Bitcoin: distributed crypto-currency

This guide will cover how to build a dedicated/headless/automated GPU Bitcoin mining rig running the Linuxcoin operating system and Smartcoin miner management.

Wait... is Bitcoin mining even worth it?

In terms of profitability, it all depends on the how much you pay for electricity, the current mining difficulty, and the BTC exchange rate. There's some calculators on the web that can help you figure out your projected costs and BTC earnings.

Would spending over $1K on a rig be profitable in the long run? Probably not, unless you're getting very cheap electricity. Also consider that the current payout for a solved block is 50 BTC. Every 210,000 blocks, the BTC payout for a solved block is halved. It's estimated that the payout will be reduced to 25 BTC in December 2013, greatly reducing the incentive to mine Bitcoins. But who knows? The exchange rate of BTC may be up to $50 USD by then.

Maybe you're not in it for profit, but instead to support Bitcoin. The security of the Bitcoin network relies completely on the collective computational power of miners. Bitcoin mining ensures that new block creation is not controlled by a single entity.

Disclaimer:
I still consider myself a Bitcoin mining newbie, and there's a good chance that the info in this post could be improved. This guide is simply documentation of how I got my rig up and running, so use it at your own risk.

Selecting the hardware

2x5970s

The goal is to choose hardware that:

• will maximize your hashrate.
• will minimize your electricity usage.
• will not break your bank account and eventually be profitable.
• could either be resold or reused in some other scenario.

Your hardware and component quality will be the primary determining factor of your rig's BTC mining performance. Certain components (GPU, PSU, MOBO) are more important in this regard than others (HDD, RAM, CPU). Here's an overview of the hardware I chose, and why:

Make sure your MOBO has enough room and slots for your graphics cards. If you get multiple graphics cards, there should be at least 1 inch of space between them so heat they generate can dissipate. Keeping the graphics cards cool is primary: it will increase both their performance and life expectancy.

I'm not going to cover assembling the hardware as that will be relative to your rig. Just make sure to put the rig in a place where it can breathe. If you put it in a closet, leave the door open, and point the GPU fans so they blow out the door.

Safety first: make sure the electrical circuit you plug your rig into can handle the load. You want to keep your max draw under 60-80% of what the circuit can handle (most regular US household circuits/breakers are rated at 15 amps). So if you've got other high wattage electronics running on the same circuit, add up all the amperage of each device (Watts / Volts = Amps). Don't start a fire.

UPDATE: I started a github repo that has some of the example files that are created/modified in this guide. I also wrote a few bash scripts that can automate many of the setup steps below. Look for the one-liner commands in this guide: they can be used instead of some of the manual steps.

Install Linuxcoin on the USB, with persistence

Your USB device will be acting as the rig's hard drive and it's recommended be at least 4GB, but you could probably get away with 2GB. It will be running Linuxcoin 02.b-final with persistence so that written data survives across boots. Linuxcoin comes with pretty much everything you need to get up and mining fast.

This guide uses a Windows 7 desktop to setup the USB. Many of the following steps came from a combination sources listed at the bottom of this page.

1. Download the following stuff:
   • LinuxCoin 0.2b-final ISO
   • BOOTICE - used to partition the USB drive
   • unetbootin - used to install Linuxcoin
2. Plug the USB stick into your desktop. Right-click the drive, choose "Format...", and format it with the default FAT32 file system.
3. Run the downloaded bootice.exe. Select the USB device from the "Destination Disk" dropdown.

4. Click the "Parts Manage" button, then "ReFormat USB Disk". Set:

   • Disk Mode = USB-HDD mode (Multi-Partitions)
   • Partition Alignment = Align to cylinder

5. Click Next and set:

   • size field of Partition 1 = 1024 MB
   • File System for Partition 1 and Partition 4 = FAT32

   Leave the remaining fields at their defaults. Click OK. You'll be prompted to confirm the reformat and partitions a couple of times, just click Yes/OK on each one.

6. When it's finished, close BOOTICE. Run the downloaded unetbootin exe file. Set the following values:

   • Diskimage option enabled
   • ISO option enabled and browse to and select the Linuxcoin ISO file
   • "Space used to preserve files..." = 1024 MB
7. Click OK and wait awhile. When it finishes, choose "No" when UNetbootin prompts to reboot.
8. Plug your USB stick, a mouse, a keyboard, network cable, and a monitor into the mining rig. I connected my monitor into one of the 5970's DVI plugs. Power the rig on. You should see the Unetbootin splash screen - if not, make sure your BIOS allows for botting from a USB device. Select "Linuxcoin Persistent".

9. Once Linuxcoin boots up, it should automatically display a desktop. If you're seeing a login screen, enter username = user and the password = live.

   Now we want to create a Linux file system on the USB's persistence partition so that data persists across boots. Click the yellow Ƀ button in the bottom left-hand corner, then goto Accessories > Root Terminal.

   * one-liner: The rest of the stuff for persistence (step #9) can be done automatically with this bash command:

   # curl https://raw.github.com/timbowhite/linuxcoin-setup/master/setup/persistence.sh | bash
   

   Or you can do it manually:

   Run df to see your mounted file systems:

   # df
   

   Look for /dev/sda2, which will be mounted on at a temporary moutnpoint like /media/AE90-5228 or something similar. Unmount it and use mkfs to create the live-rw filesystem:

   # umount /media/AE90-5228
   # mkfs.ext4 /dev/sda2 -L live-rw
   

   Edit /live/image/syslinux.cfg, and add a new boot entry so that Linuxcoin boots with persistence by default:

   label unetbootindefault
   menu label LinuxCoin
   kernel /live/vmlinuz
   append initrd=/live/initrd.img boot=live config quiet persistent rw vga=791 edd=off
   

   The above entry should be placed as the first entry underneath the UNetbootin menu section:

   default menu.c32
   prompt 0
   menu title UNetbootin
   timeout 100
   

   Example /live/image/syslinux.cfg available here.

   Reboot the machine. The bootloader will now automatically boot Linuxcoin with persistence.

Make it headless

The goal here is to be able to run the rig without it being attached to any I/O devices: monitor, mouse, keyboard, etc. You can then store the rig somewhere else and access it via SSH on your LAN.

* one-liner: most of the following steps for "headless-ness" can be done automatically with this bash command:

# curl https://raw.github.com/timbowhite/linuxcoin-setup/master/setup/headless.sh | bash

* You will still need to manually assign the correct network settings in your /etc/rc.local. The above command assigns a default 192.168.0.150 static IP address, see step #2. SSH should also be manually secured, see steps #5 and #6.

The manual steps:

1. First, update all packages the system:

   # apt-get update
   

2. Next, assign a static IP by putting the following lines in /etc/rc.local. The example below sets the static IP to 192.168.0.150 with a gateway of 192.168.0.1. Your network settings will different, make sure to change them appropiately:

   sudo ifconfig eth0:0 address 192.168.0.150 up
   sudo ifconfig eth0:0 broadcast 192.168.0.255
   sudo ifconfig eth0:0 netmask 255.255.255.0
   sudo route add default gw 192.168.0.1
   

   Example /etc/rc.local available here.

   Now run rc.local and restart the network interface:

   # /etc/rc.local
   # /etc/init.d/networking restart
   

3. The default user in Linuxcoin is simply the user user. We'll setup user as a password-less sudoer to make automation and running the miners as root seamless:

   # echo "user ALL = NOPASSWD : ALL" >> /etc/sudoers
   

4. Start SSH and enable the SSH daemon to start on boot:

   # /etc/init.d/ssh start
   # update-rc.d ssh enable
   

   You should now be able to detach the monitor, keyboard, mouse, and throw the rig in a closet somewhere. Make sure to connect it to your LAN. The remainder of the setup will all be done via SSH.

5. Now SSH into the rig from your desktop. The user password is the word live:

   # ssh user@RIG.IP.ADDRESS.HERE
   

6. Warning: this step is optional, but the current SSH configuration is very insecure. It's recommended that you use SSH auth keys to secure SSH logins, or at the very least, change the user's password.

   For instance, you can configure SSH to disable root login and use SSH auth keys by editing /etc/ssh/sshd_config:

   PermitRootLogin no
   PasswordAuthentication no
   

   Add your public SSH key to the user's authorized_keys files:

   # mkdir -p /home/user/.ssh
   # vi /home/user/.ssh/authorized_keys
   [PASTE KEY ON A NEW LINE]
   # sudo /etc/init.d/ssh restart
   

   Open up a new SSH session from your desktop and ensure you're able to SSH into the rig using your auth key.

Initialize and overclock the GPU(s)

We'll be using aticonfig to initialize and overclock the GPU(s) on your rig. Overclocking the GPU cores is optional: the higher your core clock, the higher the hashrate, the more BTC you'll make. But it's not necessarily recommended as it will increase the graphics card's running temperatures and decrease their life expectancy.

Underclocking the GPU(s) memory clocks is recommended as it lowers electricity usage and doesn't affect hashrates.

1. First, "accept" the AMD license agreement by extracting the icd-registration.tgz (this could've also been done manually via the rig's GUI desktop menu):

   $ sudo su -l
   # cd /
   # tar zxvf /opt/AMD-APP-SDK-v2.4-lnx64/icd-registration.tgz
   

2. su as user, and add the AMD app SDK library to the LD_LIBRARY_PATH system variable, export the display for all users, and reboot:

   # su -l user
   $ echo "export LD_LIBRARY_PATH=\"/opt/AMD-APP-SDK-v2.4-lnx64/lib/x86_64/:\$LD_LIBRARY_PATH\"" >> ~/.bashrc
   $ echo "export DISPLAY=\":0\"" >> ~/.bashrc
   $ echo "xhost +" >> ~/.bashrc
   $ echo ":0" > ~/.display
   $ sudo shutdown -r now
   

3. When the rig comes back up, ensure that aticonfig has detected your GPU's. Keep in mind that some cards (like the 5970) have two GPU's per card:

   $ sudo aticonfig --adapter=all --odgt
   

   You should see a line starting with "Adapter X" for each GPU, followed by some info about it and it's current temperature.

4. Overclocking

   Overclocking can be a trial-and-error process: if you overclock your GPU cores too high, your cards will experience frequent hard lockups, and can only be revived by rebooting the rig.

   If you decide to overclock your GPU cores, start out overclocking slightly above the stock core MHz value. Gradually increase the core clock while monitoring your GPU temperatures. If hard lockups start to occur, roll back the core MHz until the system is stable.

   Enable overclocking by:

   # sudo aticonfig --adapter=all --od-enable
   

   Then set the clocks for each GPU. In the example below there are 4 GPUs, and it's setting the core clock = 750MHz and the memory clock = 300MHz on each GPU. The format of the --od-setclocks parameter is core clock MHz,memory clock. Again, I recommend underclocking the memory MHz to the lowest configurable value to reduce electricity consumption:

   # sudo aticonfig --adapter=0 --od-setclocks=750,300
   # sudo aticonfig --adapter=1 --od-setclocks=750,300
   # sudo aticonfig --adapter=2 --od-setclocks=750,300
   # sudo aticonfig --adapter=3 --od-setclocks=750,300
   

   Commit the clock settings so they'll persist across reboot:

   # sudo aticonfig --odcc --adapter=all
   

Setup miners and automate Smartcoin

1. You may want to update the miners that came with Linuxcoin to their latest versions. For example, I use the Phoenix 1.7.3 miner, and updated it by:

   $ sudo su -l
   # cd /opt/miners/
   # mv phoenix phoenix-old
   # wget https://github.com/jedi95/Phoenix-Miner/tarball/master -O phoenix.tar.gz
   # tar zxvf phoenix.tar.gz && rm phoenix.tar.gz
   # mv jedi95-Phoenix-Miner-297945a phoenix
   

2. Before starting Smartcoin, let's get it updated to it's latest version and automated to start whenever the rig is booted. This way you won't have to SSH into the box to start Smartcoin whenever you the machine on.

   First, update to the latest version of smartcoin:

   $ sudo su -l
   # cd /opt/smartcoin
   # svn update
   

   Create a desktop autostart entry for user. This is run everytime the rig boots and user is autologged in to an X session:

   # su -l user
   $ mkdir -p ~/.config/autostart
   

   Paste in this text into a new file at /home/user/.config/autostart/auto.desktop:

   [Desktop Entry]
   Type=Application
   Name="autostart-smartcoin"
   Exec="/home/user/autostart-smartcoin.sh"
   Terminal=true
   

   Now create a Smartcoin autostart bash script that will start Smartcoin running under root at /home/user/autostart-smartcoin.sh:

   sleep 10
   xhost +
   sudo smartcoin --silent
   

   And make it executable:

   $ chmod +x /home/user/autostart-smartcoin.sh
   

   Smartcoin has the capability to run a custom-defined initialization bash script whenever it's started. We'll use this feature to ensure the GPU fans are always turned up to 80% when mining starts:

   Edit /opt/smartcoin/init.sh and paste the following:

   #!/bin/bash
   
   DISPLAY=:0.0 aticonfig --pplib-cmd "set fanspeed 0 80"
   

   NOTE: If you've got more than one card, make sure to add a fanspeed command targetting each card. You'll have to change the DISPLAY variable on each line to match the GPU # (ie. DISPLAY=:0.1, DISPLAY=:0.2, etc). If your card has two GPU's in it (like the 5970's), then increment the DISPLAY variable by 2. For instance, I have two 5970's, so my init.sh looks like this:

   #!/bin/bash
   
   DISPLAY=:0.0 aticonfig --pplib-cmd "set fanspeed 0 80"
   DISPLAY=:0.2 aticonfig --pplib-cmd "set fanspeed 0 80"
   

   You can optionally create a script at /opt/smartcoin/lockup.sh that gets executed whenever your GPU(s) lockup. It's useful for automatically rebooting whenever a lockup occurs. For instance, a simple lockup.sh to reboot on lockup:

   #!/bin/bash
   
   sleep 5
   /etc/init.d/gdm stop
   shutdown -fr now
   smartcoin --kill
   

   You could also get fancy with it and send yourself an email notification that reports the GPU stats at the time of the lockup, example available here.

   Make sure the above scripts are all executable:

   $ sudo chmod +x /opt/smartcoin/init.sh
   # sudo chmod +x /opt/smartcoin/lockup.sh
   

3. Now we need to install Smartcoin for the root user:

$ sudo su -l
# /opt/smartcoin/smartcoin_install.sh

During the Smartcoin install process, you should probably enter the default "yes" answer to the prompts to autodetect your rig's GPU(s), miners and environment. You might be prompted to choose the location of the Phoenix miner:

Found phoenix miner installed on local system
1) /live/cow/opt/miners/phoenix-old/phoenix.py
2) /live/cow/opt/miners/phoenix/phoenix.py
3) /opt/miners/phoenix-old/phoenix.py
4) /opt/miners/phoenix/phoenix.py
Select the phoenix installation from the list above

In which case you should choose "4", which is the path to the latest version of the Phoenix miner you previously downloaded.

I think the last prompt is how many daily mining minutes you'd like to donate to Smartcoin: your call, but donations help keep Smartcoin awesome and free.

Configuring Smartcoin

1. The following quick example will create a new Smartcoin profile that uses the deepbit.net mining pool and the Phoenix miner.

   If you're not already in the Smartcoin interface, fire it up as root:

   $ sudo smartcoin
   

   You'll be presented with the Smartcoin main menu:

   Smartcoin r657s 08:17:46
   ----------------------------------------
   1) Reboot Computer
   2) Kill smartcoin (exit)
   3) Disconnect from smartcoin (leave running)
   4) Edit Settings
   5) Select Profile
   6) Configure Miners
   7) Configure Workers
   8) Configure Profiles
   9) Configure Devices
   10) Configure Pools
   11) Update Smartcoin
   12) Set Failover Order
   13) Configure Machines
   14) Configure Macros
   15) Execute Macro
   

2. First we'll configure the Phoenix miner. Type "6" and hit enter to configure your miner, then type "e" and hit enter to edit existing miners. Then enter "1" to edit the Phoenix miner:

   Smartcoin r657s 10:05:10
   ----------------------------------------
   SELECT MINER TO EDIT
   1) localhost
   Select the machine the miner resides on
   1
   1) phoenix
   2) poclbm
   Select the miner you wish to edit
   1
   

   From there, you should accept each pre-populated value. The only value I would recommend updating is the "Enter the miner's launch string", which I change to use the phatk2 kernel for increased performance:

   python <#path#>/phoenix.py -v -u http://<#user#>:<#pass#>@<#server#>:<#port#>/ device=<#device#> worksize=128 vectors aggression=11 bfi_int fastloop=false -k phatk2
   

   Enter "y" at the last prompt to set the Phoenix miner as the default miner:

   Smartcoin r657s 10:06:01
   ----------------------------------------
   EDITING MINER
   ------------
   1) localhost
   Please select the machine from the list above that is hosting this
    miner
   1
   Please give this miner a nickname
   phoenix
   Enter the miner's path (i.e. /home/you/miner/)
   /opt/miners/phoenix/
   Enter the miner's launch string
   Note:use special strings <#user#>, <#path#>, <#server#>
   <#port#>, and <#device#>
   python <#path#>/phoenix.py -v -u http://<#user#>:<#pass#>@<#server#>:<#port#>/ device=<#device#> worksize=128 vectors aggression=11 bfi_int fastloop=false -k phatk2
   Do you want this to be the the default miner for this machine?
   y
   

3. Next, we need to setup the workers associated with your mining pool. From the Smartcoin main menu, enter "7" to configure workers, and then enter "a" to add workers. Enter in the info for your worker, and enter "n" when prompted to add this worker to the automatic profile:

   ADDING WORKER
   -------------
   1) DeepBit
   2) Bitcoin.cz (slush)
   3) BTCGuild
   4) BTCMine
   5) Bitcoins.lc
   6) SwePool
   7) Continuum
   8) MineCo
   9) Eligius
   10) CoinMiner
   11) ZABitcoin
   12) BitClockers
   13) MtRed
   14) SimpleCoin
   15) Ozco
   16) EclipseMC
   17) BitP
   18) BitcoinPool
   19) EcoCoin
   20) BitLottoPool
   21) X8S
   22) Ars Technica
   23) TripleMining
   24) Mainframe
   25) Bitcoin Monkey
   26) Best Bitcoin
   27) Eclipse MC
   What pool listed above is this worker associated with?
   1
   
   Give this worker a nickname
   deepbit1
   
   Enter the username for this worker
   foobar
   
   Enter the password for this worker
   123456
   
   Would you like this worker to be available to the automatic profile? (y)es or (n)o?
   n
   

   Make sure to enter "n" when prompted to make the worker available to the automatic profile.

4. The last step is to configure a profile, which basically groups the previous settings and assigns a worker to each GPU.

   From the Smartcoin main menu, enter "8" to configure profiles, and then "a" to add a new profile, and enter a name for the new profile:

   Smartcoin r657s 10:41:54
   ----------------------------------------
   Would you like to (A)dd, (E)dit or (D)elete profiles?
   (X) to exit back to the main menu.
   a
   1) localhost
   Select the machine you wish to add a profile on
   1
   Enter a name for this profile
   deepbit
   

   This part is a little tedious: you'll have to add a new instance to the profile for each of your GPU(s):

   Smartcoin r657s 10:42:30
   ----------------------------------------
   Profile: deepbit (adding miner instance #1)
   
   1) phoenix
   2) poclbm
   Please select the miner from the list above to use with this instance
   1
   
   1) DeepBit.deepbit1
   Please select the pool worker from the list above to use with this instance
   1
   
   1) GPU[0]
   2) GPU[1]
   3) GPU[2]
   4) GPU[3]
   Please select the device from the list above to use with this instance
   1
   

5. Finally, you're ready to start mining! From the Smartcoin main menu, enter "5" to select your newly created profile:

   Smartcoin r657s 10:46:58
   ----------------------------------------
   1) localhost
   Select the machine from the list above that you wish to change the profile on
   1
   1) Donation
   2) Automatic
   3) Failover
   4) Idle
   5) deepbit
   Select the profile from the list above that you wish to switch to
   5
   

   Then hit CTRL + a and then 1 to view the mining status screen:

   Smartcoin r657s 10:50:23
   ----------------------------------------
   Host: localhost
   GPU[0]: Temp: 45.00 load: 99%
   GPU[1]: Temp: 45.50 load: 99%
   GPU[2]: Temp: 52.00 load: 99%
   GPU[3]: Temp: 54.00 load: 99%
   CPU Load Avgs: 0.70 0.33 0.17
   
   Profile: deepbit
   --------DeepBit--------
   GPU[0]: [340.05 MHash/sec] [6 Accepted] [0 Rejected] [0% Rejected]
   GPU[2]: [329.10 MHash/sec] [6 Accepted] [0 Rejected] [0% Rejected]
   GPU[1]: [340.41 MHash/sec] [3 Accepted] [0 Rejected] [0% Rejected]
   GPU[3]: [329.29 MHash/sec] [3 Accepted] [0 Rejected] [0% Rejected]
   Total : [1338.85 MHash/sec] [18 Accepted] [0 Rejected] [0% Rejected]
   
   Grand Total : [1338.85 MHash/sec] [18 Accepted] [0 Rejected] [0% Rejected]
   

   When you're done watching your stats, hit CTRL + a and then d to exit the Smartcoin interface. Your miners will continue to run in the background! Type smartcoin with root privileges to re-enter the Smartcoin interface.

Smartcoin uses the screen utility to display it's user interface. A few quick shortcuts for navigating the interface for a single mining rig (see here for more details):

• CTRL + a then 0: goto the Smartcoin main menu.
• CTRL + a then 1: goto the mining status screen.
• CTRL + a then 2: goto the miner log screen.
• CTRL + a then d: exit the Smartcoin interface.

We've only setup one mining rig in this tutorial, but keep in mind that Smartcoin can manage multiple mining rigs if you decide to scale operations. You may also choose to setup some failover profiles in the event that your primary mining pool is down.

Tweaks

The following is some optional system setup stuff. I recommend it to get additional features and increased performance on your rig. Legend:

• lines starting with 3 ###'s = a comment
• indented lines = stuff that goes in a configuration file
• all other lines = command to run from the console

### If you're in the US, you probably want to update the apt-get
### source.list. Replace:
### deb http://www.mirrorservice.org/sites/amd64.debian.net/debian/ sid main non-free contrib
### with
### deb http://ftp.us.debian.org/debian/ sid main non-free contrib
vi /etc/apt/sources.list
apt-get update

### the munin-node software isn't really used and
### had a bad habit of crushing my CPU at 100%,
### so disable it
sudo /etc/init.d/munin-node stop
sudo update-rc.d -f munin-node remove

### disable all munin crons, comment out their cron entries
sudo vi /etc/cron.d/munin
sudo vi /etc/cron.d/munin-node

### set date/timezone
sudo dpkg-reconfigure tzdata

### keep the system clock synchronized
sudo apt-get install ntp ntpdate
sudo update-rc.d -f ntp enable
sudo /etc/init.d/ntp start

# optional: disable tor
sudo update-rc.d -f tor disable

Maintenance

A few quick tips on rig upkeep:

• Try to keep your GPU core temps under 80° C. The highest temperature they can withstand without permanent damage is 89° C.
• Smartcoin stores it's settings in an SQLite database in the ~/.smartcoin directory. In this case, you can backup your smartcoin settings by copying the /root/.smartcoin directory to another location.
• Dust your rig every 2 weeks to improve performance and life expectancy. Get some canned air, take the rig outside, and blow that shit out of there. You'll be suprised how much dust can accumulate in a short period of time: more airflow = more dust.
• If your graphics card is locking up a lot, or experiencing consistently high temperatures, replace the thermal paste on it's GPU(s). Take the card off the rig, unscrew the back of it, and use a q-tip and 99% rubbing alchohol to clean off the old, crusty thermal paste from both the GPU(s) and heatsink. Apply a new layer of thermal paste (I use Artic Silver 5) on the top of the GPU(s), and screw it back together.

Sources

• Headless Linuxcoin
• Persistence Files
• bitcointalk.org topic #7374

favicon thingy

This is a quick guide on making a favicon in Photoshop CS4. A favicon is a 16x16px website-specific icon that was originally used to identify websites in a browser's site bookmark list. In most browsers today, it is displayed left adjacent to the url bar when visiting a site.

First, you need to install a plugin that enables Photoshop to save files in the .ico format. It's provided for free by Telegraphics, and can be downloaded here. Select the version that's specific to your OS and/or Photoshop version.

The plugin should be unzipped and saved to the "Adobe Photoshop CS4\Plug-ins\File Formats" directory. On Windows, the default location of this directory is:

C:\Program Files\Adobe\Adobe Photoshop CS4\Plug-ins\File Formats

Next, fire up photoshop and start designing the icon. It's best to start with a 64x64px or similar sized canvas, so that the icon can be proportionately scaled down to 16x16px.

You must also have the following enabled in order to save in .ico format:

Image > Mode > 8 Bits/Channel
Image > Mode > Indexed Color...

When you've finished the design, scale the image size down to 16x16px:

Image > Image Size...

• Width: 16px
• Height: 16px
• Constrain Proportions: checked
• Resample Image: Bicubic Sharper (best for reduction)

Then save as an .ico file:
File > Save As...

• Format: ICO (Windows Icon) (*.ICO)

If you want to setup the favicon for your website, save the .ico file to the document root, and add the following to your tags.

<link rel="shortcut icon" type="image/ico" href="/favicon.ico">

interfaces

This tutorial describes how to permanently add any number of static IP addresses to a network interface from the console.

Login as root and open /etc/network/interfaces in your favorite editor. By default, it should look something like this:

# The loopback network interface
auto lo
iface lo inet loopback

Let's assume your network interface is eth0. To assign it a static IP address, add the following lines to /etc/network/interfaces:

auto eth0
iface eth0 inet static
        address 192.168.0.41
        netmask 255.255.255.0
        broadcast 192.168.0.255
        gateway 192.168.0.1

You should of course replace the values for address, netmask, gateway, and broadcast with values specific to your desired IP address and network.

Additional static IP's can be added by repeating the above code in the interfaces file. Just increment the interfaces using "eth0:1", "eth0:2", etc, for each IP address. Example:

# The loopback network interface
auto lo
iface lo inet loopback

auto eth0
iface eth0 inet static
        address 192.168.0.41
        netmask 255.255.255.0
        broadcast 192.168.0.255
        gateway 192.168.0.1

auto eth0:1
iface eth0:1 inet static
        address 192.168.0.56
        netmask 255.255.255.0
        broadcast 192.168.0.255
        gateway 192.168.0.1

auto eth0:2
iface eth0:2 inet static
        address 192.168.0.89
        netmask 255.255.255.0
        broadcast 192.168.0.255
        gateway 192.168.0.1

If you're not using DHCP at all, you may need to add/edit your DNS servers. Open up /etc/resolv.conf and make sure the IP's of your DNS servers are in there. For example:

nameserver 192.168.0.1
nameserver 192.168.0.2

Finally, restart the network interfaces.

/etc/init.d/networking restart

If you're SSH'd into the box, don't log out yet. Log into the box with a new SSH session to verify the network interface came back up. Once verified, try pinging the new IP from a different box.

# ping 192.168.0.41
PING 192.168.0.41 (192.168.0.41) 56(84) bytes of data.
64 bytes from 192.168.0.41: icmp_seq=1 ttl=64 time=0.051 ms
64 bytes from 192.168.0.41: icmp_seq=2 ttl=64 time=0.061 ms
64 bytes from 192.168.0.41: icmp_seq=3 ttl=64 time=0.068 ms

Optional: remove dhcp client
It may be necessary to remove the dhcp client in order to keep your IP address static. It seems that by default, dhcp3-client is setup to automatically renew it's lease 1 per day. You can remove dhcp3-client like this:

apt-get remove dhcp3-client

ifcfg-eth0

Need to permanently assign a static IP address to a Linux box? This tutorial describes how to add any number of IP addresses to a network interface from the console.

Login as root and change directory to /etc/sysconfig/network-scripts and list the configurations of your network interface. In this example, we're assuming the network interface is eth0.

# cd /etc/sysconfig/network-scripts
# ls -l ifcfg-eth0*
-rw-r--r-- 1 root root 38 Dec  8  2009 ifcfg-eth0

Take note of the files listed. In the above example, eth0 only has one config file, but there could be many.

# cd /etc/sysconfig/network-scripts
# ls -l ifcfg-eth0*
-rw-r--r-- 5 root root 235 Sep 29  2009 ifcfg-eth0
-rw-r--r-- 1 root root 237 Sep 29  2009 ifcfg-eth0:1
-rw-r--r-- 1 root root 237 Sep 29  2009 ifcfg-eth0:2
-rw-r--r-- 1 root root 237 Sep 29  2009 ifcfg-eth0:3

Make a copy of the ifcfg-eth0 file and name the new file ifcfg-eth0:X where X is the next increment of the file names. For example, if there was only a single ifcfg-eth0 file, the new file should be named ifcfg-eth0:1. If the last config file was ifcfg-eth0:3, name the new file ifcfg-eth0:4.

cp ifcfg-eth0 ifcfg-eth0:1

Edit the new file and change/add the settings below. The DEVICE, NETMASK, IPADDR, GATEWAY settings must be set to your specific values. For example:

# DEVICE should be the network interface name and increment
DEVICE=eth0:1
IPADDR=192.168.0.102
NETMASK=255.255.255.0
GATEWAY=192.168.0.1
ONBOOT=yes
BOOTPROTO=none

Repeat these steps for each IP address you need to add. For each static IP address, create a new network interface config file named with the ":X" increment, and similarly set the DEVICE directive.

If you're not using DHCP at all, you may need to add/edit your DNS servers. Open up /etc/resolv.conf and make sure the IP's of your DNS servers are in there. For example:

nameserver 192.168.0.1
nameserver 192.168.0.2

Finally, restart the network interfaces.

/etc/init.d/network restart

If you're SSH'd into the box, don't log out yet. Log into the box with a new SSH session to verify the network interface came back up. Once verified, try pinging the new IP from a different box.

# ping 192.168.0.102
PING 192.168.0.102 (192.168.0.102) 56(84) bytes of data.
64 bytes from 192.168.0.102: icmp_seq=1 ttl=64 time=0.051 ms
64 bytes from 192.168.0.102: icmp_seq=2 ttl=64 time=0.061 ms
64 bytes from 192.168.0.102: icmp_seq=3 ttl=64 time=0.068 ms

iPhone on T-Mobile

Please note:
a reliable jailbreak for the iPhone 4 is not yet available, so this article currently applies only to the iPhone 3G/3GS.

A lot of folks out there are frustrated with AT&T's iPhone service. Its got a reputation for sucking. Recently detailed in Wired magazine, Apple has seriously considering ending its exclusive contract with the telecommunications giant. Apparently, AT&T's network hasn't been able to handle the new load of iPhone users, and upgrades to it's network aren't happening fast enough.

As an iPhone user, I never really experienced any problems with AT&T's service - until I moved to my new home. For some reason, my iPhone's reception dropped from 5 bars to none whenever I was inside or within 10 feet of my house. Other cell phones got great reception there - I can only presume that something nearby was causing interference on AT&T's operating frequencies.

After the 1000th dropped call and another $85 bill, it was time to dump AT&T.

Comparing AT&T to T-Mobile

The iPhone only works with the GSM network. AFAIK, T-Mobile is the only other major carrier in the United States with a GSM network and hence was the 1st choice for an alternative carrier. Plus, my friend's T-Mobile phone worked great in my home.

Here's how my original AT&T plan from July 2009 (Nation 450 Rollover & 5000 Night/Weekend) stacked up against T-Mobile's most similar plan (Individual Even More Plus 500 Talk + Text + Web).

T-Mobile was the clear winner - but its worth reiterating it's major drawbacks:

• No 3G, slower Edge data network only
• No visual voicemail

Wait... I don't have an iPhone yet. How should I get one?

No problem. If you're looking to get an iPhone without the AT&T plan, there are a few different things you can do.

• Newegg.com:
  • new unlocked iPhone 3GS 16GB: $769
• New/used on Ebay: Depending on the model:
  • used unlocked iPhone 3G goes for about $200
  • a used unlocked iPhone 3GS goes for about $500
  • a used iPhone 4 goes for about $800
• Signup for AT&T and cancel after 30 days: According to AT&T's latest return policy, you can cancel the service after 30 days and you can keep the phone. This comes at the cost of $325 (minus $10 for each additional month you have service). So a few examples of the total cost:
  • iPhone 3G 16GB (refurb): $99 phone + $60 for 1 month of cheapest service + $36 activation fee + $325 termination fee = $520
  • iPhone 3GS 16GB: $149 phone + $60 for 1 month of cheapest service + $36 activation fee + $325 termination fee = $570
  • iPhone 4 16GB: $199 phone + $60 for 1 month of cheapest service + $36 activation fee + $325 termination fee = $620
• Signup for AT&T and cancel immediately: You can avoid the $325 "Early Termination Fee" by canceling service within 30 days of the activation date. In this case, AT&T expects you to return the phone - but since you're not going to do that, they'll charge you for it's retail price. Oddly enough, Apple could only tell me the retail prices of the following models:
  • iPhone 3GS 8GB: $499 phone + $36 activation fee = $535 *
  • iPhone 4 16GB: $599 phone + $36 activation fee = $635 *

*Note: this does not include the pro-rated amount of your plan for X number of days before you cancel.

Uh... is this legal?

Yes. Your contract with AT&T was not signed in blood (I hope). Neither Apple nor AT&T can sue you for using a phone, that you've paid for, on a different network. And although you'll have to jailbreak and unlock your iPhone, jailbreaking is not against the law. The Library of Congress recently determined that jailbreaking mobile devices is exempt from the Digital Millennium Copyright Act, and hence is legal.

Disclaimer: I'm not a lawyer, so if you're still concerned, go talk to one.

OK enough yapping - let's do this thing.

Step 1 - Jailbreak and Unlock

Jailbreaking allows you to install 3rd party applications and modify your iPhone's software. Unlocking allows you to use your iPhone with carriers other than AT&T. It's necessary to jailbreak your iPhone first, and then unlock it.

There are currently a lot of ways to do this, and they all depend on your phone's OS and baseband version. For example, I used redsn0w to jailbreak and ultrasn0w to unlock my iPhone 3G on OS 3.1.2 on baseband 05.11.07.

The easiest way to figure out what to do is to use iClarified.com's online wizards:

• Jailbreak: http://www.iclarified.com/jailbreak
• Unlock: http://www.iClarified.com/unlock

If your device/OS is not listed in iClarified, visit www.jailbreakmatrix.com for the best software to jailbreak your phone with.

Most jailbreak software requires you to have the device's .ipsw firmware file. You can download any device firmware from http://www.felixbruns.de/iPod/firmware/.

Not to worry - your iPhone will still work with AT&T's service after you jailbreak and unlock.

Step 2 - Signup with T-Mobile

Head over to t-mobile.com and choose a plan. When prompted to purchase a phone, scroll through their models until you find the SIM card - it should be free.

I was unable to complete my purchase online, their web page blew up on the very last step. So I called them up and finished the order with a live rep. The rep had no problems when I mentioned I would be using an unlocked iPhone. The whole process took about 45 minutes due to the rep's "slow system".

Step 3 - Swap the SIM card

The SIM card will take about a week and a half to arrive unless you ordered expedited shipping. When you get it, do this:

1. Power off the iPhone. Grab a paper clip, stick it in the hole in the top of the iPhone, and pop out the AT&T SIM card.
2. Carefully push the T-Mobile SIM card out of it's "credit card" sized holder. Try not to touch/damage the side with the "grid" contacts on it. Place it in the little plastic iPhone SIM tray.
3. Bonus points: if you want to boost your reception a little, attach "a small piece of scotch tape to the outer side of the iPhone's SIM card (the side that does not have metal contacts)". This apparently adds to the pressure of the SIM card's contact enough to increase the phone's reception.
4. Slide the SIM card and tray back into the iPhone and power it up.

With any luck, you should see "T-Mobile" displayed as the carrier in the upper-left corner of your phone. Try a few test calls, text messages, etc. You can view your new phone number by dialing #686#.

Step 4 - Transfer your old number

Call up T-Mobile customer service and tell them you're a new customer and you want to transfer your old AT&T number to your phone. When I called, I was transferred twice, and the last rep (who barely spoke english) set up the transfer. The call took about 15 minutes. Within 2 minutes of hanging up, I received 2 text messages from "907" stating:

"Welcome to T-Mobile! To change your plan, pay your bill, or check your minutes go to my.t-mobile.com. Your mobile number is XXXXXXXXXX."

Step 5 - Verify Edge network

You may want to test the Edge data network connectivity and compare to the previous AT&T 3G service. Disable any wifi connection, open the Safari browser app, and try to browse. In my scenario, every time I tried to browse I was redirected to a T-Mobile branded web page (http://10.176.76.164/resources/webpages/error.jsp) that stated:

"Error Unknown

Your homepage is temporarily unavailable. Please try later."

I called T-Mobile again, and they noted that even though I had the unlimited data plan, they hadn't enabled a "Smart Phone Web" option on my account. The tech specified it may take 2 hours to be activated. About 1.5 hours later the Edge connection was working and I was able to browse freely.

Step 6 - Configure MMS (picture messaging)

If you'd like to send/receive picture messages, it's necessary to manually configure your iPhone's MMS settings.

On your iPhone, goto Settings > General > Network > Cellular Data Network. Enter the following T-Mobile MMS settings into the MMS section.

• APN: epc.tmobile.com
• MMSC: mms.msg.eng.t-mobile.com/mms/wapenc

Restart your iPhone and try sending a new message. There should now be a small camera icon next to the text input box that allows you to attach pictures.

Please note that MMS messaging seems to fail if you're connected to the internet via wifi - so disable any active wifi connections before sending a message.

If sending the text message still fails, check out the lengthy MMS discussion on the T-Mobile forum.

Step 7 - Cancel AT&T service

Once you've verified everything's working, drop that AT&T plan like it's hot. You'll have to call them up to do so, and of course, the rep will try to talk you into keeping your AT&T account. Be assertive and continue to repeat that you just want to cancel your account. Don't provide any reasons if you don't want to. If you're getting pegged with an "Early Termination Fee", it will show up on your final bill. I had signed up a year ago when the fee was only $175, so my termination fee was $115 = ($175 - (12 months x $5)).

Optional final step - Visual Voicemail

T-Mobile's visual voicemail doesn't yet work with the iPhone. You'll still get notifications and Phone app's "Voicemail" button will dial to your T-Mobile voicemail.

If you really want visual voicemail, as well as a bunch of other voicemail features, try out the YouMail app. Their service is free and offers voice-to-text transcriptions of your voicemails. The app will automatically switch your voicemail over to their service, and you can cancel anytime.

Booyah, Grandma. Booyah.

Congrats - you're officially a jailbreaking, phone unlocking, freedom loving, renegade iPhone ninja. If William Wallace was here, he'd tell you:

"They may take our money, but they will never take OUR FREEDOM!"

Deluxe iPhone Pencil Stand
3 pics

If you've made it here, chances are you're a DIY-er. You hate spending money and you'd rather build stuff with your own hands. You chuckle at rich suckers who shell out $350 for a pure crystal iPhone stand. And you'd get a much more rewarding experience making your own stand using office supplies stolen from your employer.

Disclaimer: we didn't come up with the original idea for this stand. This "deluxe" model offers a few modifications of Geeky Gadget's Pencil iPhone Stand.

Features

• Uses two pencils as rear support for more sturdiness
• Front and back paper clip "guard rails" prevents the iPhone from slipping off
• Holds the iPhone in either landscape or portrait orientation

What You'll Need

• 8 rubber bands
• 2 paper clips
• 6 pencils (unsharpened for safety or sharpened for bad-assed-ness)
• Needle nose pliers (unless you're a tough guy)

Supplies

Guard Rails

We'll start by straightening out the paper clips to create the stand's "guard rails".

1. Grab a paper clip, and unbend the first 3 folds till you have a "U" shape (or upside down "U").
2. Use the pliers to get rid of that kink in the middle.
3. Optional: Notice one side of the "U" is longer than the other? You can try to make them even, but we'll be able to hide this difference in the rubber bands later.
4. Using the pliers, make a new, 90 degree outward bend in each side of the "U" about a half inch from the previous bend. This should create a more "3 dimensional" shape.
5. Repeat for the other clip. The more even and identical they are, the better the end result will look.

The Cradle

Two pencils and those bent paper clips are going to hold the cellular telephone.

1. Place 2 pencils side-by-side and loop and tighten a rubber band around each end. Make sure the distance between the rubber bands is the same as the width of the paper clips from the previous step.
2. Grab a guard rail and thread each underneath a rubber band on the conjoined pencils. The guard rail leg doesn't need to go underneath all of the loops, just one loop is fine.
3. Grab the other guard rail and thread it through the rubber bands on the other side of the pencils.
4. Try to make the paper clips as straight and even as possible. It's alright if they feel loose, we'll tighten them down in the next step.

The Frame

In this step, we'll make a frame for the phone to lean on and attach it to the cradle thing.

1. Cross 2 pencils at their pointy ends and wrap a rubber band around them. The rubber band should be wrapped about 1 inch from the end of the pencils.
2. Thread each leg of these two pencils through the rubber bands of the cradle. Again, just getting each leg under a single loop is fine.
3. Now its time to tighten this thing down. Grab a rubber band and loop it around the joint you just made. Make it tight. If you have some thicker rubber bands, now is the time to use them. If not, you can repeat this step with extra rubber bands for more stability.
4. If any paper clip ends are sticking out, try to move a rubber band loop over them. Those things can be pretty sharp.

Complete the Stand

Let's finish this thing off.

1. Grab the last 2 pencils and loop a rubber band around them just below the metal of their erasers. This is going to be the 3rd leg of the stand.
2. Spread the 3rd leg's pencils apart and thread the pointy ends through the rubber band on each side of the top of the frame.
3. Grab a rubber band and tightly wrap it around the new joint. Repeat with another rubber band. If it's still not stable, wrap a 3rd rubber band around the joint.

Done

With any luck, you should now have something that looks like this.

The finished product

If any of the joints feel loose or unstable, loop another rubber band band around it. The thicker the rubber band, the more stability you'll get.

yum fun

Every once in a while you may need to install a specific version of a package on your system. If you're using the yum package manager, it isn't blatantly obvious how to do this, even after pouring over the man pages.

So let's do this thing.

First, if you've already installed a newer version of the package, you'll have to downgrade it. In that case, make sure you've got yum's downgrade plugin installed:

# yum install yum-allowdowngrade

Second, do you know exactly which package version you want to install? If not, list the all available versions using the --showduplicates flag. This example looks for available versions of the GD extension for PHP.

# yum --showduplicates list php-gd

Available Packages
php-gd.i386                    5.2.3-3.el5s2                        testing
php-gd.i386                    5.2.6-2.el5s2                        testing
php-gd.i386                    5.2.9-2.el5.centos                   testing
php-gd.i386                    5.2.10-1.el5.centos                  testing

Yum uses the RPM package manager, and it's package names are based on the RPM package label format. In the above example, the first column displays the package name and architecture in the format NAME.ARCHITECTURE. The middle column displays the version and release info in the format VERSION-RELEASE. Don't ask me why yum rearranges the RPM package label format.

Anyways, if you're downgrading, remove the previous package:

# yum remove php-gd

Now that you know which version to install, it's a matter of specifying the package name with the version. Use the format NAME-VERSION.

# yum install php-gd-5.2.6

Or if you want to get even more specific, use NAME-VERSION-RELEASE (which might be easier to copy/paste since VERSION-RELEASE is the second column).

# yum install php-gd-5.2.6-2.el5s2

If you're downgrading, you may need to throw in the --allow-downgrade flag.

# yum --allow-downgrade install php-gd-5.2.6

And you should be good to go. If you want to lock the version you installed to prevent any future updates, install the versionlock plugin.

# yum install yum-versionlock

Then lock it.

# yum versionlock php-gd

Ping a Windows box

Platforms:
any *nix distro

What You'll Need:
Samba

In a heterogeneous LAN it is often useful to resolve network addresses by a computer's name (ie. netbios name). This is especially true if the LAN does not have a DNS server so that host names can be used instead of IP addresses (which if dynamically assigned, could change often).

To enable Windows netbios name resolution from a Linux computer, make sure that Samba is installed (although the smb service does not need to be running). The Samba suite includes winbind, which enables Windows host names to be resolved.

Then edit /etc/nsswitch.conf and change this line:

hosts:      files dns

to this:

hosts:      files dns wins

Then test by pinging the computer name of Windows machine on the LAN:

$ ping windowsbox
PING windowsbox (192.168.0.100) 56(84) bytes of data.
64 bytes from 192.168.0.100: icmp_seq=1 ttl=128 time=0.117 ms
64 bytes from 192.168.0.100: icmp_seq=2 ttl=128 time=0.127 ms
64 bytes from 192.168.0.100: icmp_seq=3 ttl=128 time=0.127 ms
64 bytes from 192.168.0.100: icmp_seq=4 ttl=128 time=0.127 ms
64 bytes from 192.168.0.100: icmp_seq=5 ttl=128 time=0.128 ms

This setting really comes in handy when mounting a shared folder of a dynamically IP'ed Windows box from Linux. Instead of using the Windows' box IP address, just specify it's netbios name. Example entry in /etc/fstab:

//windowsbox/my_share/ /mnt/my_mount_point/ cifs rw,username=xxx,password=xxx,domain=xxx 0 0

javascript array

AJAX communication between the client browser and web server is a handy method of transferring data without forcing the user to refresh the page. Typically, the data sent to the server is a simple set of form fields or flags. However, it is also possible to submit more complex structured data. This tutorial will demonstrate 4 different methods used to pass multidimensional javascript arrays using jQuery and AJAX to a PHP web server.

UPDATE: The Easy Way... for jQuery 1.4.0+

Thanks to Pirumpi for pointing out that jQuery 1.4.0 now automatically serializes arrays and objects parameters for AJAX functions. That means it's super easy to pass them to PHP:

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" dir="ltr" lang="en-US">
    <head profile="http://gmpg.org/xfn/11">
        <meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />
        <title>jQuery AJAX arrays</title>
        <script type="text/javascript" src="//ajax.googleapis.com/ajax/libs/jquery/1.4.0/jquery.js"></script>
        <script type="text/javascript">
                sendAjax = function(){
                    var data = {
                                  foo:  123,
                                  bar:  456,
                                  rows: [
                                          {
                                            column1 : 'hello',
                                            column2 : 'hola',
                                            column3 : 'bonjour',
                                          },
                                          {
                                            column1 : 'goodbye',
                                            column2 : 'hasta luego',
                                            column3 : 'au revoir',
                                          },
                                        ],
                                  test1:{
                                          test2: {
                                                   test3:  'baz'
                                                 }
                                        }
                                };

                    $.ajax({
                            type:           'post',
                            cache:          false,
                            url:            './ajax/',
                            data:           data
                           });
                }

        </script>
    </head>
    <body>
        <input type="button" style="width: 130px; height: 60px" value="send AJAX" onclick="sendAjax();" />
    </body>
</html>

When the button is clicked, the following structured data shows up in PHP's $_POST variable:

Array
(
    [foo] => 123
    [bar] => 456
    [rows] => Array
        (
            [0] => Array
                (
                    [column1] => hello
                    [column2] => hola
                    [column3] => bonjour
                )

            [1] => Array
                (
                    [column1] => goodbye
                    [column2] => hasta luego
                    [column3] => au revoir
                )

        )

    [test1] => Array
        (
            [test2] => Array
                (
                    [test3] => baz
                )

        )

)

This will only work with jQuery 1.4.0+. Otherwise jQuery simply calls .toString() on the nested array at key "rows" and nested object at key "test1", and they get passed to PHP with the useless values "[object Object]".

NOTE: The remainder of this article will demonstrate how to send objects and arrays for jQuery versions lower than 1.4.0.

Method 1 - Form input arrays

It's possible to name the keys of the JS array as HTML form input arrays (ie. named with square brackets) so that PHP decodes the array's key names as nested arrays. I don't recommend this method because it does not maintain the JS array's original structure in the Javascript. However, it can be used if you just need to get the AJAX request working without using JSON.

The original JS "data" array can be renamed/restructured as:

sendAjax = function(){
	var data = {
				  foo:  123,
				  bar:  456,
				  'rows[0][column1]':    'hello',
				  'rows[0][column2]':    'hola',
				  'rows[0][column3]':    'bonjour',
				  'rows[1][column1]':    'goodbye',
				  'rows[1][column2]':    'hasta luego',
				  'rows[1][column3]':    'au revoir',
				  'test1[test2][test3]': 'baz'
				};
	$.ajax({
			type:           'post',
			cache:          false,
			url:            './ajax/',
			data:           data
		   });
}

And the data is correctly received in PHP's $_POST variable:

Array
(
    [foo] => 123
    [bar] => 456
    [rows] => Array
        (
            [0] => Array
                (
                    [column1] => hello
                    [column2] => hola
                    [column3] => bonjour
                )

            [1] => Array
                (
                    [column1] => goodbye
                    [column2] => hasta luego
                    [column3] => au revoir
                )
        )

    [test1] => Array
        (
            [test2] => Array
                (
                    [test3] => baz
                )
        )
)

Method 2 - JSON string parameter

This method converts the JS array to a JSON string, and passes it as a POST or GET parameter, which can be manually decoded back into an array by PHP. Modern browsers (Firefox 3, IE 8) now have a native JSON object that can be used to stringify JS arrays to JSON strings. The following example uses the YUI JSON utility to convert the JS array to a string ( json2 is another populary lightweight utility for handling JSON).

sendAjax = function(){
	var data = {
				  foo:  123,
				  bar:  456,
				  rows: [
						  {
							column1 : 'hello',
							column2 : 'hola',
							column3 : 'bonjour',
						  },
						  {
							column1 : 'goodbye',
							column2 : 'hasta luego',
							column3 : 'au revoir',
						  },
						],
				  test1:{
						  test2: {
								   test3:  'baz'
								 }
						}
				};
	data = YAHOO.lang.JSON.stringify(data);

	$.ajax({
			type:           'post',
			cache:          false,
			url:            './ajax/',
			data:           {myJson:  data}
		   });
}

On the PHP side, it's necessary to use json_decode() to convert the JSON $_POST parameter string (named "myJson") to an array:

<?php
$data = json_decode($_POST['myJson'], true);
print_r($data);

And the array is good to go:

Array
(
    [foo] => 123
    [bar] => 456
    [rows] => Array
        (
            [0] => Array
                (
                    [column1] => hello
                    [column2] => hola
                    [column3] => bonjour
                )

            [1] => Array
                (
                    [column1] => goodbye
                    [column2] => hasta luego
                    [column3] => au revoir
                )
        )

    [test1] => Array
        (
            [test2] => Array
                (
                    [test3] => baz
                )
        )
)

Method 3 - JSON content-type

My favorite method is to simply JSON stringify the JS array, and pass it as the AJAX request's body. This requires that the content-type of the request be changed from the default "application/x-www-form-urlencoded" to "application/json", and that $.ajax()'s processData parameter be set to false so that the JSON data is not converted into a query string. Also, the type of request must be POST, as the GET method does not support submitting data in the request body.

sendAjax = function(){
	var data = {
				  foo:  123,
				  bar:  456,
				  rows: [
						  {
							column1 : 'hello',
							column2 : 'hola',
							column3 : 'bonjour',
						  },
						  {
							column1 : 'goodbye',
							column2 : 'hasta luego',
							column3 : 'au revoir',
						  },
						],
				  test1:{
						  test2: {
								   test3:  'baz'
								 }
						}
				};
	data = YAHOO.lang.JSON.stringify(data);

	$.ajax({
			type:           'post',
			cache:          false,
			url:            './ajax/',
			data:           data,
			processData:    false,
			contentType:   'application/json'
		   });
}

On the PHP side, it's necessary to manually read the JSON string directly from the request body using the php://input stream. The returned string can then be decoded into an array:

<?php
$data = file_get_contents('php://input');
$data = json_decode($data, true);
print_r($data);

Which gives us the array:

Array
(
    [foo] => 123
    [bar] => 456
    [rows] => Array
        (
            [0] => Array
                (
                    [column1] => hello
                    [column2] => hola
                    [column3] => bonjour
                )

            [1] => Array
                (
                    [column1] => goodbye
                    [column2] => hasta luego
                    [column3] => au revoir
                )
        )

    [test1] => Array
        (
            [test2] => Array
                (
                    [test3] => baz
                )
        )
)

Hopefully this tutorial has helped you get multidimensional array data passed to your webserver using AJAX. If you know of another method, please submit it in the comments - thanks!

PuTTY

By default, your keyboard's numlock enabled number pad doesn't work well with console applications running on a PuTTY SSH connection. The numpad's numbers seem to get mapped to inserting the characters "q" thru "y". This is especially annoying when using vim.

To get numlock working, open PuTTY's configuration window, goto Terminal > Features. Enable the "Disable application keypad mode" checkbox.

If you want this setting to persist, make sure to select a saved session and click "Save".