NiftyClusters T Mitchell wrote: > > If there are 10000 files it might look at 10000 different places > including device names.... kernel names, shell features, kernel > modules and more. > i.e all the things that 'configure' might know about and more. Yeah, that's my main worry. > /etc/redhat-release is the most common. N.B. You may need to restore > the CentOS words often to permit CentOS updates to do the right thing. I've been thinking about that too, ever since I did a yum update yesterday on my test machine. I had the centos-line commented out and instead had a rhel-line in there. AFAICT, the update went through fine, incl a kernel update. But this might be a potential gotcha'. > If /etc/redhat-release is not the answer you may have to look harder > at the failing process with strace or even SELinux tricks to see > what it does look at. Use SELinux tricks even if I don't have it installed? I'm not that familiar with SELinux, except for knowing it's a security hardened something or other. > If it is Oracle, Given the price of Oracle -- just purchase the RH > product. > It is common that the expensive packages are the most restrictive and > putative. Nah, not Oracle. We run Orcacle on Windows here. It's a constant pain for our db-admin. Main concern are a handful of proprietary molecular building and calculation suites. The *nix-group here uses computer aided drug design. There are db's involved but they are installed and run from each software suite, w/o involvement of Oracle or some such. -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/x-pkcs7-signature Size: 5118 bytes Desc: not available URL: <http://lists.centos.org/pipermail/centos/attachments/20081002/ca069417/attachment-0004.bin>