As anyone reading this is amongst those most likely to notice/complain, I thought I would just let you know that we're going to have some downtime for Goats this week.

I've decided to finally upgrade our old database machine from the MySQL [mysql.com] 3.23 lineage to the fancier/newer 4.0 line. I figured while I'm doing that, I should probably also upgrade from the 1999 installed Red Hat 6 to something a tad newer. This means I'm going to take everything down, wipe the machine, and start it over.

So, at some point today, I'm going to take down the site so I can do a "test" copy of the database onto an old desktop machine running linux I have around here. If I can get that running MySQL 4.0, then either tomorrow or Thursday we'll have some very extended downtime as I wipe the live machine and get it all upgraded.

The machine had an ...issue... as you all noticed a few days ago, when it spontaneously restarted and then decided that daylight savings is for wusses. So I decided to drop all the other fixes that I need to get done, and just get that machine happy before anything else.

I'll post a comment in this thread about an hour before actual downtime as the process continues. Obviously, I won't be able to post, and you won't be able to read while everything is down.

I should probably also upgrade from the 1999 installed Red Hat 6 to something a tad newer.

To Fedora Core 1, for example?

In Response to mattdm (#14000):

Actually, I was torn between going with the RH 8 CDs I had, and going for Enterprise 3.

The Star Trek geek in me won.

For your sakes, I thought that maybe Fedora wasn't the way to go.

I think I'll be ready to irritate you all around 11.30 or 11.45 today. This should be pretty quick. I just need to clean up all the DB connections, shut it down, copy all the files off of the machine, and then restart the DB/web servers.

In Response to phillip (#14003):

well....that took longer than expected.

There were a couple of issues with SSH compatibility for scp, that I probably should have anticipated beforehand, and man...copying over 4 Gigs of data takes a while....even on a local network.

Now to see if I can get those files working on a different machine....

It looks like everything should work. So the site will be down for between 6-8 hours tomorrow (at a guess).

My plan is to take down the servers to start backup procedures ~8:00 AM. With luck, things should be back up by 2 PM-ish. In reality, this stuff never goes right, so it's more likely that the site will be back up around 8, But I really hope not, because I have an appointment with a beer at 6.

In Response to phillip (#14039):

Bah, if you're really pressed for time, go pick up the beer, bring it back to the server room and show it how you're saving the lifestyles of millions of goats-fans. Beer really swoons at that sort of thing.

In Response to unFalln (#14043):

your assumption that we have a "server room" instead of an overheated, really noisy, black box of machines taking up too much space in my apartment seems... ill-founded.

the problems is that somehow going to "pick up the beer, [and] bring it back to the server room" is less ...interesting... when that just means a trip to the kitchen.

In Response to phillip (#14067):

well.

that was hell.

I'll fill in some of what sucked about it tomorrow, but now, I need beer.

We apologize for the inconvenience.

Today's coding appears to be a bit dodgy - the strip can be clearly viewed at http://goats.com/comix/0402/goats040213.png and yet www.goats.com brings up the 11th Feb strip. What gives, o computery godly one?

In Response to Lonely Goatherd (#14079):

I gots no problems with www.goats.com [goats.com]. Maybe you're just bringing up a cached copy. Or maybe you need to register. Whatever.

Can somebody reach Zamphir? I have a bad feeling that he is passed out on the floor atop a pile of half-eaten, squashed Danish pastries that have been scattered by his jittering DT's.

Phillip, how could you do this to him? Imagine not having a beer for 10 whole hours.

The horror.

The horror.

In Response to phillip (#14068):

ok....short version.

shutting down/backing up/copying the database all worked flawlessly. I was a happy camper....and then things went to hell.

The overall problem is that this is ancient hardware, circa early 1999. So, after spending 4-5 hours trying to install RHEL, updating the BIOS on the motherboard, updating the firmware on the RAID controller, all to versions from ~2000 instead of 1998, I finally came across this bug report [redhat.com] on RedHat's site that talked about similar problems to what I was seeing, but with RH 7. The key, though, was apparently to ignore the Release Notes for what I was installing, and instead read those for RedHat 9 [redhat.com].

If you scroll all the way to the bottom, you see:

Systems with the 440GX chipset are supported only on a best-effort basis. Therefore, we welcome bug reports regarding systems with the 440GX chipset, but may or may not be able to resolve them.

As I said, we're running ancient hardware...in fact, we don't even have Intel's 440GX chipset. We have Intel's N440BX chipset. I'm guessing that's approximately 5 versions earlier. So if they gave up on the G, the B is not likely to get too far.

(In reality, credit for uncovering most of that goes to Mike Stuhlmiller, who I definitely owe some beer to.)

At that point, I basically had to decide between compiling my own kernel, which I have no idea how to do, or just installing RedHat 8, which supports our hardware.

The determination was that learning is a waste of my time, and that all of you would probably appreciate having the site up more than me learning to compile Linux.

RH 8 installed flawlessly, I got/installed) all the updates from RedHat since then, installed MySQL (on the server, and updated the clients on all the web servers), copied the database back..and within 2.5 hours of deciding to give up on modern technology, I was on my way to the pub....with some amount of certainty that the site was probably working.

In Response to deerboy (#14087):

The twitching has mostly subsided now.