH and LD_LIBRARY_PATH are setup to take priority over any other MPI in the users PATH. I.e.
      export
      PATH=/location/of/mpich/bin:$PATH
       and
      export
      LD_LIBRARY_PATH=/location/of/mpich/libraries:$LD_LIBRARY_PATH
By default Intel® Cluster Checker is configured to use PDSH to launch data collection on compute nodes in the cluster. This can be changed to use MPI and is covered in more details in
Data Collection
 configuration section. Today Cluster Checker only supports Intel® MPI Library and MPICH for data collection via MPI. Other MPI varieties are not expected to work correctly.
YUM Repository Installing Intel® Cluster Checker 2019
In order to install Intel® Cluster Checker using the Intel® YUM repository, follow the instructions below. [Note: These steps require privileged access]:
  1. Add the Intel® Cluster Checker YUM repository: 
    
sudo yum-config-manager --add-repo
https://yum.repos.intel.com/clck/2019/setup/intel-clck-2019.repo
  2. Import the gpg public key for the repository: 
    
sudo rpm --import https://yum.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS-2019.PUB
  3. Get a list of all available versions of Intel® Cluster Checker 2019 on the YUM repository 
    
repoquery --repofrompath=reponame,https://yum.repos.intel.com/clck/2019 --repoid=reponame -a
  4. Install the desired version of Intel® Cluster Checker using the YUM package manager. The following variables are used in the installation command: intel-oneapi-<COMPONENT>,<VERSION>,<UPDATE>,<BUILD_NUM>:

sudo yum install <COMPONENT>-<VERSION>.<UPDATE>-<BUILD_NUM>
  • Example:

sudo yum install intel-clck-2019.10-060
The following table lists the available versions of Intel® Cluster Checker:
<COMPONENT>
<VERSION>
<UPDATE>
<BUILD_NUM>
YUM Repository Path
intel-clck
2019
10
060
intel-clck
2019
9
056
intel-clck
2019
8
048
intel-clck
2019
7
043
intel-clck
2019
6
038
intel-clck
2019
5
030
intel-clck
2019
4.0
027
only available through Intel® Parallel Studio XE
intel-clck
2019
3.5
025
intel-clck
2019
2.1
024
intel-clck
2019
0
015

Product and Performance Information

1

Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex.