Local usage install guide

1.1 pip

If you installed pepc, stats-collect and wult using pip then use the following commands to upgrade to the latest release:

sudo pip3 install --upgrade git+https://github.com/intel/pepc.git@release
sudo pip3 install --upgrade git+https://github.com/intel/stats-collect.git@release
sudo pip3 install --upgrade git+https://github.com/intel/wult.git@release
sudo wult deploy

1.2 Fedora

If you installed pepc, stats-collect and wult using dnf on one of the supported operating systems listed in the installation instructions, then upgrade to the latest version using the following commands:

sudo dnf upgrade pepc stats-collect wult
sudo stats-collect deploy
sudo wult deploy

2.1 Kernel sources

Wult comes with kernel drivers which have to be built for the kernel running on the SUT.

On Fedora and Ubuntu, the sources are typically located in /lib/modules/<kernel_version>/build. The <kernel_version> part must match the kernel running on the SUT. In other words, it has to be the same as what the uname -r command prints when you run it on the SUT.

Unless you are an advanced user, you probably run a stock OS kernel. In this case, it is enough to just install the kernel sources package provided by the OS, e.g.:

• Fedora: kernel-devel package
• Ubuntu: linux-headers package

Advanced users can provide custom kernel sources path via wult deploy command line options (see the wult man page).

2.2 SUT kernel version

Note that wult and ndl both have minimum kernel requirements. ndl requires that the SUT is running a kernel which is 'v5.2' or newer. For wult, the minimum kernel version depends on the method used. See the table below for details on which methods require which kernel version.

2.3 OS packages

Here are the required OS packages.

Fedora

sudo dnf install -y make gcc elfutils-libelf-devel rsync procps-ng libbpf-devel
sudo dnf install -y libffi-devel redhat-rpm-config openssl-devel
sudo dnf install -y kernel-devel
sudo dnf install -y git python3 python3-devel python3-pip python3-numpy
sudo dnf install -y python3-colorama python3-yaml python3-pandas
sudo dnf install -y python3-paramiko python3-argcomplete

Ubuntu

sudo apt install -y make gcc libelf-dev libssl-dev libbpf-dev rsync
sudo apt install -y linux-headers-$(uname -r)
sudo apt install -y procps git python3-pip python3-numpy python3-plotly
sudo apt install -y python3-colorama python3-yaml python3-pandas
sudo apt install -y python3-paramiko python3-argcomplete

Notes

1. If you do not install some of the python projects installing the corresponding python3-* OS packages, they will be pulled by the pip tool later when you install wult. However, we recommend installing numpy from OS packages, because it is optimized for the OS.
2. The git package is required to make it possible installing wult python projects directly from their git repository (see below). Otherwise it is not necessary.

3.1 pip - Recommended

Wult, stats-collect and pepc are Python 3 projects, and we recommend using the pip tool for installing them. Here is how to install them directly from the release branch of their git repositories:

sudo pip3 install --upgrade git+https://github.com/intel/pepc.git@release
sudo pip3 install --upgrade git+https://github.com/intel/stats-collect.git@release
sudo pip3 install --upgrade git+https://github.com/intel/wult.git@release

Note, we do not suggest using the --user option, because in local usage model wult has to be run with superuser (root) permissions, and --user will make this problematic.

It is recommended to install the tools using pip for quickest access to the latest release.

3.2 Fedora

wult, stats-collect and pepc are available on Fedora 38 and Fedora 39. To install pepc, stats-collect, and wult, run the following command:

sudo dnf install -y pepc stats-collect wult

If you are using Fedora 37 or older, use the 'pip' installation method.

3.3 CentOS, Ubuntu & Debian

We do not provide CentOS/Ubuntu/Debian packages, so you'll need to use the 'pip' installation method. Install dependencies using the commands in the os-packages section before continuing with the 'pip' installation method.