Warren Young wrote:
Feizhou wrote:
If you understand 'automated deployment' to mean just merely clobbering an installation with another Linux distro then I cannot help you.
You're right, we are not using the same terms. I understand "automated deployment" to include things like yum.
yum upgrade or apt distupgrade? ;)
one does not leave out staging the deployment to a staging box/farm
You're restricting yourself to a subset of the real world here.
Of course. Tom said 'servers'. Of course, that does not necessarily mean a farm of servers but hey, if you have enough of them, you don't want to have to fight fires due to human error.
Staging systems do not make sense everywhere. I cannot justify creating a staging system for my tax accountant client, where I installed a file server. There's only the one server. For your idea to work, before doing an OS upgrade, I'd have to have an identical spare machine just sitting around idle for me to test on. Tax accountants are CHEAP. I had a hard time talking the accountant into RAID. A duplicate testing box, too? Please....
Single point of failure boxes need not apply. In this case, I don't even think you worry about security updates so whether it runs Centos or Fedora is not even an issue unless one of them proves to be unstable. Unless you like to have your accounts available on the Net...