How To Do It with CentOS Linux
I've been running Eskimo North since well before the Internet was. First on a Trs-80 model III, then on a Tandy 16B/6000 running Xenix, then on a Sun 3/180 with Sun/OS 3.5 which grew into a network of Sun's running SunOS 4.1.4, and then RedHat Linux 4.1, 6.2, Aurora Linux 1.99, CentOS 5.2, and now CentOS 6.2.
Documentation has often been less than adequate. I'm hoping I can spare others some of the grief, realize the joy of getting something new and novel to work, and promote the use of Linux and other open source software solutions.
We use Linux in an ISP environment to provide all the necessary services, DNS name service, Radius authentication, ftp and web hosting, e-mail services, shell access, and more.
As of this writing, we are transitioning from a combination of Sparc and UltraSparc hardware and 32 bit Intel hardware running a combination of Redhat 4.2, Redhat 6.2, Aurora 1.99, and CentOS 5.2 onto Intel i7-2600 based hardware running CentOS 6.2.
One major thing we are doing different in this transition is employing Qemu/KVM virtualization to combine the functions that were previously distributed over a much larger number of Sparc systems onto a small number of Intel systems.
The transition from big endian to little indian hardware at the same time as making some major operating system upgrades has also presented some challenges. Many applications store data in native byte ordering and that is causing some problems migrating data from old machines to new.
Anyway, I've battled a number of issues and no doubt will battle some more so I thought by documenting them here, I can help someone else who is facing the same challenges.