Wow! That was… unexpected!

After a LONG time of not upgrading my site’s operating system, various initiatives I am undertaking required me to update from Fedora Core 5 (released March 20, 2006) to Fedora Core 9 (released May 13, 2008).

I was very nervous about this upgrade since it spanned so many versions of the operating system, but I started out with high hopes. I backed up my system to DVD using MondoArchive, an excellent bare-metal backup solution. Then, I inserted the Fedora Core 9 DVD, rebooted, and started the upgrade proceedure only to be told that one of my drives was a bit short of space (249 megs, to be exact).

The solution ended up being repartioning-in-place using a great product from Acronis which handled all the heavy lifting of the non-destructive repartition for me. I repartitioned the particular drive to a size large enough to make sure this particular problem will never surface again.

Launching the Fedora Core 9 DVD’s upgrade procedure a second time worked, using the newly-widened partition… The installer updated 890 files and rebooted successfully (I was worried about it rebooting successfully!)

However, that’s when I ran into two things… First, the webserver failed to launch because of a configuration error. Secondly, I ran into a time limit and we had to leave for dinner with friends. So, the site was down for a few hours while we enjoyed a very nice meal and evening with Brian, Milan, and Erin.

We came home, went to bed, and I was lying awake thinking of how to address the issues on the server. Next thing I knew, my feet carried me in to the server room and I started tinkering. An hour later (and about 10 minutes ago), I got the site online at last. But there is still more tweaking and testing to do… I need to ensure that the changes I made manually that got the site online will “stick” after the server is rebooted, which I will attempt to do after posting this article.

It was immediately apparent that the new operating system is considerably faster than the old one. Cold-booting takes less than 40 seconds on Fedora Core 9. I’m not sure how long it took on Fedora Core 5, but I know it was longer.

I apologize for the downtime, but it was necessary, and moving forward should be considerably smoother from now on. There is still a fair amount to do — applying non-critical patches & upgrades to the system and then another comprehensive backup, but being back online is enough for tonight.

UPDATE: Rebooting works… mostly. After rebooting, I need to enter “dhclient eth0”, “tzoupdate” and “service httpd start” to get things rolling properly. I need to determine how to automate those steps so the system reboots to an online state.

UPDATE 2: I’ve got the system configured properly now, so it boots into an online state.