Docker image usage

Launch a DaVinci container

The shortcut to run a DaVinci docker in this project is:

make docker-dv

in the project root.

The make rule is actually defined as the following:

docker run --rm -it -v `pwd`:/data -e UID=$(id -u) -e GID=$(id -g) --net=host umdlhcb/lhcb-stack-cc7:DaVinci-v45r6-SL

More generally, if you want to mount another folder, with a path <src_path>:

docker run --rm -it -v <src_path>:/data -e UID=$(id -u) -e GID=$(id -g) --net=host umdlhcb/lhcb-stack-cc7:DaVinci-v45r6-SL

Info

docker images are read-only. When we launch a docker container, a writable layer is placed on top of the image.

Use official DaVinci inside container

The command is very similar to the one used on lxplus:

lb-run DaVinci/latest gaudirun.py <ntuple_options1> <ntuple_options2> ...

since we have only 1 version of DaVinci inside the container, DaVinci/latest would always work.

Use locally compiled DaVinci inside container

Since we need additional TupleTool¹ in our reconstruction script, we also provide a locally compiled DaVinci inside the container, provided that you are using a DaVinci image with a -SL suffix.

To launch this locally compiled DaVinci:

run gaudirun.py <ntuple_options1> <ntuple_options2> ...

Note

To make our lives easier, a run.sh command is typically created in each stripping lines, such as:

run1-rdx/run.sh
run2-rdx/run.sh

If you see this file, take a look at it to see the availabe commands, and run it inside the docker container:

./run.sh <path_to_cond_file>  # inside docker

optionally, a debug flag can be supplied so that only the first couple of events will be considered.

Remove old docker images

To list all local docker images:

docker images

If you download a newer version of docker image, and want to remove the old ones to save disk space, you can:

docker rmi <docker_image_tag>
docker image prune

here <docker_image_tag> should be something like umdlhcb/lhcb-stack-cc7:DaVinci-v45r6-SL.

---

1. Such as TupleToolApplyIsolation and TupleToolTauMuDiscVars. ↩