The Trace-Cruncher project aims to provide an interface between the existing instrumentation for collection and visualization of tracing data from the Linux kernel and the broad and very well developed ecosystem of instruments for data analysis available in Python.
The Trace-Cruncher allows for sophisticated analysis of kernel tracing data via scripts, but it also opens the door for exposing the kernel tracing data to the instruments provided by the scientific toolkit of Python like MatPlotLib, Stats, Scikit-Learn and even to the nowadays most popular frameworks for Machine Learning like TensorFlow and PyTorch. The Trace-Cruncher is strongly coupled to the libtraceevent, libtracefs and KernelShark projects and is build on top of the C API of these librearies.
Trace-Cruncher has the following external dependencies: libtraceevent, libtracefs, trace-cmd, KernelShark, Json-C, Cython, NumPy.
1.1 In order to install all packages on Ubuntu do the following:
sudo apt-get update
sudo apt-get install build-essential git cmake libjson-c-dev -y
sudo apt-get install libpython3-dev cython3 python3-numpy python3-pip -y
sudo apt-get install flex valgrind binutils-dev pkg-config -y
sudo pip3 install --system pkgconfig GitPython
1.2 In order to install all packages on Fedora, as root do the following:
sudo dnf install gcc gcc-c++ git cmake json-c-devel -y
sudo dnf install python3-devel python3-Cython python3-numpy python3-pip -y
sudo dnf install flex valgrind -y
sudo pip3 install --system pkgconfig GitPython
2 In order to install all third party libraries do the following:
git clone https://git.kernel.org/pub/scm/libs/libtrace/libtraceevent.git/
cd libtraceevent
make
sudo make install
cd ..
git clone https://git.kernel.org/pub/scm/libs/libtrace/libtracefs.git/
cd libtracefs
make
sudo make install
cd ..
git clone https://git.kernel.org/pub/scm/utils/trace-cmd/trace-cmd.git/
cd trace-cmd
make
sudo make install_libs
cd ..
git clone https://git.kernel.org/pub/scm/utils/trace-cmd/kernel-shark.git/
cd kernel-shark/build
cmake ..
make
sudo make install
cd ../..
Installing trace-cruncher is very simple. After downloading the source code, you just have to run:
cd trace-cruncher
make
sudo make install
To execute the unit tests run the following from the trace-cruncher/tests directory:
sudo python3 -m unittest discover .
trace-cruncher can be installed using Docker. Run the following as root from the cloned repository:
docker build --no-cache -t trace-cruncher-image - < Dockerfile
We recommend using the --no-cache flag to ensure git updates are not missed in the build, as well as the --squash flag to shrink the final image size by removing the build dependencies.
When running trace-cruncher as a docker container, the container must be started with the --privileged flag to give trace-cruncher access to the host kernel. See below:
docker run --privileged --name trace-cruncher-container -it trace-cruncher-image
For questions about the use of Trace-Cruncher, please send email to: [email protected]
For bug reports and issues, please file it bugzilla
The trace-cruncher project team welcomes contributions from the community. For more detailed information, refer to CONTRIBUTING.md.
This is available under the LGPLv2.1 license.