Perfectly Reasonable Deviations

Server Upgrade

2011-01-17T09:03:00.000-08:00

We were using a 10 year old Dell desktop as our router/webserver/misc computer.

Our needs evolved to require running RT and Asterisk on the system. Since they're considerably higher weight applications, I decided to upgrade the hardware.

After getting my new server, I needed to clone the linux installation. Since the old server had an IDE hard drive, and the new one was SATA, I purchased a $20 USB-SATA/IDE adapter from Amazon to help with the transition.

Step 1 - Clone Drive

I used a livecd version of gparted on my new system to clone the old drive. I used the clonezilla tool that was included to do a raw copy. The result was a new installation that was literally identical to my old one

Step 2 - Resize Partitions

Since my old server was from the dawn of the centry, the hard drive was only 60G. the new machine has a 250G drive (of course, for a fraction of the price). By doing the raw copy in step #1, the hard drive looks exactly like my old one (from a software point of view), and has a single 60G partition.

I used the gparted liveCD to resize my main partition. While I'd like to simply resize /dev/sda1, the swap partition gets in the way. As such, I needed to delete the swap partition, resize my main partition, then re-add my swap partition. To do this correctly, I had to edit /etc/fstab and update the UUID of /etc/sda5(my swap partition) with the UUID of the new partition I created

Step 3 - Fix stuff that broke

Given all the hardware in the system magically changed in an instant (from the view of the linux installation) I was quite impressed that it all booted at all. The main issue i had was getting my network interfaces to work.

Linux keeps track of network adapters by MAC address. I had to edit /etc/udev/rules.d/70-persistent-net.rules to update the network interfaces with the new cards and remove the old ones (from the previous system).

After that little update, everything really just worked! It was quite impressive! I finally did a distribution upgrade that broke some perl modules and apache configuration bits, but all in all, not bad.

Forwarding XML-RPC

2010-12-15T20:38:00.001-08:00

I've always struggled with GUIs for embedded systems.

Our robots generally run linux. Generally, the systems run w/out monitors or keyboards, so the best option is to ssh into a linux box, with X11 forwarding turned on, and pretend like I was at the terminal. This is great as long as you have a laptop. The question is how I could do this in a more platform agnostic way. It'd be great if I had one GUI for windows/linux/mac/ios/etc, and if that GUI required no software installation.

Since GUI programming isn't my forte, and my GUIs literally just call my API functions with a lot of extra code for layout, this seemed like a great opportunity to fuse web and embedded software. Since we had used xmlrpc before, it seems to work OK, and there were javascript libraries for it, I decided to try writing my gui in a webpage with javascript, and write a xmlrpc wrapper for my robot api.

The first issue I ran into was a security limitation on the XMLHTTPost function in javascript. You can only post to the server the script came from. That extends to the port on the server the script came from. Since xmlrpc runs over HTTP, it has to be a different port than my GUI webpage (served by apache).

This was a pain, but I got around it by writing what is effectively a xmlrpc forwarder.

I created a php file with the following content:

1:  // Make Sure that it's a POST request that we're forwarding
2:  if ( $_SERVER['REQUEST_METHOD'] === 'POST' )
3:  {
4:       // Read the input from stdin
5:       $data = trim(file_get_contents('php://input'));
6:  }
7:  // initialize curl
8:  $ch=curl_init();
9:  // The URL of our actual server
10:  $url="http://localhost/RPC2";
11:  // The XML payload is the data that was posted, with the XML tag
12:  define('XML_PAYLOAD', '<?xml version="1.0"?>'."\n".$data);
13:  // Set all the options for cURL
14:  curl_setopt($ch,CURLOPT_URL,$url);
15:  // Running on port 8080
16:  curl_setopt($ch,CURLOPT_PORT,8080);
17:  // Get request is default, make this a POST
18:  curl_setopt($ch,CURLOPT_POST,1);
19:  curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1); 
20:  curl_setopt($ch, CURLOPT_TIMEOUT, 4); 
21:  curl_setopt($ch, CURLOPT_POSTFIELDS, XML_PAYLOAD);
22:  // This has to be set for it to be an XML post
23:  curl_setopt($ch, CURLOPT_HTTPHEADER, array("Content-Type: text/xml"));
24:  $myFile = "/tmp/xml_forward_debug.txt";
25:  $fh = fopen($myFile, 'w') or die("can't open file");
26:  fwrite($fh, XML_PAYLOAD);
27:  fwrite($fh, "\n\n*** Response\n\n");
28:  $ret=curl_exec($ch);
29:  fwrite($fh, $ret);
30:  fclose($fh);
31:  header('Content-Type: text/xml');
32:  print $ret;

My javascript uses mimic.

My C++ server uses xmlrpc for c++

My test web post calls the classic "add" method

1:       var request = new XmlRpcRequest("index.php", "sample.add"); 
2:       request.addParam(1); 
3:       request.addParam(3); 
4:       var response = request.send(); 
5:       alert(response.parseXML());

This all seems to work so far! Very exciting. Still quite a ways to go as I work out messaging, etc., but I'm hoping this was the biggest step. Will post more as the project progresses.

Displaying The IP Address of my linux computer

2010-12-03T06:35:00.000-08:00

I'm building a robot that has a computer monitor as it's head (literally), but no keyboard or mouse to support a UI. I'd like to connect to it by using the IP address for general operation, but I need to know what it is. What to do?

I've seen many posts about people looking for how to display the ip address on the desktop. My case is similar. My solution was to write a shell script that was added to the startup programs (under Settings >> Sessions & Startup) that looks like this:

#/bin/bash

w=$(ifconfig eth0 | grep "inet addr" | gawk -F: '{print $2}' | gawk '{print $1}')
wl=$(ifconfig wlan0 | grep "inet addr" | gawk -F: '{print $2}' | gawk '{print $1}')
msg=$(echo "Wired :" $w "\n" "Wireless :" $wl)

zenity --info --text="$msg" &

On startup, a message box is displayed that indicates the IP address.

wxWidgets and Failed Assertions

2009-10-22T07:25:00.000-07:00

I was writing my first GUI in C++ on linux using wxwidgets and would erratically get this message:

./../src/xcb_lock.c:33: _XCBUnlockDisplay: Assertion `xcb_get_request_sent(dpy->xcb->connection) == dpy->request' failed

it did take a couple different forms (i think there was one other), but usually had a dpy->request in it and xcb_lock.c), but this one came up quite often. it was very erratic, and would sometimes keep my gui from coming up at all.

I ultimately (thanks to the help of people in #wxWidgets on freenode) figured out it was because wxWidgets is not thread safe and my worker thread was calling member functions of my GUI.

To fix this I added wxMutexGuiEnter() and wxMutexGuiExit() around my code that access the gui and all issues were resolved.

Getting mail from gmail using fetchmail

2009-06-29T18:54:00.000-07:00

So I wanted to setup RT on my ubuntu box. It's been rough so far, but I figured I'd document this as well.

First, I was running ubuntu 7.10 and needed to upgrade to 8.04. That was a bit tricky as i tried to do 'do-dist-upgrade' but it kept failing. I finally figured out it was because i had all the repositories in /etc/apt/sources.list enabled. Needed to pear it back to just be the first one in the file.

So, after I did that I was up to 8.04. From there, I grabbed the 'request-tracker3.6' package using apt-get. This is by far the best way to do this, as it gets everything else you need as well. These instructions were well chronicled at http://www.ubuntugeek.com/request-trackerrt-ticketing-system-for-your-support-group.html. I stupidly didn't setup the database, so got a lot of weird errors.

Next, to get email integration to work... I wanted to check a gmail mailbox for our domain. To do that you need to configure postfix, these instructions (scroll down to the end where it say "Fetchmail\nWhy Fetchmail") http://souptonuts.sourceforge.net/postfix_tutorial.html. One thing it misses is that you need to remember to copy the hashed links of your certs to /etc/ssl/certs.

I have yet to get rt-mailgate to work, though.

Face Blurs in Sony Vegas 9

2008-11-08T19:31:00.000-08:00

So, it's easy to find face blur tutorials all over youtube for sony vegas <9, but what about 9? There's a checkbox that seems to have gone missing.

Anyhoo, someone made a great tutorial at http://www.broadcastnewsroom.com/articles/viewarticle.jsp?id=166551 that describes it to a T.

Making outgoing email work in vtigerCRM

2008-11-03T08:38:00.001-08:00

I was looking for a CRM for xitome and found vtiger through my hosting provider.

out of the box the outgoing email didn't work. to make it work, i found this post to be helpful:

http://forums.vtiger.com/viewtopic.php?t=19432&highlight=gmail

Getting Asterisk Running

2008-09-30T22:07:00.000-07:00

Asterisk seems like a pretty cool option for those that are interested in the masochistic endeavor of setting up and maintaining a system that is mission critical for communication.

I wanted to set one up for our house and found the following links to be useful:

http://www.geek.com/feature-how-to-build-and-customize-your-own-pbx-with-asterisk-20080812/ - This link is awesome! it gives you a shell script to run to setup your asterisk server. use the address below for the asterisk gui part.

Asterisk Gui SVN address: http://svn.digium.com/svn/asterisk-gui/branches/2.0 asterisk-gui

Editing MPEG2

2008-09-30T22:02:00.000-07:00

I just got a Canon FS100 and was excited to start editing movies with it.

Like many camcorders, it outputs MPEG2 files.

Firstly, there's something weird going on where it names the files something other than .mpg (.mod), and you use their program to import them from the camera. Once imported, you can export them to a 720x380 mpeg2 avi.

The kicker lies when you want to edit. I tried using it with Sony Vegas and it didn't quite work. I found that the video appear, but the audio didn't. After some looking around, I learned that mpeg2 is a "transport" format (i have no idea what that means), but it has something to do with the fact that it's meant for video, not both video and audio.

The solution is to transcode into something else. MP4 works, as it's both video and audio.

You can do this using ffmpeg. Once you've got your exported avi, you can issue the command:

Ffmpeg -sameq -i input.mpg output.mp4

you can then drop the mp4 file into your vegas project and you'll be good to go!

First Post!

2008-09-30T21:54:00.000-07:00

OK! so it's 1AM, i got kicked out of bed because i'm buzzed (and we have a baby, so totally reasonable), and i'm inspired..

I spend a lot of time sleuthing the internet to figure out the solutions to my issues, mostly those of a technical sort. I have this strange void in my soul that appears as i find posts of many like minded souls stuck in a similar predicament. i eventually solve my problem, but i know many others will fall into the same hole, and i hope they'll find their way out.

this blog is my contribution to the internet. i'm going to post the solutions to issues i've had and hopefully they'll be useful to someone (other than myself).