ADACS Job Controller Client

Project setup

1. Clone the repository git clone https://gitlab.com/CAS-eResearch/GWDC/code/adacs_job_controller_client.git
2. Init all submodules git submodule update --init --recursive
3. Initialise the boost installation cd src/third_party && bash prepare_boost.sh
4. Initialise the elfutils installation cd src/third_party && bash prepare_elfutils.sh

Building

For Development

The project can now be built. There are two primary targets: Boost_Tests_run, and adacs_job_client. You'll likely only want to bother building Boost_Tests_run and using this to test the functionality of the job client.

1. Make a build directory cd src && mkdir build
2. Change in to the build directory and run the cmake configure:
   • Basic build: cd build && cmake -G Ninja -DCMAKE_BUILD_TYPE=Debug ..
   • With clang-tidy: cmake -G Ninja -DCMAKE_BUILD_TYPE=Debug -DENABLE_CLANG_TIDY=ON ..
3. Build the target ninja Boost_Tests_run

Optional cmake flags:

• -DENABLE_CLANG_TIDY=ON: Enable clang-tidy static analysis during build
• -DCMAKE_BUILD_TYPE=Debug: Build in debug mode (default)
• -DCMAKE_BUILD_TYPE=Release: Build in release mode with optimizations

NB. To run the full test suite, it may be possible that a db.sqlite3 database is required with prepopulated tables. To do this, cd src/Lib/schema && . venv/bin/activate && python manage.py migrate then copy the db.sqlite3 database to wherever the Boost_Tests_run binary is.

For Deployment

Since a primary objective of this project is to statically link a binary that doesn't rely on any system libraries (Except core libc/libdl etc) and that can run on ancient glibc versions, we use a docker build using centos 7 to build the production binary. There is a convenience script for this.

bash scripts/build.sh

This will generate a directory out which contains the executable adacs_job_client which can be shipped and deployed.