First, you have to load the following kernel modules on the hardware node and/or set them to automatically start during the boot up process:

modprobe ppp_mppe
modprobe ppp_deflate
modprobe zlib_deflate
modprobe ppp_async
modprobe ppp_generic
modprobe slhc
modprobe crc_ccitt

After doing that you should see something similiar to this when running lsmod

[root@me ~]# lsmod | grep ppp
ppp_mppe 39816 0
ppp_deflate 39168 0
zlib_deflate 52760 1 ppp_deflate
ppp_async 45056 0
ppp_generic 63632 4 ppp_mppe,ppp_deflate,ppp_async
slhc 39680 1 ppp_generic
crc_ccitt 35200 1 ppp_async

Second, you have to tweak the VPS’s conf file:

vzctl stop [VEnumber]
vzctl set [VEnumber] --features ppp:on --save
vzctl start [VEnumber]
vzctl set [VEnumber] --devices c:108:0:rw --save
vzctl exec [VEnumber] mknod /dev/ppp c 108 0
vzctl exec [VEnumber] chmod 600 /dev/ppp

Once that’s done, you can follow the normal instructions.

After that, you might need to uncomment the require-mppe-128 or mppe required,stateless line in /etc/ppp/options.pptp (depending on the VPN server). You may also need to manually create a static route to the network, with route add -net 10.1.1.0 netmask 255.255.255.0 ppp0.

You can check /var/log/messages for details about negotiation errors.

I tried simply removing Sendmail, but it turns out that Cronie has a hardcoded dependency for it. At first it seemed pretty ridiculous to make such a core program dependent on something optional and obsolete, but after a bit of digging I found that the newer MTAs provide a sendmail interface, so all you really need to do is install Postfix or Exim and run alternatives –config mta. After that you can remove Sendmail and yum won’t complain.

It turned out to be an easy fix once I figured out the problem, though; I just had to change the ownership of /var/www/cgi-bin and it’s contents so that the apache user could execute cgi_wrapper.

Here’s what was turning up in the error logs:

Apache vhost error_log

[warn] [client xx.xx.xx.xx] (104)Connection reset by peer: mod_fcgid: error reading data from FastCGI server
[error] [client xx.xx.xx.xx] Premature end of script headers: index.php

/var/log/httpd/error_log

suexec policy violation: see suexec log for more details

/var/log/httpd/suexec_log

uid: (10001/username) gid: (505/505) cmd: cgi_wrapper

/var/log/httpd/suexec.log

target uid/gid (10001/505) mismatch with directory (0/0) or program (0/0)
uid: (10001/username) gid: (505/505) cmd: cgi_wrapper

If anything you try to install says that it’ll remove packages (like glibc-common), don’t do it. You could remove a lot of core packages and end up having to reinstall Endian from scratch. Not that I would know anything about that…

First run smart config –set rpm-check-signatures=False to avoid errors about invalid keys. Then you can do
smart install http://mirror.centos.org/centos/4/os/i386/CentOS/RPMS/nano-1.2.4-1.i386.rpm
smart install http://mirror.centos.org/centos/4/os/i386/CentOS/RPMS/slocate-2.7-13.el4.8.i386.rpm
etc

You can still get stuck in dependency hell, though. Smart might be able to auto download dependencies if you add a channel, but I don’t have the time/motivation to look into that right now.

tar -czf [output-file.tar.gz] [input-directory]/
mysqldump -u [username] -p [database name] > [filename]

It’s possible to configure this yourself, though. There’s a helpful post at the OpenDNS forums that describes how to configure ddclient on an Endian to automatically update an OpenDNS account whenever your dynamic IP address changes.

Here are a few additional notes that may help:

• You don’t need to query an external website for your IP address, since the Endian box already knows it. Set ddclient.conf with “use=if, if=eth0″ (or whichever NIC is your red zone).
• It won’t work if you set “ssl=no”, so you have to leave it set to yes and install the two Perl SSL modules listed in the post.
• Endian doesn’t have an SMTP server installed, so ddclient can’t send you status e-mails. Comment out the “mail” and “mail-failure” lines to avoid errors.
• Make sure you don’t have any spaces in the OpenDNS network label or ddclient conf file.

To test if it’s working, you’ll need to get a new dynamic IP.

• Turn off your cable or dsl modem.
• Spoof the MAC address on the Endian.
  • On your PC, open a command prompt and type “ipconfig /all”. Copy the value of the Physical Address line. It will look something like 00-4D-B7-D3-5F-03.
  • Go to the Network Configuration wizard on the Endian.
  • In step 4, enter your PC’s MAC into the “Spoof MAC address with” field. Change the dashes to colons so that it looks like 00:4D:B7:D3:5F:03.
• Shut down the Endian.
• Turn the modem back on. Wait 15 seconds.
• Turn the Endian back on.

Once you’ve verified it’s working, you can remove the spoof setting by repeating the steps.

Instead of displaying the errors, you can configure PHP to log them and then use the logs for debugging. For a Linux environment, you’ll need to create a file named vhost.conf inside the domain’s conf directory, /var/www/vhosts/domain.name/conf.

<Directory /var/www/vhosts/domain.name>
  php_value error_log /var/www/vhosts/domain.name/statistics/logs/php-errors.log
  php_flag display_errors off
  php_value error_reporting 6143
  php_flag log_errors on
</Directory>

Then create the file and assign it the right permissions

  cd /var/www/vhosts/domain.name/statistics/logs/
  touch php-errors.log
  chown apache php-errors.log

And the last thing is to tell Apache to reread the updated conf file

  /usr/local/psa/admin/sbin/websrvmng -a