Atomicorp

Screwed my server up pretty bad trying to go from RH9 to CentOS 3.7 and then from Plesk 7.0.4 to 7.1.7. Had some problems with bind that I've resolved, but httpd still will not start.

I have access to the Plesk control panel and email is still functioning (really the most important aspect of our server).

When I "/etc/init.d/httpd start" it just says [FAILED] and /var/log/messages just says "httpd startup failed". There are no applicable entries in /var/log/httpd/error_log. "rpm -q httpd" gives httpd-2.0.46-56.ent.centos.2

In the process of trying to fix this, I've done some fairly stupid things like trying to reinstall httpd using the rpm (both the one I have installed and ART's version) using the --nodeps option.

I'm just at a loss because the error message is so vague. Any help would be greatly appreciated.

Oh yeah, you're in bad shape then. Never use --nodeps unless you've got a really good reason. Odds are you missed some key dependencies when you overruled the check, and now you'll have to go manually resolve them. Im afraid I can't give you much help here beyond that, most likely reinstalling will be less work than fixing it manually.

Thank you for the quick response!

"Reinstalling" is kind of what I had figured. Can I completely uninstall Plesk + dependencies and then install everything from scratch safely? I only have one domain on this server.

I guess if it doesn't leave all the data where it is, or for some reason can't detect it after uninstallation, a psadump/restore would work, correct?

Yeah, Im in mid-site tuning on the forums at the moment.

It would be faster to just reinstall the box than to try and remove PSA to fix this. Odds are you missed something core, which you could probably fix in a few hours. Compare that against a reinstall in about 10 minutes (I have a script, check out AOOI on the project page). I'd just back up your data and rekick the box.

Beautiful. Thanks again Scott.

I couldn't get AOOI to work. I didn't try very hard, but it didn't work right out of the box. I ended up using EV1's Auto-Restore feature.

AOOI installed itself and changed LILO, then I watched the boot process through EV1's "remote console" connection. LILO started the "atomic" kernel but then threw a single strange y looking character and froze. I let it sit for an hour to see what it would do, but then decided to just reboot it. Upon a remote reboot, my normal kernel loaded.

Not sure if I needed to mess with some of the kickstart config files or what else happened, but I do appreciate the help and effort. All is well now on my clean RHE4 Plesk 8.0.1 box!

Ah so EV1 uses LILO? I was not aware of that. I don't suppose you have any output you could send me? What the installer should do is boot (so you'd see all the kernel messages) then pull the remote ks.cfg. If its breaking at the boot part, then there is something wrong with the CentOS kernel itself on that hardware.

There really wasn't any output. It said, "Loading atomic" or something close to that, indicating it was loading the kernel. Then a series of periods, a new line, this character (I think) : Ÿ and then froze.

I'll send you anything you need regarding my hardware, etc, but that was the only output that I received.

As far as I know, EV1 uses LILO by default but installs both LILO and GRUB rpm packages. I probably could have tried the grub script, but EV1's support Control Panel crashed yesterday for a good 6+ hours which meant that I couldn't reboot. Very aggravating, and by then I had realized that they had a tool that does what your script does, using the same method.

Ah OK, Thats a problem with lilo loading the kernel. So it didnt even get to booting in that case. Lilo is actually easier to script for anyway, so its really a good thing that they do that.