Reinstalling your dedicated server (reference)
This page shows you how to reimage your dedicated server in advanced ways – if you want the simple way see here instead where you only need to pick from a few menu options.
If you want to be able to control partitioning, processor architecture and IP configuration in a bit more detail, read on.
You can pick all the default options by simply typing:
The script will then confirm some installation details:
Initialising... <hostname>.dh.bytemark.co.uk ---------------------------- Distribution : squeeze Imager : imager3.bytemark.co.uk Net IPV4 gateway: 22.214.171.124 Net IPv4 : 126.96.36.199/24 RAID : RaidHP RAID level : raid10 Root password : 8dbSEnx6DeTW Continue? (y/N) >
The IP address and other details should be automatically filled in for you, so you only generally need to tell it the hostname. Also the system will default to installing Debian/squeeze – I’ll go through some other options in a second.
If you confirm by typing ‘y’ the system will have its drives wiped and the new OS installed within about 5 minutes. The process looks like this:
Configuring RAID... Building filesystems: /dev/sda1... /dev/sda2... Mounting filesystems... Requesting image... Imaging system... Executing post-install script... Installing boot loader... Installation of squeeze complete in 150 seconds. Uploading hardware manifest for tracking purposes: audit 188.8.131.52/24
At this point you can type “reboot” and the system will start with the new image.
Choosing Your Distribution
In the previous example we didn’t specify a distribution to install, so the imager defaulted to the latest release of Debian GNU/Linux (lenny at the time of writing).
If you wish to specify a different distribution you should make use of the
—dist parameter. This specifies the distribution you would like
to be installed. You can call InstallMeADedicatedServer
—distributions on its own to see the list of distributions available -
at the time of writing this is:
Available distributions for architecture amd64: [*] squeeze Debian 6.0 (64 bit) centos-6.3 CentOS 6.3 (64 bit) precise Ubuntu 12.04 (64 bit) wheezy Debian 7.0 (64 bit) symbiosis Symbiosis 6.0 (64 bit) centos-5.8 CentOS 5.8 (64 bit)
So far, this usage doesn’t differ from what you can do using the simple method . However there are various extra features that you can invoke:
Almost all of our dedicated servers are 64-bit capable systems, and the installer will default to a 64-bit installation.
If you want to force a 32-bit installation you can add:
Normally, the installer will take the network boot system’s current IP address as its permanent one for installations. If this is not correct (usually only if you have your own VLAN) you can specify the network parameters precisely:
--ip x.x.x.x --gateway x.x.x.x --netmask x.x.x.x --broadcast x.x.x.x
Normally the installer will make one small partition for /boot, and use the rest of the disc for the root (/). There are two ways of overriding this. We would recommend using the:
option which tells the installer to allocate the normal /boot partition, then only a 15GB root partition, and to use the rest of the disc for LVM. When the system has installed, you can them use the normal LVM tools to carve up the rest of the disc dynamically.
On Premium hosts it is also possible to allocate a second logical volume that is then used for LVM. This is only really useful where the total disc capacity is greater than 2TB, beyond the limit for traditional partition tables.
If you want the root partition to be a size other than 15GB (20GB, for example), you can specify that with the option:
Alternatively, you can do the partitioning and filesystem creation completely in advance. Once you have created your partitions, create a directory called
/tmp/target and mount your root and boot filesystems onto it. You then need to tell the installer which partitions you are using, and where you would like the boot-loader to be installed, for example:
--mounted --root-disc=/dev/sda --boot-dev=/dev/sda1 --root-dev=/dev/sda2
This tells it to skip the disc setup entirely, and trust you. Obviously if you’ve made any mistakes in partitioning, it may not complete.
While there are some other options listed, they are really only for debugging purposes, so only use at the risk of a badly set-up system, or if instructed by Bytemark support.