Hi Ewan, The ATLAS images used on various cloud resources are preconfigured SL6. They don't provide the hooks needed to contextualise as LHCb does. There is some activity to look at uCernVM but at this stage running on Vac seems beyond the effort we have available. Personally I like the idea but the best we can do is make decisions to not exclude the Vac model. If you have time to look into the details then we can help. One aspect are the wrapper scripts which are being re-jigged in ATLAS now. Cheers, Peter On 28 February 2014 16:37, Ewan MacMahon <[log in to unmask]> wrote: > Hi all, > > As mentioned in the Friday future technology meeting, we have an > install of Andrew's Vac system at Oxford that's doing good work > running LHCb jobs. > > However, I'd quite like it to be running ATLAS jobs too. We have > talked in some of the previous meetings about what it would take > to get (something akin to) the current ATLAS cloud images running > on Vac, but I'd actually like to try to push forward on this a bit > if we can. > > The LHCb 'images' are actually just plain cernvm images, with a > contextualisation script that sets up all the LHCb bits from > cvmfs; it's all very nice. What do the current ATLAS cloud images > look like? Is there any chance of being able to take a similar > approach, or will we need dedicated ATLAS images? > > Ewan