Difference between revisions of "Client software Install"

From Define Wiki
Jump to navigation Jump to search
(Created page with "==Introduction== The FabreX Operating System is the software engine that drives the performance and dynamic composability of GigaIO software-defined infrastructure (SDI) for e...")
 
 
(16 intermediate revisions by the same user not shown)
Line 1: Line 1:
 
==Introduction==
 
==Introduction==
 +
 
The FabreX Operating System is the software engine that drives the performance and dynamic composability of
 
The FabreX Operating System is the software engine that drives the performance and dynamic composability of
 
GigaIO software-defined infrastructure (SDI) for enterprise data centers and high-performance computing
 
GigaIO software-defined infrastructure (SDI) for enterprise data centers and high-performance computing
environments. This Linux-based, resource-efficient software layers onto FabreX hardware for easy-to-use
+
environments.  
 +
This Linux-based, resource-efficient software layers onto FabreX hardware for easy-to-use
 
composing of computing clusters on-the-fly.
 
composing of computing clusters on-the-fly.
The FabreX Operating System is composed of three main components: 1) Server software, 2) Switch software,
+
 
and 3) Management software. The Server software runs on Linux and provides libfabric functionality for use with
+
The FabreX Operating System is composed of three main components:  
 +
#Server software,  
 +
#Switch software,
 +
# Management software.  
 +
The Server software runs on Linux and provides libfabric functionality for use with
 
MPICH for HPC applications. It also provides a TCP/IP interface for communication between servers. The
 
MPICH for HPC applications. It also provides a TCP/IP interface for communication between servers. The
 
FabreX Switch software runs on an embedded switch controller and provides dynamic composability and
 
FabreX Switch software runs on an embedded switch controller and provides dynamic composability and
 
communications between servers.
 
communications between servers.
The FabreX Operating System and associated hardware provide extreme connectivity for the following
+
 
applications: AI/ML/DL; Cloud Computing; Edge Computing and HPC. The flexibility of the FabreX interconnect
+
{| class="wikitable"
breaks the constraints of old architectures, opening up possibilities for new configurations that maximize
+
|-
utilization of all elements within your advanced scale computing systems. Users can choose from a variety of
+
! Packs
software packages to build the cluster configurations needed.
+
!
Composer Pack
+
! Description
Single-Host Expansion
+
!
Includes: FabreX OS + Management Module + GigaIO CLI + Composition Module
+
|-
The Composer Pack features full disaggregation and dynamic composability for deep and wide IO tree
+
|Composer Pack
 +
|Single-Host Expansion
 +
|Includes: FabreX OS + Management Module + GigaIO CLI + Composition Module
 +
|The Composer Pack features full disaggregation and dynamic composability for deep and wide IO tree
 
configurations. Multiple FabreX switch configurations enable flexible IO expansion including super servers,
 
configurations. Multiple FabreX switch configurations enable flexible IO expansion including super servers,
 
cascading, and resource partitioning. FabreX OS also supports GPUDirect for direct access to third-party
 
cascading, and resource partitioning. FabreX OS also supports GPUDirect for direct access to third-party
 
memory using the standard features of PCIe.
 
memory using the standard features of PCIe.
Leader Pack
+
|-
Multi-host Clusters
+
|Leader Pack
Includes: Everything in Composition Pack + Cluster Module
+
|Multi-host Clusters
The Cluster Module provides true multi-host to host communication in single-switch systems. The Leader Pack
+
|Includes: Everything in Composition Pack + Cluster Module
 +
|The Cluster Module provides true multi-host to host communication in single-switch systems. The Leader Pack
 
adds more power for parallel processing and fast access to distributed memory by integrating MPI, NVMe-oF,
 
adds more power for parallel processing and fast access to distributed memory by integrating MPI, NVMe-oF,
 
and TCP/IP.
 
and TCP/IP.
Maestro Pack
+
|-
Large Multi-Rack Systems
+
|Maestro Pack
Includes: Everything in Leader Pack + Network Pack
+
|Large Multi-Rack Systems
For true composability in large computing systems where you need more clusters and multiple racks, add the
+
|Includes: Everything in Leader Pack + Network Pack
 +
|For true composability in large computing systems where you need more clusters and multiple racks, add the
 
Network Pack. You’ll have all the benefits of the Leader Pack but also the ability to manage multiple Leaderconfigured switches, for quick scaling of your system to meet your largest workloads.
 
Network Pack. You’ll have all the benefits of the Leader Pack but also the ability to manage multiple Leaderconfigured switches, for quick scaling of your system to meet your largest workloads.
 +
|}
 +
 +
==  Software Requirements ==
 +
 +
Software Version: FabreX OS Version 2.0.0
 +
*PCIe Gen 4 Switch support
 +
*Updated Kernel Source
 +
*Updated Fabric Management
 +
*NVIDIA's GPUDirect RDMA (GDR):
 +
 +
=== Downloads ===
 +
 +
{| class="wikitable"
 +
|-
 +
!Target
 +
!Link
 +
|-
 +
|Switch
 +
|[https://gigaio-confluence.s3.us-west-2.amazonaws.com/LSD/spaceFiles/FabreX_v2.0.0.gio?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAUJP5JYSV7VQHCCUT%2F20200522%2Fus-west-2%2Fs3%2Faws4_request&X-Amz-Date=20200522T123015Z&X-Amz-Expires=604800&X-Amz-Signature=565839752cd6115e6377ff594cbd4e37379f9b0c9b59c1f8e7ebe43196037e7e&X-Amz-SignedHeaders=host|FabreX_v2.0.0.gio ]
 +
MD5 = 66f3b9982e9f6d6ab403fd8a1bdf56c3
 +
|-
 +
|Server
 +
|[https://gigaio-confluence.s3.us-west-2.amazonaws.com/LSD/spaceFiles/FabreX_v2.0.0_CentOS7.tar.bz2?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAUJP5JYSV7VQHCCUT%2F20200522%2Fus-west-2%2Fs3%2Faws4_request&X-Amz-Date=20200522T123015Z&X-Amz-Expires=604800&X-Amz-Signature=0af76cca74b5735b78ae6ee18bf57fe8e90d2fff874d0903a729ec5db3cab7bc&X-Amz-SignedHeaders=host|FabreX_v2.0.0_CentOS7.tar.bz2]
 +
MD5 = e92cf05af33a494cedac037e37f40059
 +
|-
 +
|Switch Management
 +
|
 +
*Ubuntu
 +
**[https://gigaio-confluence.s3.us-west-2.amazonaws.com/LSD/spaceFiles/gigaio-fabrexfm-tool_2.0.0.release.22.107d16b_amd64.deb?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAUJP5JYSV7VQHCCUT%2F20200522%2Fus-west-2%2Fs3%2Faws4_request&X-Amz-Date=20200522T123016Z&X-Amz-Expires=604800&X-Amz-Signature=b3b0e8e8923d84c870a5b312e2d5bf0f84cbe55d5015ff252d169d53dbedd6b0&X-Amz-SignedHeaders=host|gigaio-fabrexfm-tool_2.0.0.release.22.107d16b_amd64.deb]
 +
**MD5 = 027c097a5f6aff8df204b87a0074d24a
 +
*Centos
 +
**[https://gigaio-confluence.s3.us-west-2.amazonaws.com/LSD/spaceFiles/gigaio-fabrexfm-tool-2.0.0.release.22.107d16b.x86_64.rpm?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAUJP5JYSV7VQHCCUT%2F20200522%2Fus-west-2%2Fs3%2Faws4_request&X-Amz-Date=20200522T123016Z&X-Amz-Expires=604800&X-Amz-Signature=93d2be8f4d3c7747fc91420e813871fc69895e1c714203be3b395eebd57f5686&X-Amz-SignedHeaders=host|gigaio-fabrexfm-tool-2.0.0.release.22.107d16b.x86_64.rpm]
 +
**MD5 = ca8926e6e749f70b39cc67dbd340ad5a
 +
|}
 +
 +
 +
===Install ===
 +
 +
Installing the Packages
 +
 +
Verify CentOS 7.7 or later with "Development Tools" installed for air-gapped environment.
 +
Copy the tarball to the host.
 +
 +
 +
 +
Go to the directory where the software package was downloaded and untar the files:
 +
<pre>
 +
tar -xjvf FabreX_<VERSION>_CentOS7.tar.bz2
 +
</pre>
 +
 +
 +
Enter the extracted directory.
 +
 +
===== On the head node =====
 +
 +
fabrexfm-tool help setup the PCIE switch to required configuration . This should be present on the head node or node different from 
 +
Compute node. Fmtool is stateless hence it doesn't save the configuration on the server-side. 
 +
 +
<pre>
 +
[root@vcontroller ~]# yum localinstall -y gigaio-fabrexfm-tool-2.0.0.release.22.107d16b.x86_64.rpm
 +
Failed to set locale, defaulting to C
 +
Loaded plugins: fastestmirror, versionlock
 +
Examining gigaio-fabrexfm-tool-2.0.0.release.22.107d16b.x86_64.rpm: gigaio-fabrexfm-tool-2.0.0.release.22.107d16b-1.x86_64
 +
Marking gigaio-fabrexfm-tool-2.0.0.release.22.107d16b.x86_64.rpm to be installed
 +
Resolving Dependencies
 +
--> Running transaction check
 +
---> Package gigaio-fabrexfm-tool.x86_64 0:2.0.0.release.22.107d16b-1 will be installed
 +
--> Finished Dependency Resolution
 +
 +
Dependencies Resolved
 +
 +
================================================================================
 +
Package
 +
      Arch  Version
 +
                    Repository                                            Size
 +
================================================================================
 +
Installing:
 +
gigaio-fabrexfm-tool
 +
      x86_64 2.0.0.release.22.107d16b-1
 +
                    /gigaio-fabrexfm-tool-2.0.0.release.22.107d16b.x86_64 7.3 M
 +
 +
Transaction Summary
 +
================================================================================
 +
Install  1 Package
 +
 +
Total size: 7.3 M
 +
Installed size: 7.3 M
 +
Downloading packages:
 +
Running transaction check
 +
Running transaction test
 +
Transaction test succeeded
 +
Running transaction
 +
  Installing : gigaio-fabrexfm-tool-2.0.0.release.22.107d16b-1.x86_64      1/1
 +
  Verifying  : gigaio-fabrexfm-tool-2.0.0.release.22.107d16b-1.x86_64      1/1
 +
 +
Installed:
 +
  gigaio-fabrexfm-tool.x86_64 0:2.0.0.release.22.107d16b-1
 +
 +
Complete!
 +
 +
</pre>
 +
 +
=== Setup Auth for fmtool ===
 +
 +
<pre>
 +
[root@vcontroller ~]# echo -n admin@gigaio.com:password1 | base64 > ~/.fabrex_auth
 +
[root@vcontroller ~]# chmod 600 ~/.fabrex_auth
 +
[root@vcontroller ~]# cat ~/.fabrex_auth
 +
YWRtaW5AZ2lnYWlvLmNvbTpwYXNzd29yZDE=
 +
</pre>
 +
 +
 +
=== Install packages on the Compute node ===
 +
<pre>
 +
root@vcontroller ~]# pdsh -w node000[1-2] uptime
 +
node0002:  10:52:39 up 4 days, 18:01,  0 users,  load average: 0.00, 0.01, 0.05
 +
node0001:  10:52:39 up 4 days, 20:20,  1 user,  load average: 0.00, 0.01, 0.05
 +
 +
[root@vcontroller ~]# scp FabreX_v2.0.0_CentOS7.tar.bz2 node0001:/root/
 +
FabreX_v2.0.0_CentOS7.tar.bz2                100% 1060MB  81.5MB/s  00:13
 +
FabreX_v2.0.0_CentOS7.tar.bz2                100% 1060MB  75.7MB/s  00:14
 +
[root@vcontroller ~]#
 +
[root@vcontroller ~]# scp FabreX_v2.0.0_CentOS7.tar.bz2 node0002:/root/
 +
FabreX_v2.0.0_CentOS7.tar.bz2                100% 1060MB  75.7MB/s  00:14
 +
 +
[root@node0001 package]# yum localinstall -y gigaio-support*.rpm --disablerepo='*' --disableplugin='*'
 +
Failed to set locale, defaulting to C
 +
Examining gigaio-support-2.0.0.release.27.ac8ffe6.x86_64.rpm: gigaio-support-2.0.0.release.27.ac8ffe6-1.x86_64
 +
Marking gigaio-support-2.0.0.release.27.ac8ffe6.x86_64.rpm to be installed
 +
Resolving Dependencies
 +
--> Running transaction check
 +
---> Package gigaio-support.x86_64 0:2.0.0.release.27.ac8ffe6-1 will be installed
 +
--> Finished Dependency Resolution
 +
 +
Dependencies Resolved
 +
 +
================================================================================
 +
Package        Arch  Version
 +
                          Repository                                      Size
 +
================================================================================
 +
Installing:
 +
gigaio-support x86_64 2.0.0.release.27.ac8ffe6-1
 +
                          /gigaio-support-2.0.0.release.27.ac8ffe6.x86_64  14 M
 +
 +
Transaction Summary
 +
================================================================================
 +
Install  1 Package
 +
 +
Total size: 14 M
 +
Installed size: 14 M
 +
Downloading packages:
 +
Running transaction check
 +
Running transaction test
 +
Transaction test succeeded
 +
Running transaction
 +
  Installing : gigaio-support-2.0.0.release.27.ac8ffe6-1.x86_64            1/1
 +
Created symlink from /etc/systemd/system/multi-user.target.wants/fabrex-tune.service to /usr/lib/systemd/system/fabrex-tune.service.
 +
  Verifying  : gigaio-support-2.0.0.release.27.ac8ffe6-1.x86_64            1/1
 +
 +
Installed:
 +
  gigaio-support.x86_64 0:2.0.0.release.27.ac8ffe6-1
 +
 +
Complete!
 +
 +
 +
 +
yum remove libfabric-1.7.0-1.el7.x86_64.  # remove existing LIbfabric
 +
 +
yum localinstall -y gigaio-libfabric-*.rpm.  #install with --disablerepo='*' -- disableplugin='*' as it has dependency on libatomic
 +
 +
[root@node0001 package]# yum localinstall -y gigaio-mpich-*.rpm --disablerepo='*' --disableplugin='*'
 +
Failed to set locale, defaulting to C
 +
Examining gigaio-mpich-3.3b2.release.13.f1f48f029.x86_64.rpm: gigaio-mpich-3.3b2.release.13.f1f48f029-1.x86_64
 +
Marking gigaio-mpich-3.3b2.release.13.f1f48f029.x86_64.rpm to be installed
 +
Examining gigaio-mpich-devel-3.3b2.release.13.f1f48f029.x86_64.rpm: gigaio-mpich-devel-3.3b2.release.13.f1f48f029-1.x86_64
 +
Marking gigaio-mpich-devel-3.3b2.release.13.f1f48f029.x86_64.rpm to be installed
 +
Resolving Dependencies
 +
--> Running transaction check
 +
---> Package gigaio-mpich.x86_64 0:3.3b2.release.13.f1f48f029-1 will be installed
 +
---> Package gigaio-mpich-devel.x86_64 0:3.3b2.release.13.f1f48f029-1 will be installed
 +
--> Finished Dependency Resolution
 +
 +
Dependencies Resolved
 +
 +
================================================================================
 +
Package
 +
      Arch  Version
 +
                    Repository                                            Size
 +
================================================================================
 +
Installing:
 +
gigaio-mpich
 +
      x86_64 3.3b2.release.13.f1f48f029-1
 +
                    /gigaio-mpich-3.3b2.release.13.f1f48f029.x86_64      537 M
 +
gigaio-mpich-devel
 +
      x86_64 3.3b2.release.13.f1f48f029-1
 +
                    /gigaio-mpich-devel-3.3b2.release.13.f1f48f029.x86_64 238 k
 +
 +
Transaction Summary
 +
================================================================================
 +
Install  2 Packages
 +
 +
Total size: 537 M
 +
Installed size: 537 M
 +
Downloading packages:
 +
Running transaction check
 +
Running transaction test
 +
 +
 +
Transaction check error:
 +
  file /usr/bin/mpiexec from install of gigaio-mpich-3.3b2.release.13.f1f48f029-1.x86_64 conflicts with file from package slurm-torque-ohpc-18.08.8-4.1.ohpc.1.3.8.1.x86_64
 +
 +
Error Summary
 +
-------------
 +
 +
[root@node0001 package]# yum remove slurm-torque-ohpc-18.08.8-4.1.ohpc.1.3.8.1.x86_64
 +
Failed to set locale, defaulting to C
 +
Loaded plugins: fastestmirror, versionlock
 +
Resolving Dependencies
 +
--> Running transaction check
 +
---> Package slurm-torque-ohpc.x86_64 0:18.08.8-4.1.ohpc.1.3.8.1 will be erased
 +
--> Finished Dependency Resolution
 +
 +
Dependencies Resolved
 +
 +
================================================================================
 +
Package          Arch  Version                    Repository            Size
 +
================================================================================
 +
Removing:
 +
slurm-torque-ohpc x86_64 18.08.8-4.1.ohpc.1.3.8.1  @OpenHPC_1.3_Update8 368 k
 +
 +
Transaction Summary
 +
================================================================================
 +
Remove  1 Package
 +
 +
Installed size: 368 k
 +
Is this ok [y/N]: y
 +
Downloading packages:
 +
Running transaction check
 +
Running transaction test
 +
Transaction test succeeded
 +
Running transaction
 +
  Erasing    : slurm-torque-ohpc-18.08.8-4.1.ohpc.1.3.8.1.x86_64            1/1
 +
  Verifying  : slurm-torque-ohpc-18.08.8-4.1.ohpc.1.3.8.1.x86_64            1/1
 +
 +
Removed:
 +
  slurm-torque-ohpc.x86_64 0:18.08.8-4.1.ohpc.1.3.8.1
 +
 +
[root@node0001 package]# yum localinstall -y kernel-devel-5.3.* --disablerepo='*' --disableplugin='*'
 +
 +
 +
 +
Complete!
 +
[root@node0001 package]# yum localinstall -y gigaio-mpich-*.rpm --disablerepo='*' --disableplugin='*'
 +
Failed to set locale, defaulting to C
 +
Examining gigaio-mpich-3.3b2.release.13.f1f48f029.x86_64.rpm: gigaio-mpich-3.3b2.release.13.f1f48f029-1.x86_64
 +
Marking gigaio-mpich-3.3b2.release.13.f1f48f029.x86_64.rpm to be installed
 +
Examining gigaio-mpich-devel-3.3b2.release.13.f1f48f029.x86_64.rpm: gigaio-mpich-devel-3.3b2.release.13.f1f48f029-1.x86_64
 +
Marking gigaio-mpich-devel-3.3b2.release.13.f1f48f029.x86_64.rpm to be installed
 +
Resolving Dependencies
 +
--> Running transaction check
 +
---> Package gigaio-mpich.x86_64 0:3.3b2.release.13.f1f48f029-1 will be installed
 +
---> Package gigaio-mpich-devel.x86_64 0:3.3b2.release.13.f1f48f029-1 will be installed
 +
--> Finished Dependency Resolution
 +
 +
Dependencies Resolved
 +
 +
================================================================================
 +
Package
 +
      Arch  Version
 +
                    Repository                                            Size
 +
================================================================================
 +
Installing:
 +
gigaio-mpich
 +
      x86_64 3.3b2.release.13.f1f48f029-1
 +
                    /gigaio-mpich-3.3b2.release.13.f1f48f029.x86_64      537 M
 +
gigaio-mpich-devel
 +
      x86_64 3.3b2.release.13.f1f48f029-1
 +
                    /gigaio-mpich-devel-3.3b2.release.13.f1f48f029.x86_64 238 k
 +
 +
Transaction Summary
 +
================================================================================
 +
Install  2 Packages
 +
 +
Total size: 537 M
 +
Installed size: 537 M
 +
Downloading packages:
 +
Running transaction check
 +
Running transaction test
 +
Transaction test succeeded
 +
Running transaction
 +
  Installing : gigaio-mpich-3.3b2.release.13.f1f48f029-1.x86_64            1/2
 +
  Installing : gigaio-mpich-devel-3.3b2.release.13.f1f48f029-1.x86_64      2/2
 +
  Verifying  : gigaio-mpich-devel-3.3b2.release.13.f1f48f029-1.x86_64      1/2
 +
  Verifying  : gigaio-mpich-3.3b2.release.13.f1f48f029-1.x86_64            2/2
 +
 +
Installed:
 +
  gigaio-mpich.x86_64 0:3.3b2.release.13.f1f48f029-1
 +
  gigaio-mpich-devel.x86_64 0:3.3b2.release.13.f1f48f029-1
 +
 +
Complete!
 +
 +
</pre>

Latest revision as of 09:52, 21 August 2020

Introduction

The FabreX Operating System is the software engine that drives the performance and dynamic composability of GigaIO software-defined infrastructure (SDI) for enterprise data centers and high-performance computing environments. This Linux-based, resource-efficient software layers onto FabreX hardware for easy-to-use composing of computing clusters on-the-fly.

The FabreX Operating System is composed of three main components:

  1. Server software,
  2. Switch software,
  3. Management software.

The Server software runs on Linux and provides libfabric functionality for use with MPICH for HPC applications. It also provides a TCP/IP interface for communication between servers. The FabreX Switch software runs on an embedded switch controller and provides dynamic composability and communications between servers.

Packs Description
Composer Pack Single-Host Expansion Includes: FabreX OS + Management Module + GigaIO CLI + Composition Module The Composer Pack features full disaggregation and dynamic composability for deep and wide IO tree

configurations. Multiple FabreX switch configurations enable flexible IO expansion including super servers, cascading, and resource partitioning. FabreX OS also supports GPUDirect for direct access to third-party memory using the standard features of PCIe.

Leader Pack Multi-host Clusters Includes: Everything in Composition Pack + Cluster Module The Cluster Module provides true multi-host to host communication in single-switch systems. The Leader Pack

adds more power for parallel processing and fast access to distributed memory by integrating MPI, NVMe-oF, and TCP/IP.

Maestro Pack Large Multi-Rack Systems Includes: Everything in Leader Pack + Network Pack For true composability in large computing systems where you need more clusters and multiple racks, add the

Network Pack. You’ll have all the benefits of the Leader Pack but also the ability to manage multiple Leaderconfigured switches, for quick scaling of your system to meet your largest workloads.

Software Requirements

Software Version: FabreX OS Version 2.0.0

  • PCIe Gen 4 Switch support
  • Updated Kernel Source
  • Updated Fabric Management
  • NVIDIA's GPUDirect RDMA (GDR):

Downloads

Target Link
Switch FabreX_v2.0.0.gio ]

MD5 = 66f3b9982e9f6d6ab403fd8a1bdf56c3

Server FabreX_v2.0.0_CentOS7.tar.bz2]

MD5 = e92cf05af33a494cedac037e37f40059

Switch Management
  • Ubuntu
    • [1]
    • MD5 = 027c097a5f6aff8df204b87a0074d24a
  • Centos
    • [2]
    • MD5 = ca8926e6e749f70b39cc67dbd340ad5a


Install

Installing the Packages

Verify CentOS 7.7 or later with "Development Tools" installed for air-gapped environment. Copy the tarball to the host.


Go to the directory where the software package was downloaded and untar the files:

tar -xjvf FabreX_<VERSION>_CentOS7.tar.bz2


Enter the extracted directory.

On the head node

fabrexfm-tool help setup the PCIE switch to required configuration . This should be present on the head node or node different from Compute node. Fmtool is stateless hence it doesn't save the configuration on the server-side.

[root@vcontroller ~]# yum localinstall -y gigaio-fabrexfm-tool-2.0.0.release.22.107d16b.x86_64.rpm
Failed to set locale, defaulting to C
Loaded plugins: fastestmirror, versionlock
Examining gigaio-fabrexfm-tool-2.0.0.release.22.107d16b.x86_64.rpm: gigaio-fabrexfm-tool-2.0.0.release.22.107d16b-1.x86_64
Marking gigaio-fabrexfm-tool-2.0.0.release.22.107d16b.x86_64.rpm to be installed
Resolving Dependencies
--> Running transaction check
---> Package gigaio-fabrexfm-tool.x86_64 0:2.0.0.release.22.107d16b-1 will be installed
--> Finished Dependency Resolution

Dependencies Resolved

================================================================================
 Package
       Arch   Version
                    Repository                                             Size
================================================================================
Installing:
 gigaio-fabrexfm-tool
       x86_64 2.0.0.release.22.107d16b-1
                    /gigaio-fabrexfm-tool-2.0.0.release.22.107d16b.x86_64 7.3 M

Transaction Summary
================================================================================
Install  1 Package

Total size: 7.3 M
Installed size: 7.3 M
Downloading packages:
Running transaction check
Running transaction test
Transaction test succeeded
Running transaction
  Installing : gigaio-fabrexfm-tool-2.0.0.release.22.107d16b-1.x86_64       1/1
  Verifying  : gigaio-fabrexfm-tool-2.0.0.release.22.107d16b-1.x86_64       1/1

Installed:
  gigaio-fabrexfm-tool.x86_64 0:2.0.0.release.22.107d16b-1

Complete!

Setup Auth for fmtool

[root@vcontroller ~]# echo -n admin@gigaio.com:password1 | base64 > ~/.fabrex_auth
[root@vcontroller ~]# chmod 600 ~/.fabrex_auth
[root@vcontroller ~]# cat ~/.fabrex_auth
YWRtaW5AZ2lnYWlvLmNvbTpwYXNzd29yZDE=


Install packages on the Compute node

root@vcontroller ~]# pdsh -w node000[1-2] uptime
node0002:  10:52:39 up 4 days, 18:01,  0 users,  load average: 0.00, 0.01, 0.05
node0001:  10:52:39 up 4 days, 20:20,  1 user,  load average: 0.00, 0.01, 0.05

[root@vcontroller ~]# scp FabreX_v2.0.0_CentOS7.tar.bz2 node0001:/root/
FabreX_v2.0.0_CentOS7.tar.bz2                 100% 1060MB  81.5MB/s   00:13
FabreX_v2.0.0_CentOS7.tar.bz2                 100% 1060MB  75.7MB/s   00:14
[root@vcontroller ~]#
[root@vcontroller ~]# scp FabreX_v2.0.0_CentOS7.tar.bz2 node0002:/root/
FabreX_v2.0.0_CentOS7.tar.bz2                 100% 1060MB  75.7MB/s   00:14

[root@node0001 package]# yum localinstall -y gigaio-support*.rpm --disablerepo='*' --disableplugin='*'
Failed to set locale, defaulting to C
Examining gigaio-support-2.0.0.release.27.ac8ffe6.x86_64.rpm: gigaio-support-2.0.0.release.27.ac8ffe6-1.x86_64
Marking gigaio-support-2.0.0.release.27.ac8ffe6.x86_64.rpm to be installed
Resolving Dependencies
--> Running transaction check
---> Package gigaio-support.x86_64 0:2.0.0.release.27.ac8ffe6-1 will be installed
--> Finished Dependency Resolution

Dependencies Resolved

================================================================================
 Package        Arch   Version
                          Repository                                       Size
================================================================================
Installing:
 gigaio-support x86_64 2.0.0.release.27.ac8ffe6-1
                          /gigaio-support-2.0.0.release.27.ac8ffe6.x86_64  14 M

Transaction Summary
================================================================================
Install  1 Package

Total size: 14 M
Installed size: 14 M
Downloading packages:
Running transaction check
Running transaction test
Transaction test succeeded
Running transaction
  Installing : gigaio-support-2.0.0.release.27.ac8ffe6-1.x86_64             1/1
Created symlink from /etc/systemd/system/multi-user.target.wants/fabrex-tune.service to /usr/lib/systemd/system/fabrex-tune.service.
  Verifying  : gigaio-support-2.0.0.release.27.ac8ffe6-1.x86_64             1/1

Installed:
  gigaio-support.x86_64 0:2.0.0.release.27.ac8ffe6-1

Complete!



yum remove libfabric-1.7.0-1.el7.x86_64.   # remove existing LIbfabric

yum localinstall -y gigaio-libfabric-*.rpm.  #install with --disablerepo='*' -- disableplugin='*' as it has dependency on libatomic

[root@node0001 package]# yum localinstall -y gigaio-mpich-*.rpm --disablerepo='*' --disableplugin='*'
Failed to set locale, defaulting to C
Examining gigaio-mpich-3.3b2.release.13.f1f48f029.x86_64.rpm: gigaio-mpich-3.3b2.release.13.f1f48f029-1.x86_64
Marking gigaio-mpich-3.3b2.release.13.f1f48f029.x86_64.rpm to be installed
Examining gigaio-mpich-devel-3.3b2.release.13.f1f48f029.x86_64.rpm: gigaio-mpich-devel-3.3b2.release.13.f1f48f029-1.x86_64
Marking gigaio-mpich-devel-3.3b2.release.13.f1f48f029.x86_64.rpm to be installed
Resolving Dependencies
--> Running transaction check
---> Package gigaio-mpich.x86_64 0:3.3b2.release.13.f1f48f029-1 will be installed
---> Package gigaio-mpich-devel.x86_64 0:3.3b2.release.13.f1f48f029-1 will be installed
--> Finished Dependency Resolution

Dependencies Resolved

================================================================================
 Package
       Arch   Version
                    Repository                                             Size
================================================================================
Installing:
 gigaio-mpich
       x86_64 3.3b2.release.13.f1f48f029-1
                    /gigaio-mpich-3.3b2.release.13.f1f48f029.x86_64       537 M
 gigaio-mpich-devel
       x86_64 3.3b2.release.13.f1f48f029-1
                    /gigaio-mpich-devel-3.3b2.release.13.f1f48f029.x86_64 238 k

Transaction Summary
================================================================================
Install  2 Packages

Total size: 537 M
Installed size: 537 M
Downloading packages:
Running transaction check
Running transaction test


Transaction check error:
  file /usr/bin/mpiexec from install of gigaio-mpich-3.3b2.release.13.f1f48f029-1.x86_64 conflicts with file from package slurm-torque-ohpc-18.08.8-4.1.ohpc.1.3.8.1.x86_64

Error Summary
-------------

[root@node0001 package]# yum remove slurm-torque-ohpc-18.08.8-4.1.ohpc.1.3.8.1.x86_64
Failed to set locale, defaulting to C
Loaded plugins: fastestmirror, versionlock
Resolving Dependencies
--> Running transaction check
---> Package slurm-torque-ohpc.x86_64 0:18.08.8-4.1.ohpc.1.3.8.1 will be erased
--> Finished Dependency Resolution

Dependencies Resolved

================================================================================
 Package           Arch   Version                    Repository            Size
================================================================================
Removing:
 slurm-torque-ohpc x86_64 18.08.8-4.1.ohpc.1.3.8.1   @OpenHPC_1.3_Update8 368 k

Transaction Summary
================================================================================
Remove  1 Package

Installed size: 368 k
Is this ok [y/N]: y
Downloading packages:
Running transaction check
Running transaction test
Transaction test succeeded
Running transaction
  Erasing    : slurm-torque-ohpc-18.08.8-4.1.ohpc.1.3.8.1.x86_64            1/1
  Verifying  : slurm-torque-ohpc-18.08.8-4.1.ohpc.1.3.8.1.x86_64            1/1

Removed:
  slurm-torque-ohpc.x86_64 0:18.08.8-4.1.ohpc.1.3.8.1

[root@node0001 package]# yum localinstall -y kernel-devel-5.3.* --disablerepo='*' --disableplugin='*'



Complete!
[root@node0001 package]# yum localinstall -y gigaio-mpich-*.rpm --disablerepo='*' --disableplugin='*'
Failed to set locale, defaulting to C
Examining gigaio-mpich-3.3b2.release.13.f1f48f029.x86_64.rpm: gigaio-mpich-3.3b2.release.13.f1f48f029-1.x86_64
Marking gigaio-mpich-3.3b2.release.13.f1f48f029.x86_64.rpm to be installed
Examining gigaio-mpich-devel-3.3b2.release.13.f1f48f029.x86_64.rpm: gigaio-mpich-devel-3.3b2.release.13.f1f48f029-1.x86_64
Marking gigaio-mpich-devel-3.3b2.release.13.f1f48f029.x86_64.rpm to be installed
Resolving Dependencies
--> Running transaction check
---> Package gigaio-mpich.x86_64 0:3.3b2.release.13.f1f48f029-1 will be installed
---> Package gigaio-mpich-devel.x86_64 0:3.3b2.release.13.f1f48f029-1 will be installed
--> Finished Dependency Resolution

Dependencies Resolved

================================================================================
 Package
       Arch   Version
                    Repository                                             Size
================================================================================
Installing:
 gigaio-mpich
       x86_64 3.3b2.release.13.f1f48f029-1
                    /gigaio-mpich-3.3b2.release.13.f1f48f029.x86_64       537 M
 gigaio-mpich-devel
       x86_64 3.3b2.release.13.f1f48f029-1
                    /gigaio-mpich-devel-3.3b2.release.13.f1f48f029.x86_64 238 k

Transaction Summary
================================================================================
Install  2 Packages

Total size: 537 M
Installed size: 537 M
Downloading packages:
Running transaction check
Running transaction test
Transaction test succeeded
Running transaction
  Installing : gigaio-mpich-3.3b2.release.13.f1f48f029-1.x86_64             1/2
  Installing : gigaio-mpich-devel-3.3b2.release.13.f1f48f029-1.x86_64       2/2
  Verifying  : gigaio-mpich-devel-3.3b2.release.13.f1f48f029-1.x86_64       1/2
  Verifying  : gigaio-mpich-3.3b2.release.13.f1f48f029-1.x86_64             2/2

Installed:
  gigaio-mpich.x86_64 0:3.3b2.release.13.f1f48f029-1
  gigaio-mpich-devel.x86_64 0:3.3b2.release.13.f1f48f029-1

Complete!