(Click to open topic with navigation)

You are here: Installation and Configuration > Manual Installation > Upgrading > Upgrade Nitro

2.9 Upgrading Nitro

This topic contains the steps and procedures to follow to upgrade Nitro using the Manual upgrade method.

In this topic:

2.9.1 Upgrade from a Version Prior to 2.0

Beginning with Nitro 2.0, the licensing procedure changed to use an RLM server. If your company already uses an RLM Server, you can skip this procedure.

The following steps are required if you are upgrading a Nitro version prior to 2.0.

  1. Install or obtain access to an RLM server. See 2.5 Installing RLM Server.

    Beginning with Nitro 2.0, the licensing procedure changed to use an RLM server. If your company already uses an RLM Server, you can skip this procedure.

  2. Obtain and install the Nitro license. This requires access to an RLM server. See 2.6.1 Obtain a Nitro License.
  3. Copy the license file to each compute node (coordinator). On each compute node, or on the shared file system, do the following:
    [root]# cp <licenseFileName>.lic /opt/nitro/bin/

2.9.2 Upgrade Nitro

On the host where Nitro resides, do the following:

  1. If you have not already done so, complete the steps to prepare the host. See 2.4 Preparing for Manual Installation or Upgrade.
  2. Back up your existing launch script in /opt/nitro/bin/.

  3. Change the directory to the root of the unpacked Nitro tarball bundle.

    [root]# cd nitro-tarball-bundle-<version>-<OS>
  4. Identify the Nitro product tarball (nitro-<version>-<OS>.tar.gz) and unpack the tarball into the same directory you created when you first installed Nitro (for example, /opt/nitro).
    [root]# tar xzvpf nitro-<version>-<OS>.tar.gz -C /opt/nitro --strip-components=1
  1. Identify the launch_nitro.sh script version for your resource manager.

    Reference scripts are provided in /opt/nitro/scripts.

    [root]# find . -name launch_nitro.sh
./scripts/lsf/launch_nitro.sh
./scripts/torque/launch_nitro.sh
./scripts/slurm/launch_nitro.sh
./scripts/alps/torque/launch_nitro.sh
./scripts/alps/slurm/launch_nitro.sh
  2. Copy the latest launch script to the bin directory. (This example uses the Torque-based launch script.) 
    [root]# cp /opt/nitro/scripts/torque/launch_nitro.sh /opt/nitro/bin/launch_nitro.sh

    This is a "copy" file operation and not a "move" operation. This allows you to customize your version of the script and always have the factory version available for consultation and/or comparison.

  3. Merge any customizations from your existing launch_nitro.sh script into the script you just copied to the bin directory.
  4. If you are not using a shared file system, copy the updated Nitro installation directory to all hosts.

    Only the Nitro bin directory with its proper path is required to run Nitro jobs. This means that you only need to copy the Nitro bin directory to the other hosts.

    [root]# scp -r /opt/nitro/bin root@host002:/opt/nitro
nitrostat                                       100%   12KB  12.0KB/s   00:00
launch_nitro.sh                                 100% 6890     6.7KB/s   00:00
nitro                                           100%   15MB  14.9MB/s   00:00

2.9.3 Verify Network Communication

Verify that the nodes that will be running Nitro are able to communicate with the Nitro ports and that the nodes are able to communicate with one another.

© 2016 Adaptive Computing