Getting started with TRK
1. TRK for Linux newbies
1.1 What is TRK? What 's a live distribution?
1.2 What is different between accessing your PC from Windows and accessing from TRK?
1.3 Getting around with common linux commands (cd, cp, mv, rm, more, grep, mount)
1.4 Reading information about your PC (dmesg, /proc/partitions)
2. TRK own commands and utils
2.1 Virusscan
2.2 Winpass and regedit
2.3 Clonexp
2.4 Mountallfs
2.5 Updatetrk
2.6 Trk2usb
2.7 Trk2iso
2.8 Ntfsundeleteall
2.9 Fileserver
2.10 Bridge
2.11 Setip
2.12 Setproxy
3. Procedures
3.1 Rescueing files of dying harddiscs (mounting network => cp, ddrescue)
3.2 Recovering deleted files or files from formatted drives (ntfsundeleteall, photorec)
3.3 Recovering lost partitions (testdisk, gpart, fdisk)
3.4 Bootsector repair
3.5 Manually cloning a Windows installation
3.6 Hardware testing
3.7 Virus scanning
3.8 Manual PC cleaning
4. Boot time options and triggers
4.1 Boot menu options
4.2 Triggers
4.2.1 The TRK options server: make your lan TRK aware
4.2.2 Scripts on the computer's local harddisks
4.2.3 Script on the TRK medium
4.2.4 Antivirus on CD
5. Upgrade, update and change of bootmedia procedures
5.1 How to install/upgrade your USB media to run the latest version of TRK
5.2 Setting up your PXE boot environment
Addendum TRK 3.3: what 's new since 3.2?

Getting started with TRK

0. Foreword and conventions of this documentation:

-Consulting help
Trinity Rescue Kit 3.2 does not contain any manpages or info files (3.3 does). For help on common Linux commands you are mostly bound to what you find on the Internet or in books. Most tools however have their own short help message baked in. Most of the time this help gets displayed when a syntax error occurs or by typing "-h" or "--help" behind the command.
The only documentation included in TRK are these helppages as an offline version of the trinityhome.org 's documentation. You might be looking at them right now. You can call on this documentation as one big page from TRK by typing 'trkhelp' at the command prompt. This will start the builtin Links browser in graphical mode opening all of the documentation at once. To be able to switch between this help and your commandprompt, you must be in text mode.  Use the key "q" to quit links and type 'trkhelp -t' to run in text mode. Once in textmode, use alt+F2 to go the second console of TRK. In total there are 6 consoles, each switchable with their respective alt+ function key.
If you are not online, you can consult the local documentation which is on the TRK medium by entering 'trkhelp -l'.
You can combine these two parameters: 'trkhelp -l -t'  gets you the local helpfiles in textmode.

-literal commands that you can execute in TRK or Linux are put between 'single quotes'. Omit the quotes when using the real commandline. Exceptions on these quotes will be mentionned (when single quotes really appear in the command).

-"double quotes" are used to emphasise words, unless they are used inside commands.

-<trkmedium> stands for the rootfolder of the medium on which TRK runs. TRK can be run from CD, usb stick/disk, fixed harddisk or from network over PXE. These specific bootmethods will be explained later in this document.

-this documentation is intended for people who at least have some experience with computer troubleshooting or know how to install their own Windows. If you have absolutely no idea of this, I recommend you call someone who knows more.

1. TRK for Linux newbies

This section gives a quick introduction about the concepts of Linux, a live distribution and how you should see and control your computer from the viewpoint of TRK

1.1 What is TRK? What 's a live distribution?

This is a definition for people that have absolutely no idea of what an operating system means. If you don 't understand what I just said, you need to read the following text. The people who do know what it all means, might take the following definition as incorrect, but that 's just so I can explain in simple language what I mean.

“Trinity Rescue Kit" or TRK is a collection of programs that can start a computer in an alternate way if it failed to start normally.

TRK is a so called "live distribution" of Linux. Linux is in fact the "brain" and "the senses" that drives your computer, all the programs are the limbs that allow you to control it. All this put together make up a so called operating system. Because of confusion in terminology and because of the so many flavors, we talk about Linux being the operating system and what makes it complete with programs is called the distribution. And as for distributions, a lot of flavors exist (hundreds, maybe even thousands). One of these flavors is Trinity Rescue Kit. TRK is a "live" distribution because it can boot from removable media, perform hardware detection on-the-fly and automate as many configuration tasks as possible.

TRK operates completely from RAM (= compare this with the short term memory of your brain). This means also it doesn ‘t touch nor change anything stored on your computer until you tell it to do so. Another implication this will have is that by starting your computer with Trinity Rescue Kit you have the guarantee that no viruses that might be on your local computer can become active. In the case of your computer running Microsoft Windows in normal circumstances, it just is impossibe by design that a windows virus can run on TRK because TRK is Linux.

You have the possibility of using 4 different antivirus scanners with TRK. One of them, the free open source Clamav is integrated in the distribution, the other 3, F-prot, Grisoft AVG and Bitdefender get downloaded from the Internet when launched.
 
More on what TRK is can be read on the frontpage

1.2 What is different between accessing your PC from Windows and accessing from TRK?

Instead of running programs on your computer using Windows, you are now starting your computer with something completely different. Trinity Rescue Kit is not designed to give you the same environment you get normally, but to provide you the means and the utilities to perform rescue and repair operations that might not be (anymore) possible on your computer in normal Windows operation mode.

Because this is Linux, you will not see your drives in the same way you do under Windows (or DOS), but they will appear as logically assigned devices. So instead of the C:-drive, you will get /dev/hda1 in which ‘hda’ is your first available harddisk ('h' in hda is for ide drives, 's' is for scsi and removable drives), hda1 is the first (primary) partition on which a filesystem may reside. The filesystem used in general for Windows is NTFS. This stands for New Technology FileSystem, but in the mean time the "New" in technology is already more than 12 years old. Nevertheless, it has gotten some improvements over the years and it is, I must admit, a good filesystem. The other filesystem supported by Windows is FAT, which come in the flavors FAT12 (for floppies), FAT16 (for small disks up to 2Gb) and FAT32 (for bigger disks).

What 's also different from Windows is that these filesystems are not accessible by default in Linux, you have to so called "mount" them. Where in Windows you will get a drive C: with your files on, in Linux you have to call the command 'mount' and mount the filesystem against a subdirectory. An example of this: let 's say your drive C: is /dev/hda1 (/dev contains the collection of device references on your system). You have a directory /mnt0. In this case you type 'mount /dev/hda1 /mnt0'. When you invoke the command 'mount' afterwards without any parameters, you will see that /dev/hda1 is mounted on /mnt0. If you cd to that directory and type 'ls' (=equivalant of 'dir' in Windows), you will get a directory listing of what 's available on that C: drive.

Now with this all explained, you should see the picture: TRK runs on your computer but treats it as a doctor inspecting a dead body: everything is there, but the person is not. You can now perform surgery on the body and try to revive it.

1.3 Getting around with common linux commands (cd, cp, mv, rm, more, grep, mount)

This is a small tour on commands you will find useful when working with TRK (and Linux in general). Let 's take as a convention that commands you have to type are put between 'single quotes'. At the commandline you omit these quotes (unless I say not to).

What I 'm going to teach here is basically how you work with files, like copying, moving, editing, etc...

First, let's start with changing directories.

People that have worked with the commandline in Windows or Dos will recognize many things. The big difference with this is that directories in Linux are separated by forward slashes instead of backslashes in Windows. Another big difference is that Linux folders and files are case sensitive: capitals have to be typed as capitals, otherwise the file or folder will not be found.

-cd

The command to change a directory (or folder how it 's called in Windows) is 'cd'
f. i. you want to change to the directory /tmp you type 'cd /tmp'. If the folder contains spaces, there are two ways you can get into them: or you put the foldername between double quotes or you use so called escape characters to represent the spaces.

Let 's say we want to cd to the folder 'Documents and Settings' you can type 'cd "Documents and Settings"' or 'cd Documents\ and\ Settings' where the backslash (\) in Linux is used to 'escape' characters, i.e. you treat the characters litterally instead of as a control character. Enough of that.

Another way of easily changing directories is by typing the first letters of it and then pressing the tab-key. F.i. type 'cd Doc' and press tab. The command will complete as 'cd Documents\ and\ Settings'. If more files or folders match, the tab command will show you the possible options left. Very nice feature, saves you lots of time and painful fingers.

-cp

File copying, the second thing you 'll probably need.
Once you 're in the right directory, you need to be able to copy files to other locations.
Here 's how to do it, together with the most important switches.
Take a file called file1, located in /tmp/ and you want to copy it to /home: simple command 'cp /tmp/file1 /home' or when you first cd-ed to /tmp: 'cp file1 /home/'. Source-target.
Suppose you want to copy multiple files, you can use a wildcard with '*'. If /tmp contains file1 and file2 and copy them both at the same time: 'cp /tmp/file* /home/'
If you want to copy files and folders recursively, together with all the attributes, use the switch '-a'
To see what gets copied while it 's busy, add the '-v' parameter too (verbose).
Say /tmp contains 'file1', 'file2', a subfolder called 'testfolder' which also contains 'file3', then perform 'cp -av /tmp/* /home' This will copy the complete contents of /tmp, including subfolders to /home.
If cp asks you to overwrite already existing files, you can force that by adding '-f' (force) to your command:
'cp -avf /tmp/* /home'

-mv

Moving files. This is the same principle as copying files but easier, e.g.
'mv /tmp/* /home' moves all the contents, including subdirs to /home
Adding the parameters -v and f moves them verbosely without prompting to overwrite existing files.

-rm

Remove files.
Remove 1 file, example: 'rm /tmp/file1'
Remove files recursively, without prompting: 'rm -rf /tmp/*'

-more

Viewing files and output of commands. This is a util you can use any time there 's too much output coming to your screen or you need to look into a file.
E.g.: 'more /tmp/file1' shows you the contents of file1, but gives it a page at a time. To go to the next page, press space. The arrows and enter key scroll down line by line.
You can use this command also in combination with other commands to halt their output so you can read what it says.
Example: 'dmesg | more' : 'dmesg' gives you the output of your kernel startup procedure and recent system messages, but it 's maybe about 300 lines of output. So in this case we so called "pipe" the output of 'dmesg' to 'more' using the "|" sign. In this way I have also explained you with an example the use of  "command piping"

-command piping

Continueing on this subject, let 's see what other uses command piping can do for us.
It can be used to filter out a certain line with a specific keyword.
Let 's say you want to know whether there 's a file called Document.doc' somewhere in a subdirectory, but you don 't know which. Then use this command from within the base directory you want to search in: 'find ./ | grep -i document.doc' (the -i parameter upper- or lowercase characters)
You can also pipe the output of a command to a file instead of the screen.  F.i. to put the complete filelisting of a directory tree to a file, do like this 'find ./ > /tmp/filelist.txt' 

-editing files

Here 's quickly how to use vi, the most common text editor on Linux. Beware: this does not edit Word documents or any other document format that is in binary format.
Open a file or create a new file: 'vi /tmp/file1'
Move your cursor around with your arrows to the line you want to edit.
To insert text, type 'i', this will put you in insert mode. To remove text, use 'x', (go out of insert mode first with escape). To remove or cut a complete line, use 'dd'. You can paste this line elsewhere with 'p'.
This is basically editing in vi. To save a document, go out of insert mode and type ':wq' (colon write quit). To exit without saving: ':q!' (colon quit exclamation mark).
If you don 't like vi, you can use pico which is a bit simpler to use, but less common on the different Linux systems.

-mounting filesystems

When working with Linux and more specifically here with Trinity Rescue Kit, it is imperative that you understand the way you "talk" to filesystems.
Whereas Windows just assigns a driveletter to any local filesystem it knows and finds (which is only NTFS and FAT), Linux does it all by invoking "mount" of a filesystem against a directory where you mount it.
Trinity Rescue Kit has a utility called "mountallfs" that searches for every filesystem on the local computer's disk drives and mounts it in a directory that has the same basename as the device where the filesystem resides. More on that later in this documentation. In other, normal Linux distributions, local filesystems are detected or created on install.

Let 's talk now how to perform manual mounting.
Mounting can be performed with any filesystem, regardless of it being local or on the network. 

* Mounting a local filesystem can be done like this:

To know what device contains te filesystem you want to mount, you can look at a file called "/proc/partitions" This will tell you the partition lay-out of your disks, which will most likely contain filesystems. A common "/proc/partitions" file may look like this:

/dev/hda
/dev/hda1
/dev/hda2

/dev/hda claims in fact the whole disk. Under Windows it is impossible to create a filesystem in there, under Linux it is possible but improbable and not recommended.
Most likely you will find a filesystem on /dev/hda1 and /dev/hda2, which will be you C: and D:-drive under Windows in general.
Mounting this is quite easy, in general you don 't have to give any paramters with it, Linux will detect the type of filesystem.
'mount /dev/hda1 /mnt0'

Trinity Rescue Kit by default has two directories for manual mounting of filesystems. You can create as many as you like, in as many subdirs as you like. That 's all I 'm going to explain about local filesystem mounting. I recommand you use "mountallfs" anyway, and even better 'mountallfs -g'. More on that later.
 
* Mounting network filesystems.

This is a very interesting bit, because with Trinity Rescue Kit you will want to evacuate your files to another computer. In TRK (and most other Linux distributions) it is possible to talk to Windows filesharing technology. For those who want to know the name of this technology, it 's called SMB (=Server Message Block). TRK can act as a client as well as a server. In this case we 're talking about TRK as a client.

Let 's say you have a running windows machine and you 've configured it to share "myshare". If you have not configured a share, you can connect to the c$ hidden share, but then you need to deactivate "Use simple simple filesharing" in the folder options of your Windows explorer. But let us take the "myshare" share.

For the ease of use, it 's a good thing to create a user on your Windows machine called "root", give him a password and make it an administrator. But that 's not really necessary, you can also use the local "administrator" account, this will just require you to add a parameter to the mount command.

Let 's take the case of the "myshare" share, your Windows pc has 10.0.0.5 as ip-address (always faster to just point to the ip-address instead of the name) and you 've created a local user called root. Then here 's the command: 'mount //10.0.0.5/myshare /mnt0'

You will get prompted for a password in if no output is given, your share should appear under /mnt0. Make sure "myshare" has enough permissions for the user "root"

If you don't want to create a user, you don 't want to create a share and you did disable simple filesharing (or your windows machine is member of a domain), you can go ahead like this:
'mount -o username=administrator //10.0.0.5/c$ /mnt0' , which will prompt for the password an give you the credentials of "administrator". You can already pass the password (e.g. blahblah) in the commandline if nobody is looking over your shoulder: 'mount -o username=administrator,password=blahblah //10.0.0.5/c$ /mnt0'

Below is a screenshot of a Windows PC that has opened the disk of a remote TRK booted machine.

1.4 Reading information about your PC (dmesg, /proc/partitions)

Let 's summarize a little what 's already been said and look at reading info on your computer.

-/proc

Standard Linux always has a filesystem called "proc", which is a virtual filesystem in which files reside that have to do with your hardware and running processes. It 's a wonderful invention. We already talked about /proc/partitions, which is a file containing all local disk partitions.

Another interesting file is /proc/cpuinfo, which gives you information about your CPU. Trinity Rescue Kit only supports a single CPU, since processing power is not a priority and it can add only complication to certain incompatible hardware.

/proc/meminfo gives you information about the memory usage. Don 't let yourself get misled by the Memfree line, which will always look very low. Actually, Linux always reserves most part of the memory so it can make use of it in a fast way. What you need to look at is the Active and Inactive.

Another useful file to read info on your cdrom drive is /proc/sys/dev/cdrom/info.

Those are about the important files in /proc you need to know about now.

 -dmesg

The command "dmesg" gives you your kernel messages. Any hardware detected will give you a message somewhere in the output of this command. If you want to know the type of network card that has been detected, perform 'dmesg|more' and look for any mentions of eth0, eth1,...

What type of harddisk controller you have: dmesg. Just use it when you find yourself stuck on hardware questions. Also disk failures will be visible with this command. Network errors, link down, etc, one command.

Another way of looking at this information is through /var/log/messages (more /var/log/messages), which on normal Linux distributions contain output logs of previous boots too.

To know what device your newly inserted USB stick has,  plug it in, let it settle for a few seconds and then run dmesg again. Or just run 'dmesg|tail' to see only the last added lines.
Here 's an excerpt of what you might read from dmesg. It tells you something about your network card:
eepro100.c:v1.09j-t 9/29/99 Donald Becker http://www.scyld.com/network/eepro100.html
eepro100.c: $Revision: 1.36 $ 2000/11/17 Modified by Andrey V. Savochkin <saw@saw.sw.com.sg> and others
ACPI: PCI Interrupt 0000:05:08.0[A] -> GSI 20 (level, low) -> IRQ 16
eth0: OEM i82557/i82558 10/100 Ethernet, 00:08:02:C6:4E:9D, IRQ 16.
  Board assembly 262285-001, Physical connectors present: RJ45
  Primary interface chip i82555 PHY #1.
  General self-test: passed.
  Serial sub-system self-test: passed.
  Internal registers self-test: passed.
  ROM checksum self-test: passed (0x04f4518b).

 -lspci

This gives you any information on what 's on your PCI bus. This doesn 't only mean what 's in your PCI slots, but everything on the bus, so also onboard ethernet and usb controllers.

 -smartctl

The smartmontools are part of Trinity Rescue Kit and not so common on normal Linux systems, yet they are a valuable addition to any system. What it does is read the s.m.a.r.t. information of disk drives so you can know when errors start to occur.

Just use it like this: 'smartctl -a /dev/sda' where sda is your first scsi or sata drive. Make sure smart is enabled in the computer's bios.

 -acpi and acpitool

Two tools to read the battery and thermal information of your computer. Type acpi --help to get more info on possible options.

acpitool can give you much more information like fan speeds and cpu. Also certain laptop types are supported for their special features like brigthness on Asus laptops etc...

 -df and du

Two standard utils provided in Linux. df shows you the usage of your mounted filesystems, du shows you the usage of a specific folder. Use it like 'df -h' and 'du -h' where "-h" stands for "human readable", making the output rounded to mega- and gigabytes.
This is in short how you can get to know your computer a little and how to jumpstart using Linux and Trinity Rescue Kit.

-lshw

Recently recommended and added, but looks very promising, lshw gives you a complete list of all your hardware in your computer. Best to pipe this to a file, because the list can get long.

2. TRK own commands and utils

Let 's look at the added value of Trinity Rescue Kit, with its own specific commands and utils.

2.1 Virusscan

Virusscan is a script that actually wraps 4 different virusscanners into one. Only one of them is actually included on the TRK cd, the others are downloaded from their website upon usage.
When running virusscan, it is highly recommended that your computer has a decent internet connection so you can get the latest virus signatures.

Scan engines

-Clamav: Included in TRK is the GPL licensed Clamav, which is a decent scanner with very up-to-date support against the biggest virus threats and worms. I personally use it on the Trinityhome 's mailserver and so far NO virus has passed since it was installed in 2004 (except when it needed to get updated since a long time). I make sure the latest clamav engine version is included in TRK.
You can manually update Clamav if you put the latest daily.cvd file on the TRK medium in <trkmedium>/trk3/clamdb
The pros of Clamav are:
* very quick on new virus outbreaks
* included in TRK
* GPL licensed, so free for everyone
The cons:
* slow and very CPU and memory intensive
* detects the least viri of the 4 scanners in TRK. Because it 's in fact a mailserver scanner, it will focus more on worms than on filth that comes from malicious websites and such.
* cannot disinfect inside files on its own. What I did in this case is quarantain the infected files into a tar.gz archive in <scandestination>/TRK-INFECTED/. Should a file be accidentally deleted, you can recover it afterwards and rescan it with another antivirustool

-F-prot
This antivirus tool and all the others are not included in TRK but get downloaded from the Internet as soon as you call upon them. They disappear after a reboot of TRK.  If you want them to be available after a reboot, you have to run updatetrk. This will be explained later in this documentation.
The pros of F-prot:
*lightweight, not a big download
*pretty fast, low cpu usage
*good disinfection method
The cons:
*does not detect everything
*their website sometimes fails and download of f-prot is aborted

-Grisoft AVG
My personal favorite. It 's the fastest of the bunch and detects the most viri. It even detects certain malware, adware and spyware, but it doesn 't disinfect them. For that you probably need their spyware util, which unfortunately only exists for Windows. But some get detected and this is written to a logfile. Virusscan is written in such a way that it offers to delete these files after scanning.
Pros of AVG:
*detects most of all 4 antivirus tools
*also detects certain malware
*fast
Cons:
*big download, since the virusscanners are unpacked into the ramdisk, it requires a computer with sufficient ram (at least 256Mb, more is recommended)
*cpu intensive, yet the computer stays very well responsive
*version and website changes often which can result in file being unavailable for download. I 've captured some of that in a way that you first get the right url from the trinityhome site.

-BitDefender Scanner
The latest addition to TRK. It has a good average between filesize, cpu/memory load and virusdetection. It can detect many different types of malware. From what I 've experienced so far, it may detect other viri and malware than the other 3. I recommend to sweep with this after another one has already run.
Pros of BitDefender Scanner:
*detects quite some viri
*pretty fast
*detects alternate malware
Cons:
*sometimes doesn 't detect very common viri
*slow update process

Ok, so far for the differences between the 4 antivirus engines

Usage

Below is the help output when you type 'virusscan -h'. It 's self explanatory, with an example

Usage: 'virusscan -a {clam,avg,fprot,bde} -c -g -n -d {DESTINATION}' where
-a: What antivirus you want to use. Takes 'clam' for ClamAv, 'fprot' for F-Prot, 'avg'
    for Grisoft AVG and 'bde' for BitDefender.
    Note that ClamAv is actually meant to plug into mailservers and block infections.
    Therefore it can only quarantain infected files not disinfect them.
    BitDefender proposes disinfection as a first action and deletion as a second
    If the parameter '-a'is omitted, ClamAv will be the default
-c: Use Common extensions:  (only with ClamAv, speeds up scan a little)
-g: Get only, just download the AV and updates, no scanning. For use with updatetrk
-d: Destination folder to scan. If no destination is given, virusscan will scan all local
    filesystems it can find. You can specify multiple destinations when you separate them
    with commas (wihthout spaces behind)
-n: No update. Don 't check for new AV signatures
-h: Displays this helpmessage
Example: virusscan -a avg -d /mnt0,/hda1/WINDOWS,/hda1/Program\ Files
    This downloads and updates AVG if not available and scans the locations /mnt0,
    /hda1/WINDOWS and /hda1/Program\ Files

2.2 Winpass and regedit

winpass

By default it will reset the builtin administrator account of a locally installed Windows, but you can specify other accounts as well at the commandline. In fact, you can add any parameter from chntpw which will be parsed to the commandline. So winpass –l will list you all usernames found in the SAM. Should you have troubles that metacharacters are present in the username (like the Ø or something), you can still use the HEX reference to the username listed next to it. Be sure to prepend that with a '0x'. More info on that can be found in the manual of chntpw. Watch out with disabling syskey! Most of the time you don 't need to disable syskey. It removes strong password encryption, so all local passwords become invalid, since the hashes don 't correspond to the encryption algorythm anymore. This has been known to cause problems with Windows XP's product activation. Read the warning well!!!

regedit

this is actually the same script as winpass, but called in this way it copies all registry hives and opens them (sam, security, system and software). Only drawback is that it cannot know which user registry to open, since they are located in different directories. Because chntpw is not scriptable, I cannot read registry hives from a shell script to determine the location of userhives.

Once running, refer to the documentation of chntpw/regedit on how to use it.

After you did your work in the registry, just exit and the changed hives will be written back.

2.3 Clonexp

Clonexp clones an ntfs partition to another computer over the network. Both PCs must be running TRK3.2. The one who is receiving the data must run with sshd enabled (option 8 in the startup menu)

Basically what it does is run ntfsclone in save mode to stdout at one computer and restore mode from stdin on the other, piping over ssh between eachother.

When running clonexp, make sure your source partition is smaller or equal to what can fit on the destination. Ntfsclone (and subsequently clonexp) cannot (yet) dynamically resize a partition, so it 's a good thing to make the destination partition equal or bigger in size than the source. Do this easily with qtparted, supplied on TRK. If your maximum possible destination size is smaller than your source, try making your source smaller with qtparted. At least, if the data on it is less than the destination and the volume is not too fragmented.

Also make sure neither of the local nor remote filesystems are mounted, or the script will exit.

Clonexp can run in two modes: without any arguments, it runs in interactive mode and prompts you for source device, destination host and destination device. It will each time test whether these things exist.

You can also run it in a single commandline, in this way (drop the <>):
clonexp <sourcedevice> <destinationhost>:<destinationdevice>
e.g. 'clonexp /dev/hda1 192.168.0.7:/dev/hda1'

It will prompt you once for the password you have entered on the remote TRK machine and then start copying the data. There is also the option '--bkupbr', which will first backup your partition bootrecord and later save that as a file on your newly created partition. Although it may never really serve, I give this opportunity should you ever be unable to boot from the partition and you want to play around a bit with your old bootrecord. This might as well completely obliterate your partition, so use with caution (backups 'n stuff ready...)
I was once able to make a cloned unbootable partition bootable this way by afterwards running from the Windows XP recovery console and do a chkdsk.
It 's recommended you use clonexp only between computers which have the same disk layout, or even better, are exactly the same hardware.
If clonexp didn 't work for you, there 's another, manual procedure, explained later in "Procedures"

2.4 Mountallfs

This script is about the most handy addon to TRK. It 's a script that scans all local storage devices and mounts their filesystems to a standard mountpoint, nl. the name of its device name.

-'mountallfs' without any options will mount all partitions to a mountpoint named after their device name. So /dev/hda1 will be mounted to /hda1
Run without any options, it will mount ntfs drives with the linux kernel builtin driver, giving you full read access but only very limited write access (overwriting files of the same filesize with ntfscp)
-'mountallfs -f' mounts any filesystem but ntfs filesystems will be mounted with the ntfs-fuse driver, which has good performance and limited write capabilities (like create maximum 10 files per directory). For full and performant write access, better use the next option.
-'mountallfs -g': finally, after years of waiting, a full fledged, fast read/write ntfs driver sees the daylight. Although the driver is still in beta, there have so far been no reports of any data corruption. Write performance is fantastic and there are no big limitations to this kind of filesystem access, except it cannot create compressed files (it can read them, but copies go uncompressed) and encrypted files (which is normal). Also filesystem rights are ignored and full filesystem access is provided.
-'mountallfs -c' will also mount any filesystem, but in particular this will mount ntfs with the captive-ntfs driver. It emulates the required subsystems of the Microsoft Windows kernel by reusing one of the original ntoskrnl.exe and ntfs.sys driver files from Windows XP service pack 2.
Because of licensing restrictions from Microsoft, I cannot provide these two files from SP2 along in the TRK iso file, since you have to be a legally licensed Windows XP user in order to use them. Read more about it on the site of the original captive developer  and my howto on adding these files to TRK after you downloaded it.
-'mountallfs -l': this activates Logical Volume Management modules, scans the local computer for any logical volumes and mounts them. This completely automates the process of mounting filesystems inside logical volumes. Very useful for sysadmins on Fedora boxes (which apparently is the default there).
There are some other options you can add to mountallfs at the end, like –q or –quiet, which make it a bit more silent and not show you the result of what has been mounted.
--ignore-memfree: normally, when using captive support for mountallfs, the script will check if your machine has enough RAM to run, since this option uses quite a lot of resources. If you have less then 64Mb of ram, it will refuse to run. If you really want to try anyway, you can add this parameter (which is otherwise harmless, it will probably result in some errors and killed processes from the kernel)

2.5 Updatetrk

Trinity Rescue Kit 3.2 is able to update itself through the script 'updatetrk'. This requires decent Internet connectivity, direct, over a router or via a proxy server (in that case, run 'setproxy' first)

The script is very simple to handle:
Suppose you 've booted from a PC with a Windows XP (2K or NT is also good), you can run updatetrk without any option.
Because TRK needs at least 250Mb to rebuild itself, it will look for a 300Mb pagefile.sys on your local harddisk and create a temporary filesystem on it. Since a pagefile doesn 't contain any valuable data when the PC is not running from Windows, it 's ideal to create a loopback device on it and create a workbench.
If you don 't have a local pagefile.sys (which can be either on NTFS or FAT32), you will have to specify a location yourself where TRK can be rebuild 'updatetrk -d /mountpoint/dir/'). Make sure this is on a filesystem somewhere where at least 250mb is available (run mountallfs -g for best write access on ntfs drives). If your pagefile is too small, increase the size from Windows. Since most recent PC have at least 256MB RAM, your pagefile will have at least that size, so you should be ok.

Updatetrk takes a few arguments. Below is the help output that explains them all.

Usage

updatetrk -b {BUILDTARGET} -i {TARGETISOFILE} (or) -u {TARGETUSBDEVICE}
-b {BUILDTARGET}: Build location where intermediate files can be stored. Should be at least 250mb
   Make sure no other files reside in that location, as they will be included as well
   If no device is given, a pagefile.sys will be searched on the local harddrives of the PC
   This pagefile will be mounted as a loopback device and have an ext2 filesystem built on
   This has no consequences for your Windows system afterwards, it will just be reused
-i {TARGETISOFILE}: Target dir for isofile to create. There should be at least 140Mb free on the
   target. The filename will be automatically given.
   If option is omitted, the file will be created in the same location as the buildtarget
   If the buildtarget is omitted, it will be created in the temp dir of the location of the
   pagefile. In general this will be c:\temp
-u {TARGETUSBDEVICE}: Target USB filesystem to update to. This does a simple copy to the location
    you specified. In general this is the partition you booted TRK from
Example 1: 'updatetrk -b /hda1/temp/trkbuilding/ -i /hda1/Docs/' will use /hda1/temp/trkbuilding/
   as construction site and create /hda1/Docs/trinity-rescue-kit-3.2-275u.iso
Example 2: 'updatetrk -u /dev/sda4' will search a local pagefile as construction site and copy
   back the files to your USB device (or fixed harddisc)
If you specify no options at all, TRK will always be created on a local pagefile as an ISO in
c:\temp. If no pagefile is found, the script will exit.
If you specified a specific buildtarget, the files will remain there afterwards at your convenience
Remark: option '-i' and '-u' are not useable together

What will 'updatetrk' do?

-Well, first it will first ask you if you want Captive NTFS support and if yes, it will run 'captive-install-acquire', a utility included from the Captive NTFS project. This tool will search your local harddrive for the best available ntfs.sys and ntoskrnl.exe. If it doesn 't find them locally, it will download them from Microsoft. Follow its instructions, it 's extremely simple (hit a few times enter and a 'd' when done)


-Then, it downloads the latest virus signatures for ClamAV

-Next, it downloads F-Prot + updates

-As a third antivirus, it fetches Grisoft AVG + updates

-As last antivirusscanner, it downloads BitDefender and starts the installation procedure. This is BitDefender 's own install procedure, just follow instructions on the screen, type "accept" for the license agreement and leave everything to the default propositions.

-Finally, it copies the complete contents of /bin, /sbin /etc and /lib to your new TRK. This means that anything you modified or added in these directories will be included in your new TRK. This gives you an easy way to make small modifications/additions to TRK.

This script can be run time after time to keep your antivirus signatures up-to-date.

When no destination parameter was given to TRK and you 're running TRK from CD, it will move the newly created isofile to the drive where it found your pagefile.sys in directory 'temp'. Most likely this will be C:\temp. Burn that to CD with your favorite burnprogram in Windows or use 'cdrecord', included in TRK.
When your TRK booted from USB/harddisk, it will update itself to USB/harddisk, unless specified differently

2.6 Trk2usb

This util puts your  Trinity Rescue Kit to a USB stick/disk or even to a fixed harddisk. This is one of the two methods you can use to get TRK running from USB. This is also the recommended and easiest method of doing it. Only downside here is that you first have to burn the iso version of TRK to CD, boot from it and run it from there.
The alternate method of putting the tar.gz file to your USB will be explained later.
Below is the usage syntax

Usage

'trk2usb -d {DEVICE} (optional) -s {SIZE}  -n'
    -d {DEVICE}: specify destination device. Specify the complete disk, not a
    device partition. Your complete disk will be erased, zeroed out and formatted
    with one partition as a FAT 16 of maximum 1Gb being the fourth primary
    (e.g. specify -d /dev/sda, you will get /dev/sda4 as a (max.) 1Gb FAT16 on
    /dev/sda4). This is for maximum compatibility with lousy BIOSes
    -s {SIZE}: specify a specific size (in Mb) for your device if you want to add
    more partitions later. Use a minimum of 128 so TRK fits on it.
    -n: noformat. Use this if your device has already the good formatting and you
    don 't want to lose your data on it. This allows you to specify your destination
      partition in your commandline, e.g 'trk2usb -n -d /dev/sda1'

2.7 Trk2iso

On special demand, I 've created this small utility that creates an ISO file of your currently running TRK. It doesn 't update trk, it just creates an isofile in the current directory from where the command was launched. Here 's how to use it best:
'mountallfs -g' => suppose /dev/hda1 is your ntfs c:-drive
'mkdir /hda1/temp'
'cd /hda1/temp'
'trk2iso'
 
Once the isofile created, you can reburn TRK to CD.

2.8 Ntfsundeleteall

This script is a simple wrapper for the utility 'ntfsundelete'

What it does is find all deleted files on an ntfs volume and tries to recover them. It does this by looking for traces in the ntfs Master File Table and sees whether the file is still available on the disk (or partially overwritten). It 's a good utility that recovers your files with the original filename but without the relevant path in front. So to avoid double recovered filenames, I 've written the script so that it prepends the original inode (=internal number for the file on the filesystem) before the filename, so you always have a unique filename. It 's up to you to rename the files afterwards, but at least you have the original name in the recovered filename.

Here 's how the script works

Suppose you have the volume /dev/hda1 in ntfs from which you wish to recover files.
Make sure this volume is NOT mounted
Prepare a place to which you can store your recovered files. Recovered files will never be written to the original partition from which you recover them because the risk exists you will overwrite other files that still need recovery. Let 's say you have space on a network volume that you 've mounted on /mnt0 (for this procedure, see the section "Getting around with commands"), with a subdir called "recovered"
Here 's the syntax of ntfsundeleteall:
ntfsundeleteall Device Savedir [minimum percentage] --force

Now here 's how we do it in our example:
'ntfsundeleteall /dev/hda1 /mnt0/recovered'

This is very simple, this will recover any file it finds, even if only 1% of the file is recoverable. You 're probably not interested in that, because these files are most likely to be corrupt and unusable, so you add a parameter on the percentage you wish the file is recoverable. If we only want 100% recoverability, make the command like this:

'ntfsundeleteall /dev/hda1 /mnt0/recovered 100'

You will see the directory /mnt0/recovered filling up with recovered files, hopefully your file is in there.

You might have noticed the parameter "--force". This will only skip the check that your filesystem is ntfs, you can use this if you think your filesystem is so badly damaged that it won 't even be recognized anymore, but I doubt you will be even capable of recovering any files then.

Another recommendation I can give you is Photorec. Photorec is a third party utility that is able to recover files in a unique way: it is independent of the filesystem, it just does raw read of the disk (tries to use as much info as it can from whatever filesystem there was on it) and tries to recognize known document formats to recover whatever it can find. Although the name implies it might only serve to recover pictures, it can in fact recover any kind of important document, including word and excel, but also file archives such as zip and tar.gz.

The only thing you won 't get from this utility is the original filename, so you 'll have to sort out your recovered documents one by one.

2.9 Fileserver

In short: share your drives like a windows fileserver
The "fileserver" util actually runs a samba fileserver on your TRK.
It read-write mounts all your local drives by calling 'mountallfs -g' and then shares all your found partitions as their device name. So if /dev/hda1 was a ntfs formatted partition on host 10.0.0.1, it would be shared as \\10.0.0.1\hda1
Fileserver can be called in two modes: in secured or in guest mode.

-Secured mode is invoked with the command 'fileserver -s'
This will prompt your for adding a user and a password that will have access to your files remotely. Just follow instructions. Here 's how the output would be if you ran it

Starting a username/password secured Samba fileserver and sharing all local filesystems
Enter a username which will be created to have access to your local files:
New SMB password:
Retype new SMB password:
Added user testuser.
Mounting all your local filesystems using mountallfs -g
These are the IP-addresses your fileserver will listen to:
192.168.81.5
Starting SMB services:                                          [  OK  ]
Starting NMB services:                                          [  OK  ]

-Guest mode is invoked with the command 'fileserver -g'
Use with caution, it allows everyone with network access to the computer to go on the local harddiscs. Use only in a trusted environment with a firewall or not connected to the Internet.

2.10 Bridge

Suppose you have a PC of which of which you like to know what traffic it generates. Or suppose you have a router on which you like to know what traffic passes by. Well, "bridge" will create you a transparent connection between ethernet cards that passes all traffic and on which you can start sniffing what passes by. You could use an old fashion hub that replicates all traffic to all of its ports, but nowadays that is almost inexistent, almost any cheap hub is in fact a switch that only replicates traffic to the port it is needed. In that case, put a computer with TRK in between your connection.

What you need

is a computer with at least 2 network cards and the latest TRK to boot from.

-Once the network cards are connected on both sides (on on the PC, one on the network), run 'bridge up'. This will setup a bridge and traffic will transparently pass from one side to the other. One note here which is quite important: make sure only 1 network card has its own ip-address, let the other one without (if they both have one, do an 'ifconfig eth1 down' f.i.). I will explain this later.

-Now, start sniffing with tcpdump. Run it like this: 'tcpdump -i eth0' In this way, you see ALL traffic headers that pass by. If you need to filter something specific, use "grep" to get it out. F.i. you need to see all http traffic going to 10.0.1.20, run this: 'tcpdump -i eth0 | grep 10.0.1.20 | grep http'

About the single ip-address: I have noticed that if you connect two network cards to the same network and let them both get an ip-address through dhcp (which will subsequently be addresses in the same network range) and afterwards connect one to a pc to start sniffing, tcpdump will go berzerk and no output will be shown but your TRK will give the impression of hanging and will not respond to a ctrl+c signal very quickly. After about 10 seconds, it will stop. But you will have noticed that sniffing doesn 't work. Well, if one card doesn 't have an address, this problem doesn 't occur. So leave at least one nic down. You can even leave both down, you don 't need an ip-address to sniff the network, "bridge" will bring up both cards in promiscuous mode and make a bridge interface "br0".

To bring the bridge down, just type 'bridge down'

2.11 Setip

Use this whenever you want to enter a fixed ip address for your network cards

Usage

setip <interface> | -h
This command sets the ip information of your network card(s)
Without an argument, it assumes your network card is eth0
'setip -h' shows the help output
Here 's what  the command does:

'setip eth1
Enter the IP-address for eth1: 10.0.0.1
Enter the subnet mask for eth1: 255.255.255.0
Enter the default gateway, <enter> for none: 10.0.0.254
Enter the IP-address of the DNS server you 'd like to use, <enter> for none: 10.0.0.3

Setting ipaddress 10.0.0.1/255.255.255.0 on eth1
Setting default gateway to 10.0.0.254
Setting nameserver 10.0.0.3 in /etc/resolv.conf'

2.12 Setproxy

This script prompts you for proxy server settings, should you be unable to go directly onto the Internet. Use it without any arguments, it will prompt you for proxy ip-address or hostname, proxyport (usually 8080) and optionally username and password (invisible). If no username/password is required, just leave it blank and hit enter.

This script can also be called upon startup in the bootmenu.

Usage: 'setproxy'
 

If you want your proxyserver to always be included, I recommend you set up your LAN to be "TRK-aware", explained later in this documentation.

Setproxy sets the proxy settings for these applications:

-wget
-links
-ClamAv
-F-prot
-Grisoft AVG
-BitDefender

3. Procedures

Let 's talk a little on what actions you should take in time of need, how to best use TRK to perform troubleshooting, or how you even might work with tools other than TRK.

3.1 Rescueing files of dying harddiscs (mounting network => cp, ddrescue)

Cool it!

When you 've got a harddrive that 's starting to die on you (bad clusters, physical damage), it 's imperative that you "upset" the disk as little as possible. In most cases the cause of your dying disk is heat: hard drives can get very hot, especially when heavily used and poorly cooled. A good consideration is to make sure that your disks are placed freely floating so heat can get away easily. Also make sure that all of your fans in your computer blow in the same direction. A standard here is that fans in the front "inhale" air, while fans in the back of your computer "exhale". If you 're considering of placing a fan inside your computer directly on your disk, it 's best practice to let it blow away from the disk. This way you create a vacuum and no air anywhere, not even in the smallest wholes can sustain in a vacuum.

Another good practice is to attach your disks firmly to the computer, so the iron from the chassis can guide away heat from the disk. But do this only by attaching the sides of the harddisks to the chassis.

Now, in case it 's already too late and your disk is dying, it might be that it works fine the first minutes and starts to make funny noises afterwards: probably the bearings have worn out or the platters start to bend and the heads are grating the surface of the disk.

Thing here is to hurry up and get those files off that disk.

One suggestion I might give here might sound weird, but it worked for me already in many cases:
put the disk in the fridge for a while and then boot it up! Really, I 'm not kidding. But don 't freeze it, that will make it only worse. Put it in a normal refrigerator and let it cool down there. Also put a towel around it to avoid condensation.
Another possibility here is to connect your disk externally through a disk-to-USB cable that you can buy for 20$ or less at your local computershop. (or connect the disk to an IDE cable that hangs out of your PC like guts)
Now, should you have that, go and look for that winecooler pack you got for your birthday, you know, those plastic things you put around a bottle to keep it cool (or go and buy it at your local grocerie store). A gelpack to ease the mussles can also do the trick, but don 't get it out of the freezer where it 's minus 18° Celsius, that 's way too cold and might damage your disk even more. And always wrap a towel around your disk first.

Now, once this is in place, you will have a better timewindow to start recovering files.

What you can try, from good to bad to worst case scenario.

In the best case you 'r still able to mount the filesystem on the disk. Use mountallfs to search for all local filesystems.

Make sure you have a location where you can copy your recovered files to. This can be another local disk or a remote network filesystem like a Windows share. How you should reach these safe haven is explained earlier in this document (Getting around with commands).

So let 's assume a few things:

-mountallfs manages to mount your filesystems on your dying disk
-you mount a remote network share: 'mount -o username=administrator,password=mypassword //10.0.0.5/c$ /mnt0'
-'cd /hda1' where hda1 is your dying disk
-do an 'ls' to see that you can still see files there.
-go to the directory from which you like to recover documents. Don 't start to try and copy all files, your time here is limited and it 's very likely that your disk is damaged on places where it has read and written many times. In the case of a Windows system, this is the WINDOWS system directory. Stay out of it, there are normally no documents there.

More likely you will do something like this: 'cd /hda1/Documents\ and\ Settings/John\ Doe/My\ Documents'

-If you don 't know how many documents you are going to save, do an 'ls -l'. If you see there are many subfolders, do an 'ls -lR' 'or du -h'. 'ls -lR' is the dos equivalent of a 'dir /s', but better is to do 'du -h'. This will tell you how much data there is in this folder, whether there are big files, small files, many files, etc...
-Once you know this, estimate what you have to rescue. Listen to what the disk has to say, if it starts making funny noises already, try to save important data first.
-Let 's say all is well, start a copy (first make a remote dir to save your files to):
'mkdir /mnt0/recovered'
'cp -rvf * /mnt0/recovered'
This command starts a copy of all files (*) to /mnt0/recovered, all subdirs (r), shows the progress file by file (v) and forces overwriting files on the remote location (f).

If this works out: good for you.

Otherwise...

If your disk starts to make funny noises and you see error messages like "Disk Seek Complete Error", this means we 're running into bad clusters and your copy operation will get slowed down dramatically.
In this case, try to start copying files or folders more separately. Skip the files first on which it runs into errors. There are other means to (maybe) recover them later.
If that works out, but you still need the other files, let 's try someting else. I recommend you let the disk rest for a while by turning it off for about 20 minutes and put it in the fridge wrapped in a towel.
Once you 've started it up again, you might have another go at copying the bad files again, but cancel it as soon as you run into the same problems again.

In this case go over to using ddrescue.

Ddrescue takes an input file and copies it to an output file, just like a normal dd would do. But in this case there are a few differences: it tries to copy raw data from a file, but it will not try too long when running into bad, completely unreadable clusters; it will fill up the unreadable spaces with zeroes.
Example: 'ddrescue /hda1/Documents\ and\ Settings/John\ Doe/My\ Documents\outlook.pst /mnt0/recovered/outlook.pst'
You can see here that the rescue is file by file. Here 's a quick oneline to rescue a complete directory.
'cd /hda1/Documents\ and\ Settings/John\ Doe/My\ Documents\; for i in *; do ddrescue "$i" "/mnt0/recovered/$i; done"'

Another way of trying to rescue your data is to go and copy complete disks or partitions. Just so you can be able to use Windows' chkdsk later on the rescued filesystem, ddrescue the contents to another partition on another disk. It 's also possible to rescue the filesystem to a file and try to mount the file as a loopback device from Linux/TRK, but you won 't be able to let Windows access it.
So here goes an example (make sure nothing is mounted, neither source or destination): 'ddrescue /dev/hda1 /dev/hdb1'
In this case you will raw copy the complete contents of your partition /dev/hda1 to /dev/hdb1. Make sure /dev/hdb1 is equal or bigger dan /dev/hda1
Boot into Windows and let it run a chkdsk on it (or use the Windows install CD in recovery mode).
If your filesystem is native Linux, like an ext2 filesystem, you can rescue it to a normal file somewhere and run e2fsck on it like that.
Example: 'ddrescue /dev/hda1 /mnt0/rootfilesystem.img logfile-hda1'

If ddrescue was unable to rescue all of your hda1 because it had to skip certain sectors, try going into raw mode, which will bypass your kernel disk i/o layer and read sector per sector. In raw mode it 's way slower, but ddrescue keeps his already done work in a logfile and only does the sectors that were skipped.
Use it then like this:
'modprobe raw'
'raw /dev/raw/raw1 /dev/hda1'
'ddrescue -r3 /dev/raw/raw1 /mnt0/recovered-hda1.img logfile-hda1'
Once you got this back, restore the image to another disk that boots into a normal PC into Windows and try running chkdsk against it.

If it was a Linux filesystem (ext2 f.e.), there 's no need to restore it:

'e2fsck -v -f /mnt0/ rootfilesystem.img'
'mount -o loop /mnt0/rootfilesystem.img /mnt1'
<= check your files...

NOTE: it is not recommended to try and perform a filesystem check that starts having physical I/O errors. It will probably make things even more unrecoverable.

3.2 Recovering deleted files or files from formatted drives (ntfsundeleteall, photorec)

In TRK, there are several ways you could go and undelete files. What you should use depends on the state of your harddisk/filesystem and recoverability can vary between various methods.

Ntfsundeletall

First method already explained earlier is ntfsundeleteall (section 2.8), so I 'm not going to re-explain this. What I should mention about this method is that it has a high recoverability ratio on recently deleted files (deleted in a normal way, not because of a crash or filesystem corruption). It does however a poor job on drives that have already had more write I/O over them. It 's completely unuasable on formatted drives.

Photorec

In all other cases, you can try the photorec utility, which tries to recover any important kind of document from any filesystem. Be sure to run it by first changing to a place where you can write your stuff, because it saves its recovered output to a subfolder relative from where it was launched. Example:
'mount //10.0.0.1/share /mnt0'
'cd /mnt0'
'photorec'

If you suspect that your filesystem is corrupt, I would first try to run a chkdsk on it and see what it recovers. Then afterwards, see what show up and depending on what 's become visible, use the ntfs capabilities of TRK or use photorec... But do not run a chkdsk if your disk is physically dying!!! Try to do as little I/O possible on that disk. Running a chkdsk will probably render it even more unusable in that case.

Maybe ddrescue first

If photorec is unable to recover because of too many corrupt sectors, you can try ddrescue.
This util tries to recover files and filesystems that are on locations with bad sectors. If its really unable to read the bad sectors, it will pad them with zeroes. Your files will possibly be corrupted, but at least you 'll have saved whatever there is to save. If you ddrescue a complete filesystem (e.g. 'ddrescue if=/dev/hda1 of=/mnt0/rescued-image.img), you can afterwards mount that image as a loopback filesystem ('mount -o loop /mnt0/rescued-image.img /mnt1') and rescue/undelete files like that.

3.3 Recovering lost partitions (testdisk, gpart, fdisk)

It can happen you did something stupid or something inexplainable happened and your partition table is gone or corrupted.

To recover your partition tables, there are a few methods.

Fdisk

I 'm not talking about the Microsoft dos version of fdisk, but of fdisk on Linux.
Actually what you do with this is manually redesign your partition table. Of course, the risk here is that you have absolutely no idea of what your partition table was before if you had more than one partition. So this is good in case of a disk with a single partition.
And let 's say in this case that partition contains an NTFS filesystem on /dev/hda1
Make sure nothing is mounted (run umountallfs)
Run fdisk:
-'fdisk /dev/hda'
-In fdisk: 'p' to print your actual partition table
If your actual partition table is a bad one, erase by removing each partition with 'd' and answer with the right partition number (in case of logical drives, first remove everything equal and above 5).
-'n' for new partition
Choose primary partition, press enter for start and end sector. Fdisk automatically assigns it as a Linux native partition.
-'t' to toggle the partition type. Ntfs is '7'
-'a' to make it active (choose '1'), since it 's probably the bootpartition
-'w' to write your changes

If this was your previous partition layout, you should now already be able to mount the ntfs partition. If not, you probably won 't be able to boot from it either. But it should be, since this is the way to do it when the problem was simply a lost partition table.

Testdisk

This is the more advanced method of recovering partitions. Run 'testdisk' at the commandline, the rest speaks a little for itself (Intel partition table => Analyze => etc.). Read the testdisk documentation at  www.cgsecurity.org
Testdisk is also very nice to repair bootsectors. Just play with it a little on a testmachine to see what it all does. I 've played around with it myself, and although there 's no guarantee that you will be able to repair the bootsector all the time, it still did a pretty good job where others failed.
Another thing I can recommend you about testdisk: if it doesn 't find all of your partitions in the first minute, there 's probably not much left but some free disk space without a valid filesystem. It 's safe to cancel already if you think it found all you need.

Cfdisk

A menubased fdisk replacement, easier to use than standard fdisk

Gpart

Guess partitions: a utility that might recover your partitions too, but not maintained since a few years and haven 't tested it myself yet.

3.4 Bootsector repair

About bootrecords

Several utilities can be applied to repair a bootsector and there 's a lot to be said about it.
I 'm only describing a few procedures for Windows and Linux bootsectors.

In a PC style (x86) environment there are 2 ways a machine can boot from a harddisk: or from the MBR (Master Boot Record) OR from the bootsector on the active partition. Historically, Windows has always booted from the active partition bootsector, but it also writes some stuff to the MBR of the disk. I don 't know all the internals of the different bootprocesses, so please correct me if this documentation has it wrong.
Continuing: Linux can boot in the two different ways: or it writes its code to the MBR (e.g. /dev/hda), or it writes it to the active partition (e.g. /dev/hda1).

Check partitioning

Sometimes the failure to boot can simply be a partitioning problem. In that case, refer to the previous section. If you rebuild your partitioning like it was before, it is most likely to boot again. Never forget to set the right active partition!

Windows boot

Let 's take a quick look on procedures how to repair Windows bootrecords.
As I already said Windows booting depends on both MBR as partition bootrecord code.

-ms-sys
This utility can write new MBR code to your disk. Type 'ms-sys -h' to get some help on the different parameters, but in most cases you run it first to analyse your disk 'ms-sys /dev/hda', which will tell you how your MBR.
'ms-sys -m /dev/hda' will write an XP style MBR to your disk.
Remember that this utility handles only about the MBR. For the windows bootsectors you must refer to other utils.

-testdisk
I 've tried testdisk to write new MBR code and it worked like a charm for me. What 's also nice about it is when you go into advanced mode, you can analyse also the bootsector of the partitions and repair them if needed.

-the Windows bootcd way
It might happen that none of these methods get you back into booting Windows, then there is one last method that you can try, which actually has nothing to do with TRK: boot from the Windows XP cd and go to the recovery console. Make sure you know the password of the Windows system on the local disk. Otherwise, reset it first with 'winpass' on TRK.
-Now, booted into the recovery console, it 's a good thing to run checkdisk first:
'chkdsk c: /p'
-Next run 'fixboot'
-And finally run 'fixmbr'
-'exit' to reboot

-if really nothing worked: FAT32
Finally, if nothing did the trick, it means something is wrong in the way, I don 't know, your disk geometry was laid out, filesystem were created, etc, etc. Sometimes it 's even a mystery to me why a Windows won 't boot when everything else says it should. But it often happens after I cloned a PC with clonexp or Partition Image. Sometimes it works, sometimes it doesn 't.
In that case there is one last resort: recreate your Windows on a fat32 partition.
For documentation on that, see the section:  3.5 Manually cloning a Windows installation

-untested: relocntfs

I haven 't had the chance to test this utility, but this should be able to make your windows bootsector bootable again by relocating it to the correct addresses. This should be particularly handy after cloning of Windows systems.

3.5 Manually cloning a Windows installation

This method describes how you can migrate a Windows installation to another computer and be sure it will boot as long as the Windows installation itself isn 't corrupt. Additionally, I will also describe how you could make your system bootable again with only a single computer/disk available.

The way we approach this type of "cloning" is very simple: we create a fat32 partition, copy all data from the NTFS partition to it, let the Windows bootcd repair the MBR and bootsectors and Windows boots. Afterwards you can convert the filesystem back to ntfs from Windows.

The procedure

Let 's assume a few things first: you 've got two harddisks attached to 1 PC of which one might be dying or chkdsk doesn 't repair it and you can 't use any cloning tool since it finds errors and exits (happens with Ghost and PQ Drive Image). In the case of laptop, you don 't have the opportunity to attach a second disk unless via an external USB enclosure, which is not recognized under dos based utils.
In TRK you can OR mount a remote disk with two PCs running TRK of which one is running in fileserver or secure shell server mode (see other sections on how to do that). The possibilities are myriad.
But our example here works with two local disks, the source disk being an NTFS partition, the destination disk empty or erasable.
Disks are respectively /dev/hda and /dev/hdb, NTFS partition is /dev/hda1
-Make a large enough vfat filesystem on /dev/hdb. Use qtparted to do it graphically or use
'fdisk /dev/hdb',
'n' new primary partition
-'1'
-'t' to toggle, code 'c'
-'a' set active '1'
-'p' to print all what 's about to be done on the screen
-'w' to write and quit the partition table
-'mkdosfs -F 32 /dev/hdb1'
-mountallfs
-cp -a /hda1/* /hdb1
-Shutdown the PC, change the harddisks so your new disk becomes the primary (/dev/hda)
-Now, reboot with the Windows CD so we can make the disk bootable (it 's the best way I 've experienced).
Follow the procedure described in 3.4 Bootsector repair .i.e. fixboot and fixmbr.

Your Windows should now start. Once booted into windows, go to a commandline and run 'convert c: /fs:ntfs'
Reboot and let your disk convert to ntfs

I 've noticed that the Linux 'cp' makes the windows systemdir (in most cased named WINDOWS) lowercase. It 's good practice to make it uppercase afterwards. Although Windows is supposed to be case independent (but still case aware), I 've noticed some services wouldn 't start until I uppercased it
-'cd /hdb1'
-'mv windows/ WINDOWS/'

One note here: Windows XP will not boot from FAT 32 partitions bigger than 32Gb. So don 't create them bigger than this. If you have more data to copy, I suggest you move your big files somewhere in a directory that you then omit copying.

3.6 Hardware testing

Although TRK doesn 't claim to be the best hardware testmethod there can be, there are still a few utilities and procedures you can use to test the health of your computer.

A lot of information about hardware and eventual errors can already be obtained by reading out the bootprocedure and system information, described in section   1.4 Reading information about your PC (dmesg, /proc/partitions). Be sure to read that first.

Here we 're going to put ourselves a little more on stress testing.

Harddisk testing: Bonnie

Bonnie is a utility that stress tests I/O of your harddisks. If you want to measure performance or you just want your disk to run very hot, just mountallfs your filesystems (for ntfs volumes run 'mountallfs -g') cd to a dir on the filesystem (e.g. 'cd /hda1') and execute 'bonnie'
Here 's what it could show you:
[root@trk]:(/hda1)# bonnie
File './Bonnie.3078', size: 104857600
Writing with putc()...done
Rewriting...done
Writing intelligently...done
Reading with getc()...done
Reading intelligently... done
Seeker 1...Seeker 2...Seeker 3...start 'em...done...done...done...
              -------Sequential Output-------- ---Sequential Input-- --Random--
              -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine    MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU  /sec %CPU
          100  4500 16.3  6820  8.3  5955  9.5  9151 11.6  9985  5.1 132.0  1.5
[root@trk]:(/hda1)#

Bonnie tests a number of write operations on your disk.
If you want it to run infinitely, just type 'run-bonnie'

CPU testing: burnP6

This is a tool for overclockers and anyone who likes to know there computer doesn 't crash under heavy CPU load. It uses 100% cpu and makes bad cpus go mad (probably).

Network testing

Actually, a real network testing utility isn 't included, but you can put a lot of stress on your network card just be transferring big files for instance.

Memory testing

The latest version of memtest+ is included. Just select it from the start menu of TRK (scroll down all the way to the bottom). It tells you information about your memory speed performance and tests to find any possible errors.

3.7 Virus scanning

See section 2.1 Virusscan for all the information you need on virusscanning your PC with TRK.

3.8 Manual PC cleaning

4. Boot time options and triggers

This section gives you more info on the internals of TRK, how and why to use the differen boot menu options and the triggers TRK has builtin for customisation.
For advanced users who want to automate more or need alternate ways of booting.

4.1 Boot menu options

The boot options of the latest TRK 3.2 contain a lot of  different things, but most of the time the default startup will do.

Nevertheless, here are the options explained


-Default startup:
This will run TRK 3.2 with all the default options enabled. This means it will run partly from CD/usb stick, keeping your CD/usb stick unavailable for other things in the mean time, but will boot faster. Keyboard will be default qwerty and fontsize about 3/4 of a normal font in a screensize of 800x600 in VESA compatible framebuffermode. Some videocards are not so VESA compliant and can therefore not benefit from the graphical possibilities of TRK (like the Intel 815 chipset). They will have to run in textmode, preventing them from using qtparted and links (and my nice background logo)

 
1) TRK 3.2 with proxyserver support enabled
Start TRK with default options and prompts you for a proxyserver. In fact, it just calls for "setproxy", which prompts you for the proxy server address (hostname or ip), portnumber and optionally username and password (leave blank for none)

2) TRK 3.2 running from RAM (best >= 256mb, 192mb min)
This option will run Trinity Rescue Kit completely from RAM, which means it copies all of its files into memory so your CD or usb stick becomes available for disconnection. In the case of PXE booted TRKs, you could disconnect the netwsork at that moment. This requires at least 192Mb RAM, but recommended is 256Mb and more. Especially updated TRK versions require about 64Mb more because of the included antivirusccanners..

3) Instead of my preferred smaller 12 pixel font, use the standard 16 pixel font.

4) Run immediately in VGA mode with small fonts
If you know your machine won 't boot in framebuffer mode. As I said before: no links and no qtparted. This option is also useful in case your TRK doesn 't initiate any output on the screen and you suspect it to hang on kernel initialisation. Many times it means there 's a problem with the ACPI interface of the machine. I 've noticed this quite regularly on HP/Compaq machines.

5) Run with Belgian azerty keyboard.
More keyboards can be called upon by small scripts that stand for the initials of your country:

-be: belgium
-uk: united kingdom
-dk danmark
-it: italian
-fr: french
-pt: portugal
-nl: netherlands
-de: deutschland
-es: espana
-se: sweden
-no: norway
-us: united states

Please note that this is only a keyboard remapping, but this does not support any localised characters. This is a rescue distribution, not a desktop replacement

6) Virusscan all drives (non interactive)
runs virusscan on all your drives. Stupid option actually, but this is about the only thing you can do that is set-and-forget without the need for any interaction. This runs virusscan without any arguments, so with ClamAV. You will get no login as long as this is running.

7) Try more pcmcia and usb nics
I found that kudzu is not really good with PCMCIA cards and USB stuff and I found out that the easiest way fix this was to just have support for pcmcia bridges compiled in my kernel, start cardmgr and try every module I find until success. Same with USB nics. This will only try this for PCMCIA and USB ethernet cards/adapters.

8) TRK 3.2 with a secure shell server enabled
Run a ssh server, a very nice feature, giving you the ability to remotely work on a machine that your nexter has started up for you.

For security reasons, you will first get prompted to create/change the password, after which in green will be displayed the IP-addresses your TRK will listen to.

9)  Execute local scripts on harddrive of PC
Run local scripts from ./trk/trk3local.conf found on any local filesystem. In this way you can make computer specific scripts and put them on the harddisk. More info, see section "Triggers"

 10) Fileshare all drives, secured with user
This will run a TRK with a so called Samba fileserver, which is the same as a Windows filesharing server. In this mode, you will get prompted to create a user + password that can access your server. Next, 'mountallfs -g' will run and share all your local drives as their respective device name.

You can start this option afterwards from your TRK console with the command 'fileserver -s'

11)  Fileshare all drives as guest, no security
This is the same as the previous option, but this will share all your drives without any security boundary or username. Be careful with this, use only in a trusted environment.

12) Single user mode
Starts TRK in very basic mode, sometimes useful when your PC hangs or crashes on hardware detection or some other error.

13) Noacpi/noapic PCI=bios - Problematic machines 1
Some problematic machines with lousy ACPI controls might boot with this option. Happens regularly on some older Compaq machines. This will disable acpi and apic and will use the bios' PCI routing table. Try this when TRK hangs on early startup.

14) Noacpi/noapic PCI=any - Problematic machines 2
This is almost the same option as the previous one, but I let the kernel decide what PCI mode should be used. I needed this mode on the brand new HP DC7700 machines with Intel Duo Core because it hung on kernel initialisation in the PCI part.

15) Verbose startup for debugging
I use this to see what scripts are doing or sometimes your PC might hang on something but you don 't know what: try this, but don 't get frightened by all the output.

16) SSH server and run from RAM
This is a combination of option 2 and option 8. This is a quite common combination you would need when cloning PCs from 1 TRK to another and you only have one TRK cd handy.

17) SSH server, run from RAM and run a secured fileserver
This is a combination of option 2, 8 and 10. If you want complete remote control of the TRK machine.

18) Memory tester memtest+ v1.65
Almost every distribution comes with this on startup, but this way it 's included in this one too. Nice to test your memory banks with, but real life tests are still much more certain. Still, it might do the trick for you and find you a bad module.

4.2 Triggers

TRK contains a number of so called "triggers", things it checks for and reacts upon when existing.
Very handy for regulaer TRK users who want to customise it a little.

4.2.1 The TRK options server: make your lan TRK aware

TRK has quite a few ways you can customise it or have custom scripts be run. These scripts can run automatically because I 've built in numerous placeholders for them which I call "triggers"

If you are a regular user of Trinity Rescue Kit, you might find it useful to have some default parameters set according to your environment i.e. lan
Ever since TRK 3.1, there 's a feature that makes your lan "TRK-aware" since a system administrator can configure his lan in a way any TRK3 can take advantage of it in terms of proxy configurations and user optional script executing, a much wanted feature from the past, but hard to implement since you 're working with read-only media.
Finally, the solution is kind of revolutionary: by activating an otherwise unused parameter in you DHCP server, you can have a TRK point to a server where it can find its config and scripts.
This option is from an obsolete RFC from 1983 called 'Resource location server', which was a UDP protocol that could tell you where to find things like routes and smtp servers. I 'm using it now to provide TRK with an IP-address where it can find a simple webserver on which resides /trk/trk3options.conf
If you 're running ISC dhcpd, add this line to /etc/dhcpd.conf:
option resource-location-servers your-trk-webserver-ip-address;
On a Windows server, go to Administrative tools=>DHCP => go to your scope => scope options => option 11 Resource Location Servers.
I wouldn 't worry too much about enabling this option to break something in your network. I 've been running this for some time now at my office where there are about 100 Windows PCs on dhcp. Nobody hasn 't complained so far.

Anyway, when a TRK 3.2 boots and finds a 'resource location server' from DHCP, it will indeed look for resources from this configuration file.
Below is an example of such config file, it more or less speaks for itself.
The file is /var/www/html/trk/trk3options.conf on my local webserver

TRKSECTION WGET BEGIN
http_proxy=http://10.34.5.201:8080
proxy_user=johndoe
proxy_passwd=doesinc
timeout = 10
tries = 2
TRKSECTION WGET END

TRKSECTION LINKS BEGIN
http_proxy 10.34.5.201:8080
TRKSECTION LINKS END

TRKSECTION CHECKUPDATESARGS BEGIN
-proxy-server=http://10.34.5.201:8080
-proxy-username=johndoe
-proxy-password=doesinc
TRKSECTION CHECKUPDATESARGS END

TRKSECTION TRKSCRIPT BEGIN
#!/bin/bash
#
if [ -f /etc/proxy.conf ]; then
echo "Setting proxy params in freshclam.conf"
if [ ! -f /etc/freshclam.conf.bak ]; then
cp -f /etc/freshclam.conf /etc/freshclam.conf.bak
fi

if ! [ "rjohndoe" = "r" ]; then
sed "s/# Proxy settings/\nHTTPProxyServer 10.34.5.201\nHTTPProxyPort 8080\nHTTPProxyUsername johndoe\nHTTPProxyPassword doesinc/" /etc/freshclam.conf > /etc/fresclam.conf~
else
sed "s/# Proxy settings/\nHTTPProxyServer 10.34.5.201\nHTTPProxyPort 8080/" /etc/freshclam.conf > /etc/fresclam.conf~
fi;
mv -f /etc/fresclam.conf~ /etc/freshclam.conf
chmod 700 /etc/freshclam.conf
if [ -f /etc/avg.conf ]; then
echo "Setting proxy params in avg.conf"
if [ -f /etc/avg.conf.bak ]; then
# Take a backup of the original avg.conf if necessary
cp -f /etc/avg.conf /etc/avg.conf.bak
fi;
sed -e s'/proxy = off/proxy = 10.34.5.201:8080/' /etc/avg.conf > /etc/avg.conf~
mv -f /etc/avg.conf~ /etc/avg.conf
chmod 700 /etc/avg.conf
sed -e s'/proxyLogin = off/proxyLogin = johndoe:doesinc/' /etc/avg.conf > /etc/avg.conf~
mv -f /etc/avg.conf~ /etc/avg.conf
fi;

if [ -f /opt/BitDefender-scanner/etc/bdscan.conf ]; then
echo "Setting proxy params in bdscan.conf"
if [ ! -f /opt/BitDefender-scanner/etc/bdscan.conf.bak ]; then
cp -f  /opt/BitDefender-scanner/etc/bdscan.conf /opt/BitDefender-scanner/etc/bdscan.conf.bak 2>/dev/null
sed "s/#HttpProxy =/\nHttpProxy = johndoe:doesinc@10.34.5.201:8080/" /opt/BitDefender-scanner/etc/bdscan.conf > /opt/BitDefender-scanner/etc/bdscan.conf~
mv /opt/BitDefender-scanner/etc/bdscan.conf~ /opt/BitDefender-scanner/etc/bdscan.conf

fi;
fi;
fi;
TRKSECTION TRKSCRIPT END


-TRKSCRIPT is a section where you can do basically anything you want, so be very carefull with what you put in here because it will have an effect on any machine you boot with TRK3 . It will basically execute as a script in the same environment from the last script run by TRK, /etc/init.d/trklocal
What I put in the example here is to download an adapted version of the ClamAV config file 'freshclam.conf'. Refer to the ClamAV manpages for help on this file, but in here I 've put my proxy parameters too.
I don 't need to tell you what other opportunities this opens up.

If you think this file is too complicated for you to create, don 't worry: just start a TRK, run 'setproxy' and this will generate the file /etc/trk3options.conf
Copy this file to your webserver and you 're all set.

If you run setproxy afterwards, you will override alls settings given by your trk3optionsserver.

4.2.2 Scripts on the computer's local harddisks

Another trigger is to have a script executed from the computer 's local harddisk, allowing you to run commands specifically for the computer. It is in the same format as the trk3options.conf: you have TRKSECTION TRKSCRIPT (only this section) where you write your parameters. This will be generated into a script called /bin/trklocscript-<name of the drive> (f.i. /bin/trklocscript-hda1) and executed at the end of the TRK startup procedure.
The location of this script has to be in the folder .trk (dot trk) on the root of the local partition. In this folder you put the file called trk3local.conf.
A sample trk3local.conf could look like this:

TRKSECTION TRKSCRIPT BEGIN

#!/bin/bash
# Script for maintenance and backup on Windows machines
# Clean up this drive a little and do a virusscan on it
cd /hda1/Documents\ and\ Settings
for i in *; do rm -rf "$i"\Local\ Settings\Temp\*; rm -rf "$i"\Local\ Settings\Temporary Internet Files\*;
done;
virusscan -a fprot
# Now back up the userdata of all local profiles to a network drive
mount -o username=john,password=doe //10.0.0.5/backups /mnt0
for i in *; do mkdir /mnt0/"$i" 2>/dev/null;  echo "Backing up documents of $i"; cp -rvf "$i"/My\ Documents /mnt0/"$i"; done
umountallfs

TRKSECTION TRKSCRIPT END

4.2.3 Script on the TRK medium

If you want some modifications to TRK and don 't want to fiddle directly into TRK's own scripts, you can add a script to the CD/usb stick/NFS share as  /trk3/trklocscript. Unlike the previous trigger, this has to be a normal bash script.
The advantage of having scripts outside the TRK files is that you can upgrade your TRK afterwards, without having to redo your work every time: in case of a USB stick or NFS share, you just leave the script where it is, in case of a CD, you re-add the trklocscript to the multisession bootcd.
From this script you can trigger any other script you like, f. i. adding more software to your TRK.
Example trklocscript:

#!/bin/bash
# Go to /tmp, which is on virtual shared memory, about half the size of your RAM
cd /tmp
mkdir /tmp/extrabin
cd extrabin
# Use the variable $TRKMOUNTDIR so you always copy/execute from the right location
tar xzf $TRKMOUNTDIR/trk3/extrabinaries.tar.gz
# This will set the PATH globally
export PATH=$PATH:/tmp/extrabin

4.2.4 Antivirus on CD

The existence of the 3 extra virusscanners on the TRK medium is enough to have them unpacked and ready to be used upong startup of TRK.
If you run updatetrk and your virusscanners get added to your TRK medium, you can later upgrade TRK with a newer version and still keep the antiviri available.
You can also update the antivirus signature files by manually updating them directly into the archive. But I doubt you would do it like is.

5. Upgrade, update and change of bootmedia procedures

TRK 3.2 is able to boot from more than CD alone. You can boot it from a USB stick, USB disk, even a fixed harddisk if you like. And the most nifty feature: boot it from network over PXE!
This part of the documentation describes how to change over to different bootmethods, upgrade and customise more of TRK.
Putting TRK to a USb stick shouldn 't take more than 5 minutes.
Making TRK run from PXE takes a little longer to set up, but it 's worthwhile when you have a big lan to manage. Once set up, upgrading to new versions is piece of cake.

5.1 How to install/upgrade your USB media to run the latest version of TRK

Installing from Windows

This procedure is also valid for new TRK installations on any bootable media that has been:
-formatted as FAT16 or FAT32 (Syslinux documentation recommends FAT16, but FAT32 is tested and works as well)
-smaller or equal than 1Gb (=max 16k clusters). Bigger (and fat32) is allowed but is subject to less compatibility. Please report me on any PC that refused to boot in that way.
-has this partition made active
-has TRK_3-2 as a volume name (exactly like this). But this is not guaranteed to boot from any PC that can boot USB. Better is to use trk2usb from TRK itself (when booted from CD of course).

The rest of the installation course is like upgrading, so follow the guidelins below. If you 're unable to bring it to a success using this procedure, I recommend you download the ISO version of TRK, burn and start from that, plug in your USB stick and run trk2usb

Upgrading from Windows 

Prerequisites: Winzip, Winrar or any archive application that can open tar.gz files. With winimage you could also open the iso file version should you have downloaded that one. This results to the same.
-Download trinity-rescue-kit.3.2-build-274.tar.gz (or the latest version)
-Plug in your USB stick with the old version of TRK 3.2
-Open the file trinity-rescue-kit.3.2-build-274.tar.gz and extract it to your USB drive root (we will call it the G: drive in this case)
-Open a command prompt  'cmd'
-Go to your USB drive: 'G:'
-'cd trk3'
-'syslinux G:' ('syslinux -f G:' if it refuses because windows reports it to be a fixed drive)
-'exit'
-eject the device and try booting from it (use the 'safely remove hardware' from the system tray)

Installing from Linux

I 'm not describing in detail how you should install TRK from Linux, since it might require you to install special packages and perform complicated operations. Someone who doesn 't fear this will have enough with this short procedure
Here goes:
-prerequisites: mtools, syslinux 3.31
-figure out your USB stick device (most of the time /dev/sda)
-run 'mkdiskimage -4 /dev/sda -s 0 64 32' (from the syslinux package). This will completely erase and format your drive to maximum 1Gb. If you want to pass this limit, use fdisk and mkvfat -f 32 /dev/sda1, but beware this can be less compatible.
-edit /etc/mtools.conf and add the line: drive c: file="/dev/sda"
-'mlabel c:TRK_3-2'
-the next steps are the same as upgrading from Linux, so read below

Upgrading from Linux

-Download trinity-rescue-kit.3.2-build-274.tar.gz (or the latest version)
-Plug in your USB stick with the old version of TRK 3.2
-Check what device id it has been assigned using 'dmesg | tail'
-Assuming your USB stick is /dev/sda and your TRK partition /dev/sda4 and we use /mnt/disk as mountpoint: 'mount /dev/sda4 /mnt/disk'
-'cd /mnt/disk'
-assuming trinity-rescue-kit.3.2-build-274.tar.gz is downloaded to /home/user/:
'tar xzf /home/user/trinity-rescue-kit.3.2-build-274.tar.gz '
-'cp trk3/syslinux /tmp' *
-'cd /tmp'
-'umount /mnt/disk'
-'/tmp/syslinux  /dev/sda4'

*: you need to copy the syslinux program of the TRK medium because syslinux will complain that the medium is mounted. So you need to umount it afterwards.

5.2 Setting up your PXE boot environment

Ever wondered what you could do with that PXE stuff from your network card at boot time? Well you could run a TRK on your network.

1. Prerequisites

Here 's what you need:
-the latest TRK
-A (wired) local area network
-Computers with PXE compliant hardware
-A decent configurable dhcp server
-A Linux machine or any OS that can run an NFS server (not Windows, or else a third party soft)
-A TFTP server (these last 3 things can all run on the same server)


Time to install and configure everything: 15 minutes (if you understand everything directly)

2. Put the TRK files somewhere

Download the latest tar.gz distributed TRK version (version 275 and up) and unpack it somewhere on your Linux nfs/tftp server. In this case we 're going to assume unpacking it to /home/trkfiles. User and group ownership is ok for root, just make sure it 's world readable (is normally ok, but just in case)
commands:
'cd /home/trkfiles'
'tar xzf /tmp/trinity-rescue-kit-3.2-build-275.tar.gz'

You can also copy your files from your TRK CD, usb stick or mount the iso file with a loopback device and copy everything from there.

Keep in mind: This is the location where all of your TRK files will reside. The folder you are about to use will be the same for your TFTP server as for your NFS share. Make sure they are the same, not a subfolder of the other!

There 's another thing you should do, that 's generating a default pxelinux config file  (/home/trkfiles/pxelinux.cfg/default). I 've created a small script that does it for you, based on the normal isolinux/syslinux config files. This is something you need to do everytime you upgrade TRK
-Go to /home/trkfiles/trk3 'cd /home/trkfiles/trk3'
-'./mkpxelinux'
The script will prompt you for an nfs path to specify. This is the ip-address of your server combined with the path where TRK resides. This is necessary for the secondary startup phase of TRK. If you omit this, you will get weird results when booting.
An example you could fill in is '192.168.81.5:/home/trkfiles'

3. Setting up DHCP

I 'm going to discuss 2 DHCP servers.

First the ISC dhcp server, which is about the standard in most Linux distributions.
Here 's what the dhcp.conf should more or less look like, depending on your site configuration.
It 's what 's in bold that 's important.
Note that in this situation ALL of your machines will be able to boot from PXE.
If you omit the PXE parameters ('next-server' which is your TFTP server and 'filename') from the general dhcp parameters and put them in the section 'group' (what' s commented out with #), you can specify PXE booting per host based on the host's mac address.

allow booting;
allow bootp;
ddns-update-style none;
subnet 192.168.81.0 netmask 255.255.255.0 {
    # default gateway
    option routers 192.168.81.2;
    option subnet-mask 255.255.255.0;

    option domain-name "trinityhome.local";
    option broadcast-address 192.168.81.255;

    # Seting up an ip address is better here
    option domain-name-servers 192.168.81.2;
    option nis-domain "trinityhome.local";

    range dynamic-bootp 192.168.81.128 192.168.81.254;
    default-lease-time 21600;
    max-lease-time 43200;
    # PXE directives
    next-server 192.168.81.5;
    filename "/pxelinux.0";

#    group {
#    host testtrk {
#    hardware ethernet 00:0C:29:A1:E9:E5;
#    fixed-address 192.168.81.253;
#    next-server 192.168.81.5;
#    filename "/pxelinux.0";
#        }
#    }
 
Second DHCP server is with a Microsoft DHCP.
-Run  the DHCP snap-in
-Go to  your active scope => scope options => right click in it and select  'Configure options'
-Activate option '066 Bootserver hostname' and option '067 Bootfile Name'.
-Put for hostname the IP-address of your TFTP server alltogether.
-For bootfilename '/pxelinux.0'

4. Setting up a TFTP server (primary bootprocess)

I 'm using the tftp-hpa-0.43a TFTP server, which is recommended by the syslinux developer. Download it from kernel.org: http://www.kernel.org/pub/software/network/tftp/
Download, unpack, configure and make && make install it.
It 's also possible it was already in your distribution, so you don 't need to compile it.
To have the tftp server run from the commandline as a daemon, using /home/trkfiles as the dir where your TRK resides, type 'in.ftpd -l -s /home/trkfiles'
In fact, this whole procedure so far is also explained on the syslinux homepage.

5. Setting up NFS (secondary bootprocess)

This is for a very basic NFS setup. TRK only needs read only access on NFS, so there 's no big security issue for this simple setup.
NFS relies on 1 configuration file for its shares: /etc/exports
Edit /etc/exports and add the next line, again assuming /home/trkfiles is where your TRK resides. This is the same folder as your TFTP server root!
'/home/trkfiles   *(ro,no_root_squash)'
Restart your NFS service.


That 's it, you 're all set. Take a machine in your lan, power it on, press F12 and boot from network with TRK. It should boot even faster than from CD or disk.
When a newer version of TRK appears, all you need to do is download and unpack in the same dir where it resided before and regenerate your PXE config with mkpxelinux

In the event you gave wrong parameters or your NFS has any problem at all to get mounted, TRK will drop to a basic shell allowing you to debug what's happened.

6. Custom security and triggers

Setting it up

In some cases you cannot have the physical console of a machine, but it can be remotely booted from network with TRK  on PXE.
You could run a ssh server, but TRK by default will not run one because there would be no security (the default password for root is 'trk').
In this case I 've developed a possibility to have a custom shadow file (=holds encrypted versions of passwords on a Linux machine) or even better the use of RSA public keys.
Furthermore, there 's also a check for a custom, per host trk3options.conf file, should you not have a trk3optionsserver environment (feature since TRK 3.1).
Also it will look for a custom trklocscript, so you can launch anything else you like.
Here 's the way to set it up:
-In the directory where your TRK files reside, create a directory called 'pxeconfig'
-in the directory pxeconfig, create directories named according the mac-address of your custom hosts.
If your hosts network card is 00:0C:29:A1:B9:E5, then create that directory, but use hyphens instead of colons, just to avoid filesystems that don 't accept colons. So in this case create directory '00-0C-29-A1-B9-E5'

Security and other triggers

-Per host shadow file
In the mac-address based directory you can now put a custom shadow file. This will be detected upon startup of TRK on the host with the matching network interface and copied locally.
Once copied, TRK will start a secure shell server allowing remote control of the machine with the matching root password from the custom shadow file.

-Per host authorized_keys file
Another way of  having custom security for ssh is the use of rsa key authentication. This is fairly easy to setup. The method of public/private key authentication is standard secure shell and is explained anywhere on the internet. I will therefore not explain how you should generate a public/private keypair.
What 's important is that you copy the public key of the root user to a file called 'authorized_keys'. In general, one key is enough, so it 's ok to copy the file id_rsa.pub or identity.pub to 'authorized_keys' into our /<trkpxefiles>/pxeconfig/<mac-addres>/ directory.
Upon startup, this file will be detected and gets precedence over a custom shadow file. Once this file is copied, TRK will disable password based authentication, so you can only login with your private key

-Per host or general trk3options.conf and trklocscript file
Just like with a trkoptionsserver, you can have a custom trk3options.conf file in which you can specify custom proxy settings or a complete script. Just put the file in /<trkpxefiles>/pxeconfig/<mac-addres>/  for per host trk3options.conf or in /<trkpxefiles>/pxeconfig/ to have this file for general use for all local TRK hosts.
The same principle goes for trklocscript, a script that will get executed when found. Place this script in /<trkpxefiles>/trk3/ and it will get executed on startup.

Addendum TRK 3.3: what 's new since 3.2?

This document is an addition to the documentation of TRK 3.2. It tells you breefly all you need to know about the new features and how to use them.

-5th virusscanner: Vexira. Run virusscan -a va to get the newly added Vexira AV downloaded and run on your computer

 -updatetrk -z / zimrun:  Novell Zenworks Imaging (only TRK 3.3 build 310, broken in newer builds), a wonderful tool to replicate PCs over multicast (=an infinite number of replicas in just one run!). The tool is free but not quite GPL, so it 's not included by default into TRK. Use updatetrk to have it integrated. The syntax for it (skipping all antiviri) is 'updatetrk -s clam,fprot,avg,bde -z -b /hda1/temp' <= where "-b /hda1/temp" is the specific buildlocation on the computer 's local harddrive, since it needs quite some space to get built. Once booted from your updated TRK, type 'zimrun' to get Zenworks Imaging. It 's a menu based tool and self explanatory.
As stated above, this functionality is broken because of kernel/application incompatibilities. As of TRK 3.3 build 321, please use mclone (which is much faster and has better features  too).

-updatetrk -t:  Seagate/Maxtor Seatools harddrive testing. This is a tool completely outside of TRK that runs under FreeDos. I 've added this to updatetrk. Add the "-t" flag to get it downloaded from the Seagate website and have it integrated. When booting an updated TRK, you will see an extra option from the initial TRK bootsplash menu.

-rkhunter and chkrootkit: rootkit detection utilities, only for unix/linux machines unfortunately

-manpages for all common commands now included

-fileserver -r: filesharing can be run as read-only (option -r)

-tryscsi: try to detect your scsi controller, also available as startup option

-dd_rescue: alternative to ddrescue

-ide/sata raid support: mountallfs -l scans for software raid configurations, puts them together and mounts them

-trkbootnet: boot TRK over the network from another TRK: use the script  trkbootnet to boot a TRK from CD or USB stick and then boot other machines over PXE without any change to your network environment (no dhcp modifications, no resident tftp nor nfs server)
How does it work?
Well, the initial bootprinciple is similar to Microsoft RIS or Novell Zenworks network boot.
Thanks to a long existing patch from Michael Johnston for the ISC dhcp server, this dhcp is able to act as a proxy dhcp server. In short it means that when a PXE client receives boot parameters from a dhcp server like an ip-address but no PXE information, it broadcasts for the presence of a proxy dhcp server that could provide it with additional information like the location of a PXE server.
The built-in PXE server from this ISC dhcp then responds with parameters about a TFTP server and a bootfilename.
Next, the client gets its bootfile from the TFTP server, which is also running on TRK and boots the TRK kernel.
The "TRK server" is also acting as an nfs server so TRK can also run the second bootstage.

-TRK 3.3 is now volume label independent: you can put any volume label you like on your CD/USB stick. You just have to pass the volume label in your syslinux.cfg/isolinux.cfg by adding vollabel=YOURLABEL to the command line. Interesting for people making multiboot CDs.

-winpass rewritten (+ backup option): winpass is for 80% rewritten. It is now completely case independent, with much  better code, so it should give no problems anymore on windows installations not being found. It also now makes a backup of the original hive and has a restore mode (winpass -r). So you can reset a password, get in the pc and put back the original password.

-trinisup: This script allows to give remote support on a running TRK. It makes a connection to an intermediate server and opens up a reverse ssh port. Use any port between 6000 and 6500 as a remote port to connect to. This port will get mapped to the ssh server running on TRK. It uses the same principle as the Trinity Remote Support Pack (read all about that on the trinityhome.org new pages if you want to know more about it). This is to be used in cooperation with the Trinity support team (the people from Opencomputing). You can get an account to use it for a small fee or use the 2 minutes limited demo account (password "gogogo")

-disable autorun from windows: If you put in the TRK cd from Windows, you
will get a warning that TRK is not meant to run from Windows. Furthermore, it tells you that autorun is a bad thing (autorun viruses and such) and asks you if it can disable autorun altogether.