In any case setting up a LUKS + LVM root on Debian is a breeze, in fact the installer does everything for you. And the Ubuntu Desktop CD? Well, not so much, unfortunately. You could use the alternate or server CD, but this was not possible for me, I only had the desktop CD on a USB Stick. In any case, since Ubuntu also uses apt, its really just more or less a matter of installing what’s missing, dpkg will do the rest!

It goes a little something like this…

Pre-flight

Before we begin…

1. Boot the Ubuntu Karmic CD into the “LiveCD” mode this is the one called, “Try Ubuntu without any change to your computer”.
2. You will need an internet connection to download the latest versions of the missing programs, make sure you have connectivity from the LiveCD.
3. Make sure you know what the drive name is of the drive you are installing to in this howto it will be /dev/sda
4. You will need a separate unencrypted boot partition, it is entirely possible to have this on a different device, such as a USB Stick, however here we assume that the boot partition is the first partition of the internal harddrive, i.e. /dev/sda1
5. Open up a Terminal (Man up sissy boy – Applications > Accessories > Terminal)
6. (Optional) fill your drive with random data dd if=/dev/urandom of=/dev/sda. This prevents people from seeing how much data, but not what data has been written to the drive. This will take forever, especially if you have a large drive, like a 1TB.

Getting the tools

Now we need to get the tools, this is relatively simple to do. Since we will need root privileges for almost the entire process, i recommend dropping to root immediately. To do this, simply type:

sudo su -l


We will assume that you are root for all commands below!!!

Now we need to update the packages and install cryptsetup and lvm2. To do this first type in:

1. apt-get update
2. apt-get install cryptsetup lvm2

Setting up the encrypted volume

Partitioning the disk

First we need to create a partition table, to do this we fire up fdisk. In this tutorial the disk we will partitioning is /dev/sda.

1. type: fdisk /dev/sda
2. Type in p to print the current partition table and use d to delete any existing partitions: Command (m for help): p
3. Create the first, primary, bootable, 250MB partition.
   1. type: n
   2. then: p
   3. then: 1
   4. then hit enter
   5. then type: +250M
   6. then: a
   7. then: 1
   8. and type: p, you should get something like:
      
Device Boot Start End Blocks Id System
/dev/sda1 * 1 33 265041 83 Linux

4. Now create a single partition, taking up the rest of the space:
   1. type: n
   2. then: p
   3. then: 2
   4. then hit enter
   5. then hit enter again
   6. and type: p, you should get something like:
      
Device Boot Start End Blocks Id System
/dev/sda1 * 1 33 265041 83 Linux
/dev/sda2 34 2610 20699752+ 83 Linux

5. now press w

Create the encrypted Volume

Next step is to setup the encrypted volume. We “luksFormat” the partition we want to be our encrypted volume, in this example this is /dev/sda2. To encrypt the volume, you should select a nice password. If you are not good at making your own pseudo-random passwords, download pwgen AND DON’T LOSE THIS PASSWORD!!!

1. apt-get install pwgen
2. pwgen 20

Now we wipe the partition. Enter your super-duper password at the prompt for the LUKS volume.

cryptsetup luksFormat /dev/sda2


Mount the encrypted volume

Now all that is left is mounting the encrypted volume. The first parameter of the command is the physical partition or device that the encrypted volume resides on, the second parameter is the alias for the decrypted volume in /dev/mapper. In this example I called my decrypted volume sda2_crypt (this happens to also be the Debian naming scheme) however you can chose whatever you want. Enter the LUKS volume password at the prompt.

cryptsetup luksOpen /dev/sda2 sda2_crypt


Now that we should have the encrypted volume up and running, we can check by typing cryptsetup status /dev/mapper/sda2_crypt. You should see a similar output.


/dev/mapper//dev/mapper/sda2_crypt is active:
cipher: aes-cbc-essiv:sha256
keysize: 128 bits
device: /dev/sda2
offset: 1032 sectors
size: 41398473 sectors
mode: read/write


Setting up LVM

Perfect, now we can make this a physical volume for LVM. If you have never used LVM before, its about time! I’m not gonna go into great detail about how it works, there are plenty of resources online explaining the differences between physical volumes, volume groups and logical volumes. Very briefly though, each logical volume resides on a single volume group which reside on one or more physical volumes. So a volume group is a pool of physical devices which we can use to create logical volumes. In our case we will have one physical volume (/dev/mapper/sda2_crypt), one volume group (i called mine crypto in this example) and several logical volumes.

Create the Physical Volume

Lets go! Create the physical volume with:

pvcreate /dev/mapper/sda2_crypt


Create the Volume Group

Create a volume group with:

vgcreate crypto /dev/mapper/sda2_crypt


Create the Logical Volumes

And now create your logical volumes. You need at least a root and a swap, but you can create more. In this example i also created a /home and a /tmp partition (albeit pretty small). To create a logical volume use lvcreate -n<volume_name> -L<volume_size> crypto, where <volume_name> is the name of the volume (i.e. root), where <volume_size> is size of the volume (i.e 10G or 512M) and where crypto is, of course, the example name of our volume group from the previous step.

root@ubuntu:~# lvcreate -nroot -L10G crypto
Logical volume "root" created
root@ubuntu:~# lvcreate -nswap -L1G crypto
Logical volume "swap" created
root@ubuntu:~# lvcreate -nhome -L4G crypto
Logical volume "home" created
root@ubuntu:~# lvcreate -ntmp -L4G crypto
Logical volume "tmp" created


Now we’re ready to install!

Installation

The installation should be completely normal, the only changes you have to make are at the Partitioning screen.

Partitioning

Select: Specify Partitions manually at this screen

Formatting

At the next screen remember to set mountpoints for your partitions and for them to be formatted. I chose to use ext2 for my boot partition (/dev/sda1), mostly for legacy reasons. Make sure the swap logical volume is formatted swap, otherwise format the other partitions (i.e. root, tmp and home) whatever you want, i chose ext4. Again, MAKE SURE YOU ARE SETTING MOUNTPOINTS and DO NOT format /dev/sda2!!!

Remember to select mountpoints, filesystems and format the partitions at this screen.

Finishing the Installation

Now click forward and let the installer finish. Remember to NOT RESTART at the very end, select “Continue Testing”.

Select "Continue Testing" here!

Getting the missing bits and pieces

Now, fire up a Terminal again, drop to root (sudo su -l). We will need to mount the new system, create an /etc/crypttab entry and install some missing packages.

Mount the new system

To mount the partitions do the following

1. create a mount point: mkdir -p /mnt/newroot
2. mount the root: mount /dev/crypto/root /mnt/newroot
3. mount proc: mount /proc /mnt/newroot/proc -t proc
4. mount your boot partition: mount /dev/sda1 /mnt/newroot/boot
5. mount any other partitions (i.e. /tmp, /home, /var…): mount /dev/crypto/<vol_name> /mnt/newroot/<mount_name>

chroot into the new environment

Now we are going to chroot into this environment, to do this type: chroot /mnt/newroot if you got no errors, and the command pwd returns /, you should be good.

Create an /etc/crypttab

Now we need to create the /etc/crypttab entry and install the missing packages:

To create an entry in /etc/crypttab simply (replacing any discrepancies in naming) do:

echo "sda2_crypt /dev/sda2 none luks" >> /etc/crypttab


Your /etc/crypttab should now look something like this, when you type cat /etc/crypttab.

# <target name> <source device> <key file> <options>
sda2_crypt /dev/sda2 none luks


Install the missing packages

Now we only need to install the missing packages, the rest will be done by dpkg’s scripts.

1. apt-get update
2. apt-get install lvm2 cryptsetup

Unmount the partitions

Thats it! Now exit chroot and unmount all the volumes

1. to exit chroot simply type: exit
2. to unmount all the mountpoints within /mnt/newroot, type comma-seperated in curly-brackets what you want to unmount: umount /mnt/newroot/{boot,home,tmp,proc}
3. you should now be able to unmount /mnt/newroot, if not check if you unmounted all partitions in step 2: umount /mnt/newroot

Reboot… pray

Now type in reboot and cross your fingers! Hopefully you are welcomed by a screen like this:

If all went well, Ubuntu's splash screen should ask you for the password

Troubleshooting

Ideally it all went well, and you don’t have any complaints, however, I will create a troubleshooting section as comments rain in, so please do comment.

Thanks for reading, can’t wait to see some comments.

Just add all the links you want to download to a text file and save it with .txt extension. Now run this script and input the .txt file as the parameter, i.e:

./txttohtml links.txt

Now you should find an html file in the same folder with the text file, the filename is the, but with an .html extension. Anyway, here’s the script:


#!/usr/bin/ruby
# Dmitry Nedospasov
# Text to HTML
# This script converts a .txt file to an html file with each
# line of the file as a link
#
# I wanted this so that i could use rapidshare with DTA
# Script needs one argument, the .txt file
require 'ftools'
def parseargs(arguments, stdin)
# puts arguments
txtFile = arguments[0]
if txtFile == nil
puts 'Please enter a filename'
elsif txtFile =~ /.txt$/
puts 'Checking the following file: ' + txtFile
convertFromTxtToHtml(txtFile)
else
puts 'Please enter a file that ends with .txt'
end
end
def convertFromTxtToHtml(txtFile)
htmlFile = txtFile.sub(/.txt$/, " .html" )
puts 'Writing results to: ' + htmlFile
myHtmlFile = File.new(htmlFile, " w+" )
myTxtFile = File.open(txtFile, " r" )
if File.zero?(myTxtFile)
puts 'the .txt file is empty'
elsif File.exist?(myTxtFile)
myHtmlFile.puts(" <html>" )
counter = 1
myTxtFile.readlines.each { |curLine|
newLine = '<a href=" '
newLine += curLine
newLine += '" >'
newLine += " #{counter}"
newLine += ' - '
newLine += curLine
newLine += '</a>'
myHtmlFile.puts(newLine)
myHtmlFile.puts(" <br />" )
counter += 1
}
myHtmlFile.puts(" </html>" )
else
puts 'File not found'
end
end
parseargs(ARGV, STDIN)


Now you can fire up firefox and open the html file and start dta, however the “cool” way to create the html files and launch firefox from terminal is, in Mac OS:


./txttohtml.rb *.txt && open -a Firefox *.html


And in linux:

./txttohtml.rb *.txt && firefox *.html

Have fun and here is the file for download

Enter 3 parameters:

1. input directory
2. search string
3. replacement string

Although the hardware is detected by default, somehting isn’t supported and causes the broadcasts from avahi and Mac OS to be dropped. You will need XCode to build the sources, or look for pre-compiled ones on the net. More on my hackintosh later.

Haven’t written in a really long time. Anyway, We recently did a project that generates Mosaics in C++. You can check out the results here. And a full write up should be up pretty soon with some detailed explanations and links to paper’s that will get you started on a similar project.

http://pastebin.ca/966414

My configuration

Just in case you were wondering what configuration i used, and continue to use for that matter Here it is:

• Debian Etch
• Self-compiled netatalk (for Leopard compatibility) here’s a good howto.
• 330GB Volume on Raid5 formatted ext3

So, if you have the first two you’re in business, but even if you don’t use netatalk. All you have to do is mount everything as described in “Mounting”.

Setting up the Harddrive

Now first things first, boot from your cd. To do that just press and hold the c-key immediately after powering on your computer. You should eventually be greeted by a language selection etc. After all of this you will be welcomed by the installer. We don’t need it at all. At the top you will see “Utilities”, where you will find among others Disk Utility, Terminal and Restore from a Time Machine Backup.You will probably want to start off, especially if its a new harddrive by checking out Disk Utility and creating any volumes you want to install your system. After you have them configured your pretty much ready to go.

Network and Jumbo Frames

So now you should connect make sure you are connect to your network. You can open up disk utility to check connectivity. If you need anything special, like setting up specific network settings, these can be done in terminal. I for example like to have jumbo frames, since all my gigabit devices support it. Open up Terminal from the utilities menu, and type.ifconfig en0 mtu 9000and you should be set.

Mounting the Volumes

Let me reitterate that I have not yet tested anything other than netatalk (afp), but this should for any network file sharing protocol. All you have to do is mount your Time Machine volume under /Volumes. The syntax for mounting an afp share ismount -t afp afp://username:password@hostname/ShareName
/Volumes/ShareMountImportant here is to remember that the ShareName is a case-sensitive string. So make sure you type \ before a space, and to check for proper capitalization. Otherwise you get some mount errors, which occur because a volume wasn’t found with that name, since the actual name doesn’t match the one you typed. ShareMount is any string you want, it doesn’t matter what you call it. Same goes for spaces here. If you mount your afp volume just /Volumes (i.e. without ShareMount) you might see multiple TMBackups, as is the case with the images bellow. Still works, just shocking to see 2 of the same thing. Thats basically it, you’re almost done…

Restoring…

Before you close Terminal, make sure the share actually mounted properly, you can simply ls /Volumes and see if your mount is there. If you didn’t mount it at /Volumes/ShareMount your /Volumes folder might be litered with the contents of your afp volume, this shouldn’t cause any problems, just doesn’t look neat at first site. If everything seems to be there, then select “Restore from a Time Machine Backup” from Utilities. Click through the menu and you will come to a screen called “Select a Backup Source”, with your shares appearing properly and being recognized. Don’t ask me why there are two, probably because MacOS in general uses a lot of symlinks… Anyway, important thing is that it works. Select the first volume, and after selecting the you will come to a screen called “Select a Backup”, where you get to chose which backup you wish to use. Afterwards, the system will calculate if there is enough space and will afterwards let you restore. This will now take some time and it went up for me after about 25mins, so don’t hesitate to let it sit there for 2-3 hours. Afterwards, you will just need to reboot, and you’ll be back where you left off.

Hey, I’ve seen this before!

One thing some will notice, is that things like Mail and Itunes, will need to reconfigure themselves, i.e. reimport all your emails. Having said that, I’m still blown away by how well this went over, even if this is “unsupported”. Oh and for those of you wondering how to backup to an unsupported volume in the first place, well its simple. All you have to do is:defaults write com.apple.systempreferences TMShowUnsupportedNetworkVolumes 1Hope this helps some people, feel free to leave a comment.

NFS and OS X

Anyway, some of us have had Leopard for a couple of weeks, and i recommend it for anyone who has a fairly new mac, by which i mean that the only consideration should be if you have enough performance to get everything out of leopard. Leopard has a lot of new “features”, one of which I’ll be explaing today.

NFS isn’t something new, and its not even something new for Mac OS. The client to mount NFS is included on almost every *nix, and doesn’t require any additional installation (usually). It also has better performance than Samba (thats what windows uses to share files), and has a fairly extensive list of configuration options. Of course, nfs isn’t perfect for everything, but if your computers run *nix (aka BSD/Unix/Mac/Linux), then NFS might be the easiest file sharing protocol to configure. After all, its basically already there, so you save yourself the install.

I’m going to assume you have NFS configured on the server, this is the computer hosting the files. If not, well its actually pretty simple, and NFS has probably the best howtos out there. You can find it here. Anyway back to the client side, running OS X. So in 10.4, mounting NFS was kind of, well a pain. Mac OS connects to an “insecure” port by default. In other words it doesn’t send a request to the server for a port. This means if you run a firewall, you can’t connect via finder, because Mac OS will attempt to connect to a port that will be blocked, since its not in the firewall rules. Not only that, you have to configure your your shares to be “insecure” otherwise /var/log/messages will get something along the lines of:

russo@marx:~$ cat /var/log/messages | grep nfsd
Nov 8 16:56:15 marx kernel: nfsd: request from insecure port (192.168.0.4:49204)!

Directory Utility

There was a solution in 10.4, and it was fairly complicated and involved setting everything up with NetInfo Manager. This procedure was quite tedious, as you can see in this howto. But we’re not in 10.4 anymore (thank god!) And guess what… NetInfo is gone! And replaced by something, much more user friendly. The new replacement is called, Directory Utility. And can be found in your Utilities folder, imagine that! Fire it up, lets take a look!

So at first site, there probably won’t be much there, especially on a home network. But, this doesn’t matter we’re not configuring Directory Services, we’re here for the NFS mounts. Just click “Show Advanced Settings” and you’ll see “Mounts” appear up top. Click that. Next you should see something like this

Click the lock in the bottom left corner, to allow you to add/edit mounts. Enter the username and password of an Administrator on the computer (Chuck Norris administers mine). Now you should be able to click the + to add a mount. By the way, with Directory Utility you don’t have to reboot to see changes. You just click that apply in the corner. How sweet is that?

The syntax here is a bit different than mount in CLI, however the syntax still contains that which made many new users make a mistake. The syntax is

nfs://<hostname or ip>/<full path to directory>

Remember, /full/path/to/dir is the full path on the server!

Also enter, the directory where you want this share to be mounted. Its probably best to keep everything under either /Volumes or /Network. This way all users can easily access to the shares. You can always add aliases to your home directory, or wherever you want these mounts to actually appear. Now if you, just want to automount “insecure” mounts, than you’re basically done at this point. However we want to add “secure” mounts. This means we need to specify the “Privelaged” flag to the mount. This will cause it to use a port designated by the server. Click the little triangle next to “Advanced Mount Parameters” And you’ll see something like this:

all we have to do is add, “-P”. This makes the mount use a privelaged port. Now click “Verify”. At this point, its good to mention that if Directory Utility only checks if the server that you entered is running nfs and not if the mount is actually functional. Now click “Apply”

Nothing Happend…

Well it did, but you might not be able to see it. At the very least the mount point has been created, and you can check this by opening finder and going to either /Volumes or /Network (EDIT: or wherever you mounted you shaere). To get there you’ll have to type command (apple) + shift + G, and then type in where you want to go. Okay, so now you should see it there, BUT you probably won’t be able to write to even if you didn’t mount it read-only. Why is that, you ask? Well you don’t have permissions, silly! And even if you your user does have permissions to write, even if your user exists on both machines, he probably won’t have the same gid/uid. How can you check? Just use the id command:

lenin:~ russo$ id
uid=501(russo) gid=20(staff)

Just as a side note Mac OS begins with 500, whereas most linux distros begin with 1000. So now you can either change the gid/uid of one of the machines so that they do match, or run “chmod 777″ the directory on the server… that should be it. You should be able to write now!

EDIT: There are also ways to map IPs, hostnames and authenticated users to users on the machine. Maping ips and hostnames is fairly easy, you simply add anonuid=youruid, anongid=yourgid to the options in the /etc/exports file. (Thanks Paolo)

How do i get the share in my Finder Sidebar?

Simple! Just click the folder and drag it while holding apple + option. You can place it under “Places” in this manner. Even if you’re not an admin user.

EDIT: This wasn’t clear enough for some. You cannot create aliases for the actual mounts, however you can create a folder enclosing the mounts and create an alias for that. I.e /Volumes/Hostname and put all the mounts in /Volumes/Hostname. I hope its a little bit easier to understand now…

Here are the links to the .pls files, which you can open with winamp or iTunes among anothers, to tune in to the stream.

Ibiza Global Radio.

Ibiza Sonica.

Blogged with Floc