How similar should my rpm build machine be to my target deploy machines? Like, do you have to build on a multi-core machine if you plan to run on a multi-core machine? Or as long as the arch is the same, nothing else matters?
johnn
On Tue, Mar 18, 2008, Johnny Tan wrote:
How similar should my rpm build machine be to my target deploy machines? Like, do you have to build on a multi-core machine if you plan to run on a multi-core machine? Or as long as the arch is the same, nothing else matters?
IHMO, it's always best to build on identical machines.
In theory they should work, but I have had problems when building packages where 32-bit AMD and 32-bit Intel systems had some incompatibilties that resulted in illegal instruction traps.
One package I know gave me this was gmp, GNU Arbitrary Precision Arithmetic Library, required by clamav.
Pretty much everything we build, we do with the OpenPKG portable package management system where the basic philosophy is to make it easy to build on the target machines thus avoiding this type of problem.
Bill -- INTERNET: bill@celestial.com Bill Campbell; Celestial Software LLC URL: http://www.celestial.com/ PO Box 820; 6641 E. Mercer Way FAX: (206) 232-9186 Mercer Island, WA 98040-0820; (206) 236-1676
If the future navigation system [for interactive networked services on the NII] looks like something from Microsoft, it will never work. (Chairman of Walt Disney Television & Telecommunications)
Johnny Tan wrote:
How similar should my rpm build machine be to my target deploy machines? Like, do you have to build on a multi-core machine if you plan to run on a multi-core machine? Or as long as the arch is the same, nothing else matters?
Sounds like a question for a more rpm specific list.
For CentOS, I'd recommend you just match the arch, and use mock to setup buildroots to build in. And make sure redhat-rpm-configs is installed in the buildroot. Apart from that you should be fine.
-
Bill Campbell -
Johnny Tan -
Karanbir Singh