To upgrade, or not to upgrade...

Those that follow Linux news will certainly know that Ubuntu 8.04 "Hardy Heron" is coming at the end of April. This release is a major milestone for Ubuntu, because it's the second Long Term Support (LTS) release, after 6.06 "Dapper Drake". That means they'll be keeping it up to date for 18 months as normal, but will provide security updates for 3 years on desktop installations and 5 years for servers.

For those wondering, yes, I do host my website on my own server and it is running Dapper. Previously I had made up my mind that I was going to continue using Dapper for the duration of it's support cycle mostly because it is proven and stable but also because it's a major hassle to upgrade. I'm reconsidering this for a number of reasons, but the main one is simply because I enjoy being on the bleeding edge, and a lot of things on my Dapper server are very out of date and won't ever be updated unless I can find backports, or compile programs myself, neither of which I want to do.

So what to do? I just can't decide. Upgrading will involve having to move everything off to a temporary server while I do a fresh install of Ubuntu on my main server. Then I'll need to remember all the little tweaks and customizations I've done especially in the areas of Apache/MySQL/PHP performance. I'll have to import and setup my shell scripts, my cron jobs. Oh and let's not forget the shell accounts I've given to some of my friends. SSH keys and authorized user files since I exclusively use public-key authentication. And what if my PHP applications are incompatible with newer versions of PHP and MySQL? There are lots of questions that won't be answered until I get a chance to test the environment on another computer, which I hope to do soon.

Simply put, there is a lot of work to do if I upgrade, so that leaves me wondering: what's the point? Logical thinking tells me that newer versions of the applications I'm using will have newer and better features as well as bugfixes that were not deemed critical enough to be incorporated into older, legacy versions. New features in the Linux kernel will add to the functionality of the server, and maybe even give a performance increase.

And to summarize the cons: There may be incompatibilies. I will certainly forget a few of my customizations which could lead to lower performance. Data could be lost. I'm sure there are more I haven't even thought of. Most of all, if it isn't broken, why fix it, right?

For the time being, I am still undecided. I'll have to see how my tests go to get a better idea of whether or not this upgrade would be worth the time, effort, and risk. I'll also take others' opinions into consideration, go ahead and comment on this post.