tag:blogger.com,1999:blog-6996317921980758688.post3391784165451072479..comments2022-07-15T06:34:26.145+01:00Comments on NGS: Good news, bad newsDavid Meredithhttp://www.blogger.com/profile/14618673864658094727noreply@blogger.comBlogger1125tag:blogger.com,1999:blog-6996317921980758688.post-1017394603475395762011-05-20T16:33:27.688+01:002011-05-20T16:33:27.688+01:00For sake of completeness I should probably point o...For sake of completeness I should probably point out that for full-blown gLite style job submission the HPC worker nodes will need (access to) the glite-WN tools, otherwise they won't be able to access any grid resources.<br /><br />There are several 'shared' clusters running in GridPP as well as the dedicated PP ones, and a common model is to install the 'tarball' copy of the worker node tools on an NFS server, then have all the worker nodes mount it. It allows the grid stuff to be kept on a system the grid folks have access to, with minimal changes needing to be made to the worker nodes by the HPC admins.<br /><br />It sounds like what you're doing is quite similar to Edinburgh's ECDF system and Birmingham's 'Bluebear'.Ewanhttps://www.blogger.com/profile/17124871481877068938noreply@blogger.com