Setup machines with ansible - SEAPATH project

This repository contains different playbooks, used to automate the configuration of the machines used in this projetct. The setup is detailled below. To run a playbook, you can use the following command:

ansible-playbook -i inventory.yaml playbooks/<your_playbook>

Setup - inventory

Machines

Three machines are used in this setup:

• A PTP master, used to synchronize the two other machines with the PTP protocol

  • This machine uses a Debian 11 distribution with the ptp4linux package installed

• A sender machine, used to send IEC61850 frames (Sampled Values)

  • This machine must be preinstalled with a linux distribution

  • PTP is running as a service (ptp4l running as a slave), and synchronizes the system clock (phc2sys)

  • RTE’s OUISTITI docker container is installed

• A SEAPATH machine, used to host different VMs, which will receive the frames

  • This machine will be flashed and configured via ansible playbooks

Configure the PTP grand master clock

Run the playbook setup_grand_master.yaml. When the PTP master is started, it serves its system time through the PTP protocol.

Fetch generated images

Images are generated on the build server jean. Two playbooks allow to fetch Host and Guest images from this build server.

SEAPATH machine flash

PXE Repository

In order to automate the flash, the machine which will run the SEAPATH distribution is booted via PXE. The container repository of the Seapath project must be used.

• Github repository : https://github.com/seapath/containers

A DHCP server will set an IP that you need to provide in the rte.yaml inventory. Then seapath-flash-pxe image will be loaded on the machine, and booted.

Flash and "static" setup

When the machine has booted via PXE, use the playbook 2_flash_host.yaml to flash the seapath-host image, and configure the host. A few variables need to be set according to your setup :

• First, in the inventory rte.yaml:

  • In the group pxe_machines:

    • The IP address of the machine, once booted via PXE

• In the group hypervisors:

  • The name of the network interface you want to use for SSH connection

  • The fixed IP address this interface should be assigned to

  • Then, if you need to, change custom kernel boot parameters by editing extra_kernel_parameters. Those will also be changeable with another playbook.

To find the IP of the PXE machine and the name of the network interface, you will need to log into the system with a separate screen or a serial connection on the first time. Once the flash has finished, the playbook will setup the fixed IP, reboot into the host system, setups PTP, and sets custom boot kernel parameters.

Network Configuration

Once all this configuration has been done, the network must be configured with 3_deploy_config.yaml As multiple configurations should be tested, with or without DPDK, with or without SR-IOV, this playbook takes the variable config as a parameter The aim is to be able to deploy different kinds of network on the same host at the same time, unfortunately, this feature has not been developped so far. Currently, available options are just dpdk, sriov and 2nd, to add another network interface.

VM Deployement and setup

• VMs are deployed with 4_deploy_vm.yaml

• This playbook uses the VMs variables defined in the inventory, in the VMs group, and the role deploy_vm. This role customises the xml libvirt configuration file in templates/config.xml.j2. Especially, the following variables can be modified:

  • port: Port used to report logs (detailled later)

  • cpuset: Set of host CPUs used by the guest, e.g. [4,5]

  • ovs_port

  • mac_address

  • guest_image

  • vm_features: Dictionnary of features. Available values are:

    • 'rt'

    • 'isolated'

    • 'dpdk'

    • 'pci-passthrough' and 'sriov' are not currently implemented.

• Before being managed with ansible, VM’s must be assigned a fixed IP. To do so, a recipe was added to the seapath-guest-efi-dbg image in the Yocto Project, adding a fixed IP to the interface enp0s2. On the first boot of a VM, a systemd service launches and adds the following ip address to the interface: 192.168.203.<last byte of the MAC address>. So, be sure to set ansible_host last byte to the last byte of the mac address.

VM Postinstall

SVTools

The SV receiver is provided with a docker container (where the libiec61850 is pre-installed). Build it using the SVTool repository: https://github.com/vermeulenthi/svtools Before using deploy_svtools.yaml, a local registry containing the image must be created on your machine:

docker run -d -p 5000:5000 --name registry registry:2
docker tag svtools_dkimg:latest localhost:5000/svtools_dkimg:latest
docker push localhost:5000/svtools_dkimg:latest

(Following the docker documentation : https://docs.docker.com/registry/) After having launched the local registry, you can use the playbook 5_vm_postinstall. This will setup PTP sync on the VM and deploy svtools. * The role setup_ptp_vm loads the kvm_ptp module, which creates a virtual PTP clock under /dev/ptp0 and starts phc2sys service to synchronize the systemclock with this paravirtualized PTP device. The playbook then deploys svtools software. * Then, svtools is pulled from the registry * Finally, the binary is copied from your local directory to the VM