Well, the main eC site is once again alive and well. Everything seems to be functioning properly.

If I recall correctly, I left you all off with the dist-upgrade in my last entry. So here’s what happend next.

The dist-upgrade went smoothly. When we went to install a new kernel (and thus, need to reboot), all hell broke out. There was a small error in the lilo config (forgot to define root). After a few panics, we got that sorted out, and all was well. For a few minutes.

With 2.6 running, I removed some extra packages and then installed Apache 1.3 to put up a “Just a bit longer, kids!” page. After that, I did the first firewall ruleset. When I activated this rule set, the server promptly dropped off of the internet. A bit too secure, perhaps?

I ran the ruleset by the suave and debonaire linuxwolf from EFnet, and he made a number of modifications. After about two hours, EV1 finally rebooted the server, and the new ruleset went in with minor tweaking. (Like fixing stop. When I say stop, I mean stop filtering, not stop networking. Damn ruleset.)

Five hours of sleep, and right back to work. I started the backup uploads from Kato to Plain, then slimmed down the apache1.3 config by about 600 lines. It had a few errors, but nothing that couldn’t be fixed in a minute or two. So, with Apache 1.3 running (and supporting virtual hosts, hooah!), I began on MySQLd.

Oh boy, was that sure fun. It installed no problem, but when it came time to restore the db backup…

You see, the backup was ~2gb text file contaning all of Plain’s databases. One. Single. 2GB. File. This wouldn’t be a problem under normal circumstances, but MYSQLDUMP ALLOW SYNTAX ERRORS WHEN IT WAS CRAPPING IT OUT.

Seriously, the next time I see a MySQL AB employee, he’s getting beat with the stupid stick. SInce it had about 10 errors spread out between 7.5 million lines, I did not have much fun.

I opened it on the server with nano, which took about twenty minutes of 70% CPU usage and 80% memory useage on the part of the text editor process. After that, I had to search for each CREATE TABLE section and check the column names against the mysql reserved word list. Then I wrote the file out, which took forty minutes. Of course, I didn’t fix all of the errors at once (stupid me), so this process took eleven hours.

Finally MySQL was set up. I installed php4-mod for Apache, then started her up. Behold, eCri…wait, what the fuck? The pages using the database were all blank…

Leif popped on from some Windows XP box in CT at this point and told me to install php4-mysql. Hah, whoops.

NOW, behold, eCritters! For…one minute? What the fuck?

Well, apache has decided to stop serving requests after about a minute. I beat the config file to death with a spoon, but it would still keep getting lazy after a minute. Not knowing what to do, and being 3am, I got SIX hours of sleep (yay!).

So I woke up, prodded the site to see if EV1 was just having network trouble last night, and found the same problem. So I got pissed off, dyked Apache 3.1 out, slapped Apache 2 on the server, whacked the php modules into line, and modified the config files slightly.

Tada. Instant eC. And it works. I think there’s room for some small improvement, but damnit it works. I’ll tune it up later.

After this I realized that register.php needed GD, so I installed that and restarted apache. No problems, and register displayed fine.

Restocks, on the other hand…Hah, I have no idea if any of the crons are going to work. I mean, are they coded as part of the site? Or do they need a cronjob to run them? Leif didn’t tell me, so if there aren’t any restocks, oh well, too damn bad, life goes on.

All that’s left to be done is redeploy the other sites (easy enough to do). But that happens when Leif gets back, because he only uploaded the eC code.

Yea. I still need to set up an ftpd/pop3/imap/smtp server. But not now, I’m tired and cranky and not in the mood for fucking around.