They also offer more powerful Enterprise versions of their NAS devices, which run the same operating system but with much faster hardware. I’ve yet to test them in a production environment, but given my experience in the lab, I am sure they would be a competitive solution.

In this post I will show you how to set up a PXE boot server that will let you perform a network installation of Centos 6.3 using your Synology NAS.

What is PXE?

PXE (pronounced pixie) stands for Preboot eXecution Environment. It’s a technology that can be used to boot a computer into an operating system from it’s network card without needing anything to be installed on the computer’s local storage devices in advance. Most modern servers come with PXE support as standard.

It’s incredibly useful if you wish to automate the deployment of many servers without having to attend each one with an installation CD / DVD / USB stick. With a little work, you can also configure custom kickstart files to be served to each server, to save having to enter all the installation options manually.

How to set up your Synology NAS as a PXE boot server

Step 1 – Install DSM 4.2

Upgrade your Synology device to DSM 4.2 beta if you haven’t already. Follow the download links for your region, download the appropriate firmware that for your model of device, then upload it in the DSM admin panel – control panel – DSM update screen.

Step 2 – Set up the DHCP Service on your NAS

I would recommend you set up the DHCP server on your Synology first and test it works. If you are running this on your main LAN, you will need to disable the DHCP server on your router so they don’t conflict. You can download the DHCP server package in Package Center.

You will need to configure the relevant primary and secondary DNS, start and end IP addresses, netmask and gateway settings.

Once you are happy this is working, you can move on to configure the TFTP and PXE servers.

Step 3 – Set up the TFTP and PXE Services.

Tick the Enable TFTP service box. You also need to specify a folder somewhere on your NAS that can be used as the TFTP root folder.

Tick the Enable PXE service box. In the boot loader box type ‘pxelinux.0′. Fill out the remaining fields using the same settings you used for DHCP in step 2. This will override the DHCP service settings.

This will set up a DHCP service which sets DHCP 67 (boot filename) in it’s DHCP offers to be PXELINUX.0. If the server making the DHCP request is performing a PXE boot, it will attempt to retrieve and load this file via TFTP from the DHCP server IP address. It is possible to tell the server to use a different server for TFTP using DHCP option 66 – but this is not necessary in our case because the Synology NAS is performing both functions.

Step 4 – Upload the PXELINUX scripts and PXE menu to your tftp folder.

In order to get PXE boot working, we now need to upload the PXELINUX.0 and a few associated files from the SYSLINUX project to the TFTP share. I’m sure you could use other boot loaders, but I have never tried any, so I’m going to stick to what I know!

According to the Centos wiki, the minimum required files to perform a PXE network installation of Centos 6.3 are:

• pxelinux.0
• menu.c32
• memdisk
• mboot.c32
• chain.c32
• pxelinux.cfg/default
• path/to/your_kernel_of_choice
• path/to/your_init_ramdisk_of_choice

You could download these yourself and edit pxelinux.cfg/default as necessary, but this is out of the scope of this blog, so to speed things up I have created a Github repository with all the files necessary for a Centos 6.3 install.

Simply download this repository as a ZIP file and copy the files inside your tftp folder.

This perfoms a network install using a kickstart I’ve created which will set up Centos 6.3 with a few KVM packages for use as a hypervisor. NB: The default password is changeme1122

Step 5 – Attempt to PXE boot a server.

All you need now is a server. Ensure the server is connected to the LAN with your Synology NAS on it, then power on the server and instruct it to perform a network boot. It should make a DHCP request to the NAS, and then perform a PXE boot using the files that we copied to the TFTP server.

If you want to load a different operating system, you need to copy across the relevant kernels / initial ramdisks for the distribution of your choice and then edit the PXE menu in pxelinux.cfg/default. You may also wish to either remove the kickstart parameter, or refer to a different kickstart of your own creation.

Permission denied (publickey,gssapi-keyex,gssapi-with-mic,password).

And you may see an error in the audit log (/var/log/audit/audit.log) on the server..

type=AVC msg=audit(1358012203.073:43414): avc: denied { read } for pid=5945 comm=”sshd” name=”authorized_keys” dev=dm-1 ino=25583 scontext=system_u:system_r:sshd_t:s0-s0:c0.c1023 tcontext=unconfined_u:object_r:admin_home_t:s0 tclass=file

The way to fix this is to run…

restorecon -R -v /root/.ssh

… substituting /root/ if necessary for the relevant home dir.

How to setup SSH public key authentication with SELinux enabled

The full steps to setup an authorized keys file from scratch would therefore be:

1) Create the .ssh folder

mkdir -p /root/.ssh
chmod 755 /root/.ssh/

2) Set up the authorized_keys file (remember to paste in the relevant key in vim)

vim /root/.ssh/authorized_keys
chmod 600 /root/.ssh/authorized_keys

3) Fix the SELinux file contexts

 restorecon -R -v /root/.ssh

If you would like to create a bootable USB drive from OSX, you will need:

• Unetbootin
• Syslinux
• Disk Utility (pre-installed on OSX)

Step 1 – Format the disk in Disk Utility, with the correct MBR

Assuming you have already inserted your USB drive into your Mac…

a) Open Disk Utilty

b) Select the USB device

c) Click Partition

d) Select 1 partition in the partition layout

e) Select “Master Boot Record” in the options

f) Select MS-DOS (FAT) in the format type.

g) Click Apply, then Partition

h) Close Disk Utility

This will wipe the USB disk and set it up with the correct boot record.

Step 2  – Install the MBR binary from the SysLinux project

Open up a terminal and then

a) Use the command line diskutil to find the device name for your USB drive.

diskutil list

a) Umount the USB drive with the command line. NB: Be sure to swap the device reference (in my case it is /dev/SOMEdisk2) with the correct one for your usb key that you identified in the previous step – this will change for each machine.

diskutil unmountDisk /dev/SOMEdisk2

b) Mark the partition active, then unmount it again

sudo fdisk -e /dev/SOMEdisk2
print
f 1
write
print
exit

diskutil unmountDisk /dev/SOMEdisk2

c) Download Syslinux

mkdir -p ~/Documents/BootableUSB
cd ~/Documents/BootableUSB
curl -O http://www.kernel.org/pub/linux/utils/boot/syslinux/syslinux-5.00.zip
unzip syslinux-5.00.zip -d syslinux-5.00
cd syslinux-5.00/mbr

d) Install the MBR – NB: Update the device name (/dev/SOMEdisk2) to the one you identified in the first step!!!

sudo dd conv=notrunc bs=440 count=1 if=mbr.bin of=/dev/SOMEdisk2

Step 3 – Use UnetBootin to install your OS install files

a) Download and install UnetBootin if you haven’t already from http://unetbootin.sourceforge.net

b) Load the application, choose your preferred distribution, and then click OK.

c) When it’s finished, eject the usb key and use it!

References

Thanks to a tip I found on http://perpetual-notion.blogspot.co.uk/2011/08/unetbootin-on-mac-os-x.html

df -lP | awk '{total+=$3} END {printf "%d G\n", total/2^20 + 0.5}'

What this does is…

• df -lP … shows a disk report of all local disks, in posix format (e.g. one line per volume)
• | awk ‘{total+=$3} END {printf “%d G\n”, total/2^20 + 0.5}’ … this takes the output of the df command, pipes it to awk which then sums the 3rd columns into a variable called total, and when it’s finished it prints out this number converted to Gigabytes.  To get to Gigabytes, we divided by 2^20 (1024*1024), and we also add 0.5 so that it is effectively rounded to the nearest whole number.

This is particularly helpful if you have a lot of volumes on a system.

1) Click the details button in the virtual machines virt-manager window.

2) Navigate to the Display VNC option in the left hand menu.

3) You should now see a Keymap drop down box. Select the correct Keymap that matches your keyboard, in my case it was en-gb.

4) Click Apply.

Good luck, it worked for me!

Then we came across, Ansible. It’s developed primarily by the guy who wrote Cobbler (our previous favourite deployment tool), and it’s been built from scratch to be much simpler to use.

Although it only took me just a few minutes to get it up and running, there were a couple of dependencies required to install it on OSX, so I thought I would document the steps here in case it helps someone else in the future:

Step 1 – Install Dependencies

Check that you have Xcode installed with , try typing gcc -v from the command line. If it says command not found, then ensure you have downloaded Xcode from App Store, then once it’s installed go to Xcode > Preferences > Downloads  and install the Command Line Tools.

Install pre-requisite Python modules on your main OSX machine that will be running ansible.

sudo easy_install jinja2
sudo easy_install PyYAML
sudo easy_install paramiko

If you get an error saying configure: error: no acceptable C compiler found in $PATH then you should double check you have Xcode and it’s command line tools as per the instructions above.

And on the managed nodes, assuming it’s a Centos/SL/RedHat node:

yum install python-simplejson

Step 2 – Checkout latest version of Ansible

Download the latest version from Github. Assuming you are installing into ~/github

cd ~/github
git clone git://github.com/ansible/ansible.git
cd ./ansible
source ./hacking/env-setup

Set up your bash profile to load the environment variables when you load a terminal

# Setup Ansible
cd ~/github/ansible
source ./hacking/env-setup
cd ~
export ANSIBLE_HOSTS=~/ansible_hosts
# End Ansible Setup

If you wanted to install it properly, rather than run the latest code from checkout you could do so with sudo make install. In this case you wouldn’t need to run the hacking/env-setup script to modify the environment variables, since ansible would be located within the ordinary search paths.

Step 3 – Test it works

Add some hosts to your ansible_hosts file, and then ping them.

echo "192.168.0.62" >> ~/ansible_hosts
ansible all -m ping

Assuming the servers you need to add they key to are called:

server1.example.com
server2.example.com
server3.example.com
server4.example.com
etc…

And your colleagues key is in a file called key.pub

To add to a list of consecutive servers, use:

for i in {1..10}; do ssh-copy-id -i key.pub root@server$i.example.com; done

If you wanted to list specific servers, you could use:

for i in {1,4,7,10}; do ssh-copy-id -i key.pub root@server$i.example.com; done

You can then check it worked with the following: (replace colleagues-key-comment with the comment that identifies their public key):

for i in {1..14}; do echo server$i; ssh root@server$i.example.com cat /root/.ssh/authorized_keys | grep colleagues-key-comment; done

I’ve also measured the heights to be the following:

• Ethernet port – 13mm tall
• Usb 2.0 ports – 15mm
• Audio jack – 10mm
• RCA video – 13mm
• HDMI – 6 mm
• SD Card – 4.5 mm (underneath)
• Micro USB power – 2.5 mm

sudo /opt/vc/bin/vcfiled
sudo /opt/vc/bin/ilplayer /home/pi/big_buck_bunny_1080p_h264_ac3.mkv

Big Buck Bunny is a short animated film by the Blender Foundation - (c) copyright 2008, Blender Foundation / www.bigbuckbunny.org.

The production boards should be going on sale in the next few months for £16 / £25 each depending on the model you choose. For those of you who are looking forward to ordering your own, I have taken some photos of it against my iPhone so you can get a better idea of the size for yourselves.

The first thing that strikes you about the Raspberry Pi board is how small it is. I had seen pictures of it online, but I’m sure it looked bigger! It’s great to see how far computing has evolved since I got my first BBC Model B computer approximately 25 years ago.

You can see the full size Raspberry Pi board diagram here.

Although these boards were designed for schools, I think they will transform the world of computing in myriad other ways. A few years ago the OLTP foundation set out to build a $100 laptop. 2012 marks the beginning of the $20 desktop.

What is the Raspberry Pi?

The Raspberry Pi board is effectively an entire computer on a credit card size board. Like a cut down Mac Mini, all you need is a keyboard and a monitor and you can run a full Linux desktop operating system like Fedora or Debian. It’ll even play Quake 3 and full 1080p films!

Re-using these photos

I’m releasing all these photos under the Creative commons attribution license:

Raspberry Pi Beta Board #8 Photos by Paul Maunders is licensed under a Creative Commons Attribution 2.0 UK: England & Wales License.