<div dir="ltr">On Mon, Nov 19, 2018 at 2:57 PM Frank Thommen <<a href="mailto:f.thommen@dkfz-heidelberg.de">f.thommen@dkfz-heidelberg.de</a>> wrote:<br><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">But this also means ignoring reality, as there can be very good reasons <br>
not to update (or to update with considerable delay). E.g. we run a big <br>
HPC cluster based on CentOS and we /cannot/ update due to drivers for <br>
special hardware (Infiniband) which usually is one or two point releases <br>
behind.<br></blockquote><div><br></div><div>If that is the case, you should be running your own staged/versioned repos of CentOS and elrepo repositories. Otherwise, there's no easy way to reproduce an HPC node from scratch or to add new nodes using the same configuration. I'd assume you'd want to do that anyway, since loading/updating nodes is much easier if you're running against a local source with fast interconnect.</div><div> </div></div>-- <br><div dir="ltr" class="gmail_signature" data-smartmail="gmail_signature">Jonathan Billings <<a href="mailto:jsbillin@umich.edu" target="_blank">jsbillin@umich.edu</a>><br>College of Engineering - CAEN - Unix and Linux Support<br><br></div></div>