/
PerfSONAR Install Guide Using Package Manager

PerfSONAR Install Guide Using Package Manager

Setting up a PerfSONAR Node using the Package Manager
Introduction: This document provides instructions on setting up perfSONAR nodes on an already existing Linux computer using the Linux package manager. Use these instructions if you are using an already existing Linux server.

Requirements:

  • System requirements of the node:
    • OS
      • A RedHat variant computer (Rocky or Alma Linux) version 9
      • Debian 11-12
      • Ubuntu 20-22
    • A network interface with a public IP address
    • Min: 4 cores
    • Min: 8 GB RAM
    • Min: 2 GHz (general); 2.8 GHz (10G path)
  • A super-user (sudo) account on the node

Rocky Install

  1. Start by configuring dnf. Run the following commands.

    sudo dnf install epel-release
    sudo dnf config-manager --set-enabled crb
    sudo dnf install http://software.internet2.edu/rpms/el9/x86_64/latest/packages/perfsonar-repo-0.11-1.noarch.rpm
    sudo dnf clean all
  2. To install the full perfSONAR suite, which includes the webUI and local storage, run the following command.

    sudo dnf install perfsonar-toolkit

Debian/Ubuntu Install

  1. Configure apt to pull from PerfSONAR Repos.

    curl -o /etc/apt/sources.list.d/perfsonar-release.list http://downloads.perfsonar.net/debian/perfsonar-release.list
    curl -s -o /etc/apt/trusted.gpg.d/perfsonar-release.gpg.asc http://downloads.perfsonar.net/debian/perfsonar-release.gpg.key
    1. On Ubuntu, you may also need to enable the universe repository with the following command.

      sudo add-apt-repository universe
  2. Update apt.

    sudo apt update
  3. To install the full perfSONAR suite, which includes the webUI and local storage, run the following command.

    1. Note that the password variable is necessary to install OpenSearch correctly. It will be automatically replaced with a more-secure random password once installed.

      env OPENSEARCH_INITIAL_ADMIN_PASSWORD=perfSONAR123! apt install perfsonar-toolkit

Post-Install Configuration

  1. Configure an admin account for the webUI by running the following command

    sudo /usr/lib/perfsonar/scripts/nptoolkit-configure.py
    1. When prompted with the screen below, type in "2" and hit "ENTER"
    2. When prompted with the screen below, type in "1" and hit "ENTER"
    3. Enter the username and password for the admin account you want to create. Make sure to pick a strong password.
    4. Enter "0" a couple of times to exit the script and return to the command line
  2. (Optional) Check that the following services are running using the following command

    systemctl status SERVICE-NAME
    1. The following are service names you can use in the above command to ensure the node is up and running
      1. pscheduler-scheduler
      2. pscheduler-runner
      3. pscheduler-archiver

      4. pscheduler-ticker

      5. psconfig-pscheduler-agent

      6. owamp-server

      7. twamp-server

      8. perfsonar-lsregistrationdaemon

Finalizing perfSONAR Setup (via WebUI)

  1. You can now log in via the web UI at "https://<hostname or IP address>/toolkit" using the web admin credentials you made in step 4.
  2. Click the "Configuration" button in the top right-hand corner. You will be prompted to enter in your web administrator username and password you made previously.
    1. Here, you can add information about your organization name, administrator name, administrator email, etc.
    2. You'll need to mark the checkbox "I agree to the perfSONAR Privacy Policy" and then click the "Save button" at the bottom of the screen to save these changes.

  3. After some time (up to 24 hours but usually within an hour or so) you should be able to go to the perfSONAR global registry (http://stats.es.net/ServicesDirectory/+) and search for your node.
    1. You can either use the search bar to find your node or scroll on the online map to where your node is physically located to find the kit
  4. You should now be set up and can begin adding tests to your new perfSONAR node.

Configuring tests with other hosts (via Web UI)

  1. Click the "Configuration" button in the top right-hand corner to configure tests with other hosts.

     














   2. Click on "Tests"

  3. Click on "+ Test" to add a test.

  4. Select the type of test.

  5. Enter the test name/description and add Test Members

  6. Enter Hostname/IP and Host description (optional) and click add host.

 7. Click OK.

At this point, your PerfSONAR node will start testing with the hosts added and you should be able to see the results in the "Test Results" section of your toolkit homepage.

Configuring Tests with KYCyberteam

To configure tests, you need to download the configuration file given below.

  1. Contact a member on the KYCyberteam to let them know you want join the KYCyberteam tests
    1. Current contacts for the team
      1. Tyler Burkett (tyler.burkett@uky.edu)
    2. Make sure to include the IP address and hostname of the perfSONAR node you want to use
  2. Edit the NTP configuration on your node to (usually /etc/ntp.conf or /etc/chrony.conf) to set "maddash.ccs.uky.edu" as a source (preferably the only source).
    1. Restart the NTP service with one of the following

      sudo systemctl restart ntpd
      sudo systemctl restart chronyd
  3. On the command line after you receive confirmation from the administrator, run the following commands to add the KYCyberteam test configuration file to your perfSONAR node.
    1. Substitute the URL in the command with the one provided by the administrator.

      sudo psconfig remote add --configure-archives "https://maddash.ccs.uky.edu/psconfig/kycyberteam-config.json"
  4. Run the following command to restart pScheduler, the service which reads the configuration file and automatically begins running the tests.

    sudo systemctl restart psconfig-pscheduler-agent 
  5. After a couple of minutes, you should be able to see test results in the "Test Results" section of your perfSONAR web interface.



For more information, visit the links below:


Related content

Center for Computational Sciences