The installation in the buildtree environment is not useful as all is pre-installed in the image.

sqplus is needed to clean up the database.

Now we need to install XRootD with SSI from our own respository

added a /CANSTART file so that eos waits for all keys to be OK before starting.
/bin/sh -> /bin/bash to allow function names with `-`...
yum and yum-config-manager overridden when in BUILDTREE environment
ctafrontend not booting needs further work

The usage of this user was added earlier. The user and group are created in RPM post install scripts.
For this reason, only the buildtree tests we affected (as the post-install script behavior needed
to be duplicated.

Fixed home directory not readable (prevented the frontend from running).
Added installation of CASTOR's rmcd to the image.

Fixed missing dependencies in image.

Fixed directory name in docker which did not match the top one.
Split the image create to allow fast script fixing.

Also moved image build out of vmBootstap directory.

Refactoring layers to limit their number and catch possible errors using && instead of ; between commands

Cleaning up run scripts for containers: going along Dockerfile and in /opt/run/bin: /shared still available for local scripts developments.

After merge with master, CTA required xrootd >= 4.4.1-1 otherwise ctafrontend xroot plugin is crashing

versionlock the version of EOS that works and cache its files locally so that they are not lost if we remove files from a repo by accident...

using versionlock for eos since the last rpm in citrine-commit repo is broken and FST does not start. I am switching to the tagged version which is tested. Anyway, we ll need to download the RPMs locally and add them to our local repo so that we can reinstall those: they are disappearing after few weeks from jenkins repo and are lost forever otherwise...

solving xroot version conflicts in RPMs... using version provided by eos-citrine-depend repo for all images

add castor repo for tpsrv pod and install cta-cli everywhere so that we have at least some RPMs already installed in the docker image: otherwise it takes too much time to install all rpms in parallel on all the pods as it puts the VM on its knees...

cleaning in same line as install to limit number of layers + remove cta package (installed later so that we have just one docker image

instead of using the jenkins repo (disabvled in the repo file)

Environment variables must be declared with the ENV instruction otherwise those are taken for commands

CONTEXTDIR does not allow to retrieve artifacts from directory under it, therefore move Dockerfile location is the answer (officially)