Setting up NAT and MASQUERADE for sharing USB Tether connection over LAN

I the only source of Internet connection I have currently is my phone. I wanted to share this network with other systems, via a lan/wireless router. So here is a basic setup:

• Android Phone with USB Tethering
• Laptop ( Fedora 18 )
• Wireless Router with Lan
• A desktop machine ( Fedora 17 )

Setting up Internet gateway: Laptop + Android Phone

I connected Android Phone via USB cable to the laptop and enabled USB thether, so I have a usb device /dev/usb0. This was automatically detected by Network Manager. Also I connected Laptop to the router via a LAN.

I setup the NAT on my Laptop using this script as follows:

# cat nat.sh
IFOUT=usb0
IFLAN=em1
echo 1 > /proc/sys/net/ipv4/ip_forward &&
sudo iptables -t nat -A POSTROUTING -o $IFOUT -j MASQUERADE &&
sudo iptables -A FORWARD -i $IFOUT -o $IFLAN -m conntrack --ctstate RELATED,ESTABLISHED -j ACCEPT &&
sudo iptables -A FORWARD -i $IFLAN -o $IFOUT -j ACCEPT

So all I had to do to setup is do this:

# sh nat.sh

And IPTABLES take care of masquerading the IP addresses to-and-fro from local network to the internet. I received IP address 192.168.1.2 for the Laptop, which becomes a default gateway for rest of the machines connected to the router.

Setting up Internet user: Desktop machine

Now on the desktop machine ( with IP address 192.168.1.3 ), I only had to setup the routes and DNS server:

# route add default gw 192.168.1.2
# route -n
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
0.0.0.0         192.168.1.2     0.0.0.0         UG    0      0        0 em1
0.0.0.0         192.168.1.1     0.0.0.0         UG    0      0        0 em1
192.168.1.0     0.0.0.0         255.255.255.0   U     0      0        0 em1

# cat /etc/resolv.conf 
# Generated by NetworkManager
#nameserver 192.168.1.3
nameserver 208.67.222.222
nameserver 208.67.220.220
nameserver 192.168.42.129

Thats all, that was needed. I had done the NAT + MASQUERADE setup earlier, however it was on Fedora Core 2 and some commands didn’t match up, esp. how to use conntrack module.

References:

• NAT (Internet connection sharing) switching between multiple public interfaces
• Old NAT script

Understanding big Lucene index by inspecting a portion of it

I was wondering if I could get a sample out of many huge Lucene indexes and inspect them with Lukeall on my machine. I quickly realized, that copying such indexes over network would be time consuming.

First I googled for a ready-made solution so that I could copy on a few documents from the whole index into a separate ( small ) index. That way I could quickly understand the document structure. I came across this blog which mentions only how to backup a Lucene index. My use-case is to get only a portion of it. However, it also mentions how to use Lukeall to export an index ( or a portion of it ) in XML format. That seemed to be in right direction. But, there is no way I could import it back into a Lucene index. Well, that is was a stumbling block. There is a defect open in Lukeall for precisely this feature here.

For my purpose I have created a Scala script to copy first few documents of an index into an output index. The script is located here for now.  So, now I can get the first few documents from an index on a remote machine and inspect them locally.

Are there any better ways ?

On choosing Play2 framework

I was wondering whether or not to learn Play2 Framework. I already know Ruby On Rails. The main criteria right now is to run a web-application that will use lots of code already written in Java.

Which one of Rails/JRuby or Play2 would be better when deploying on JVM? That is my main question. So far I have stumbled upon similar questions as below:

• Play vs JRuby on Rails ??
• advantages of play framework for people coming from php / ruby / python

There are two criteria that I considered for selecting Play2:

• Learning curve
• Performance

For learning curve, definitely it would take more time for me to learn Play2 than just using Rails on JRuby. However, since Play2 is quite similar to Ruby On Rails, it would be fairly easy to map the concepts. Play2 provides type-safety, so some bugs will appear earlier at the time of development itself. I can choose either Scala/Java for development and also I can select different templates for HTML or JavaScript etc. For performance, I found something very interesting here.

• Performance Comparison: Rails, Wicket, Grails, Play, Tapestry, Lift, JSP, Context

I am getting a feeling that Play framework will perform significantly better that Rails/JRuby. Well, that remains to be seen for now.

I hope Play2 would be nice to learn.

Emacs and Akka Pi example

I was trying The first Akka Example here

Install the TypeSafe stack on RPM distro:

$ sudo yum install typesafe-stack

Setup a g8 project:

$ g8 typesafehub/akka-scala-sbt
Akka 2.0 Project Using Scala and sbt 

organization [org.example]: in.tuxdna
name [Akka Project In Scala]: akka-pi
akka_version [2.0.1]: 
version [0.1-SNAPSHOT]: 

The I added the Pi.scala to src/main/scala folder. Then I tried to compile it from within Emacs, to check if it will be easier navigate compilation errors in the code. I got wierd output. Emacs compilation buffer was printing escape characters.

For this I found two solutions:
* Disable the colors generated by SBT in the first place: http://www.scala-lang.org/node/2198

 sbt compile -Dsbt.log.noformat=true

* OR Enable ANSI colors in Emacs compilation buffer: http://stackoverflow.com/questions/13397737/ansi-coloring-in-compilation-mode

 (require 'ansi-color)
(defun colorize-compilation-buffer ()
  (toggle-read-only)
  (ansi-color-apply-on-region (point-min) (point-max))
  (toggle-read-only))
(add-hook 'compilation-filter-hook 'colorize-compilation-buffer)

And here is the final output after I had resolved all the compilation errors. ( because I didn’t copy/paste all the code, I typed it out ):

Enter number: saleem@impetus-D853:~/Downloads/akka-pi$ sbt run 
[info] Loading project definition from /home/saleem/Downloads/akka-pi/project
[info] Set current project to akka-pi (in build file:/home/saleem/Downloads/akka-pi/)

Multiple main classes detected, select one to run:

 [1] in.tuxdna.Akkapi
 [2] Pi

Enter number: 2

[info] Running Pi 

	Pi approximation: 		3.1415926435897874
	Calculation time: 	991.0
[success] Total time: 4 s, completed 15 Mar, 2013 8:04:33 PM

Value of PI : 3.1415926435897874

Attending Jamia Developer Day

Students at JMI University had organized JDevDay ( formerly CONF@JMI). This was a two day conference on FOSS topics. The agenda is listed here:

https://jdevday.wordpress.com/2013/03/05/talks-in-jdevday/

I spoke on “Introduction to Scala”, the slides of which are uploaded to slideshare and on github.

It was so nice to meet all the speakers and attendees. I took a few photos which I have uploaded on flickr.

JDevDay Rocks!

Emacs Org Mode for creating and generating presenations

Last time I created a presentation in Org Mode was in 2010 here.

Creating presentation in Emacs Org Mode and publishing it as Latex, PDF and HTML, is so awesome.

Here are the links to get you started:

• http://orgmode.org/worg/exporters/beamer/tutorial.html

• http://orgmode.org/worg/org-tutorials/non-beamer-presentations.html

And follow these simple steps ( I am using Ubuntu right now but the steps should be easy on Fedora too.):

Create a slides.org file as in the tutorial above:

Install Beamer class for Latex.
$ sudo aptitude install aptitude latex-beamer

And export it to PDF using “Ctrl+c Ctrl+e p” in Emacs.

Your presentation is in slides.pdf in the same folder as slides.org.

Where the heck did agave go from Fedora 17 repos? Yes I am asking the same question already been asked in fedora forum [1].

Apparently it doesn’t [2] have any maintainer hence no build for Fedora 17.

[1] http://forums.fedoraforum.org/showthread.php?t=276162
[2] https://admin.fedoraproject.org/pkgdb/acls/name/agave
[3] http://home.gna.org/colorscheme/

SBT and Specs for BDD in Scala

I had used RSpec earlier for Behaviour Development Development for a Ruby on Rails project. Today I learnt how to do BDD in Scala.

Chapter 4 of Programming Scala introduces Traits and Specs for testing the code. Here is how I set it up:

I already had Scala installed. So the first step was to setup sbt ( Simple Build Tool for Scala ). Setup was easy ( described in detail here ):

$ sudo yum localinstall http://scalasbt.artifactoryonline.com/scalasbt/sbt-native-packages/org/scala-sbt/sbt//0.12.2/sbt.rpm

Then, I created a sample project which is hosted on github ( https://github.com/tuxdna/trait-with-sbt-spec ):

$ git clone git://github.com/tuxdna/trait-with-sbt-spec.git
$ cd trait-with-sbt-spec
$ sbt test

And here is the output

...
[info] Compiling 9 Scala sources to ...
[info] Compiling 4 Scala sources to ...
[info] ButtonClickableObserverSpec
[info] A Button Observer should
[info]   + observer button clicks
[info] ButtonObserverSpec
[info] A Button Observer should
[info]   + observe button clicks
[info] ButtonClickableObserverVetoableSpec
[info] A Button Observer with Vetoable Clicks should
[info]   + observe only the first button click
[info] ButtonObserverAnonSpec
[info] A Button Observer should
[info]   + observe button clicks
[info] Passed: : Total 4, Failed 0, Errors 0, Passed 4, Skipped 0
[success] Total time: 19 s, completed Feb 21, 2013 9:04:57 PM

Setting up SBT was easy, only that it took some time to fetch the libraries from maven repos. Its configuration is very similar to Maven POM files. And then adding specification was only a matter of extending ‘Specification’ class.

References:
[1] http://examples.oreilly.com/9780596155964/README.html
[2] http://www.scala-sbt.org/release/docs/Getting-Started/Setup
[3] http://code.google.com/p/specs/wiki/QuickStart
[4] http://www.scala-sbt.org/release/docs/Getting-Started/Library-Dependencies.html
[5] http://twitter.github.com/scala_school/sbt.html

Apache Mesos – incubator project; and Spark Project

Apache Mesos, written primarily in C++ is a cluster resource allocation framework. It is used by the Spark Project ( majority of Spark project is written in Scala ).

Apache Mesos – http://incubator.apache.org/mesos/
Spark Project – http://spark-project.org/
Tech Talk: Matei Zaharia (UC Berkeley) — “Spark: A Framework for Iterative and Interactive Cluster Computing” – http://vimeo.com/20757432

Scala script tool seems to be broken Fedora 17

I am running Scala on Fedora 17 and I seem to face a problem. I created a /tmp/hello.scala file, as follows:

println("Hello world")

This works just fine if I run it on Ubuntu 12.04 i.e. it prints `Hello world`. However, when I try to run it on Fedora, I don’t get any output and the script keeps waiting forever.

So I tried to inspect what’s happening here:

$ bash -x scala /tmp/hello.scala
+ scala_exit_status=127
+ saved_stty=
+ trap onExit INT
++ stty -g
+ saved_stty=6d02:5:4bf:8a3b:3:1c:7f:15:4:0:1:ff:11:13:1a:ff:12:f:17:16:ff:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0
+ [[ ! -n 0 ]]
+ [[ -n '' ]]
+ cygwin=false
+ case "`uname`" in
++ uname

... OUTUPT SKIPPED ...

+ declare -a scala_args
+ CPSELECT=-Xbootclasspath/a:
+ '[' 1 -gt 0 ']'
+ case "$1" in
+ scala_args=("${scala_args[@]}" "$1")
+ shift
+ '[' 0 -gt 0 ']'
+ set -- /tmp/2.scala
+ exec /usr/lib/jvm/java//bin/java -Xmx256M -Xms32M -Xbootclasspath/a:/usr/share/scala/lib/jansi.jar:/usr/share/scala/lib/jline2.jar:/usr/share/scala/lib/scala-compiler.jar:/usr/share/scala/lib/scala-dbc.jar:/usr/share/scala/lib/scala-library.jar:/usr/share/scala/lib/scala-partest.jar:/usr/share/scala/lib/scalap.jar:/usr/share/scala/lib/scala-swing.jar -Dscala.usejavacp=true -Dscala.home=/usr/share/scala -Denv.emacs= scala.tools.nsc.MainGenericRunner /tmp/2.scala

It keeps waiting on the last line for ever.

However, if I add -nc option to scala command, it works fine:

$ scala -nc /tmp/hello.scala
Hello world

So I guess, the issue has something to do with the compilation daemon as below

-nc no compilation daemon: do not use the fsc offline compiler