To avoid meltdowns like we had in JBQA-2144, to reduce disk usage, and to speed up the rollout of new hudson jobs, I propose we create two folders:

/home/hudson/hudson_workspace/workspace/jbosstools-3.0-jbdevstudio-2.0-REQUIREMENTS
/home/hudson/hudson_workspace/workspace/jbosstools-3.1-jbdevstudio-2.1-REQUIREMENTS

In there, we will place a copy of all the requirements used by the paired builds, and instead of burning time and bandwidth fetching the 3G of zips every time we start a new job, we'll simply symlink to that folder so new reqs will be added.

cd /home/hudson/hudson_workspace/workspace/jbosstools-nightly-3.0.x/jbds-build
ln -s /home/hudson/hudson_workspace/workspace/jbosstools-3.0-jbdevstudio-2.0-REQUIREMENTS requirements

Thus, for the 4 parallel jobs for JBDS2.0.0/2.0.1 & JBT3.0.0/3.0.1, we'll use about 3G of disk instead of 12. For the 2 parallel jobs for JBDS2.1.0 & JBT3.1.0, we'd save 3G.

For the case where this is run on a local box (eg., my own Hudson) we just need a catch that says "if /home/hudson/hudson_workspace/workspace/jbosstools-3.1-jbdevstudio-2.1-REQUIREMENTS doesn't exist, create it, symlink to it, and begin downloading" .. of course the path there would be ~/.hudson/jobs/jbosstools-3.1-jbdevstudio-2.1-REQUIREMENTS, but it's the same relative path.