First of all, you don’t need to enable or install FTP to copy files from one place to another. SSH (and SCP) can accomplish this task (better than FTP can…). Second, you don’t always need to connect to Amazon Linux as ec2-user. This is really only necessary the first time.

To FTP, or not to FTP

If you absolutely must have FTP, configuring FTP for Amazon Linux is as easy as installing ‘vsftpd’ package, making some subtle configuration file updates, and opening ports to your EC2 instance. There is a great tutorial on vsftpd installation on Amazon Linux at That’s Geeky. FTP is fine under some circumstances, but from a security standpoint, it comes with its share of flaws. FTP requires opening an absurd number of ports (or switching to active mode - which inevitably confuses your users), it sends data in the clear, and I just don’t like it. Not to mention the fact that SCP is already available, since it uses SSH protocol and port 22 (which you have already in the Amazon Linux AMI).

…

So I joined Google+; more on that later. I joined with one of my non-primary gmail accounts, thinking it would shield me from some things - the effectiveness of this strategy is up for debate, but that is what I did. I thought it would be great to download the Google+ mobile app, and that’s when I remembered how much I hate my Samsung Captivate.

Apparently, there is a standard issue problem with adding Google accounts (in the Gmail app) on Samsung Captivates after the Android 2.1 update. Since this is my non-primary account, I never bothered to add it, so here I am.

…

Choosing a replication system

I needed to replicate (master-slave) a particular group of tables from one PostgreSQL database server to another. WAL shipping and integrated PostgreSQL streaming replicatino updates in 9.0 both require full database replication, so they were out. The primary candidates for replication at the table level are (IMO): Slony, Bucardo, and Londiste (part of SkyTools). These are all trigger based replication systems, and the advantages/disadvantages among the three of them are minimal, and seem to depend on fringe case scenarios that are extremely particular to specific implementations. That said, my decision went like this:

…

I admit, I have fallen short of my 2011 New Year’s resolution. Nonetheless, I continue to use, monitor, and promote open source options; and I have tried to stick around as a passive observer in a lot of the open source communities in which I initially embedded myself. One of the things I promoted in a couple of places was the concept that bug/issue trackers are not a tool that we must limit to development teams. Documentation, edits, administration, and other aspects of projects (and not just in open source!) all have issues. An issue tracker is a natural place to store these ‘bugs’.

…

Distance Matrix seems to have limited use for most, and a few niche markets that could really take advantage of the service. The problems I encountered using the service are that samples and documentation are minimal in comparison to other Google services, and the request limitations are quite restrictive (100 results per 10 seconds with a 2500 daily maximum for non-Premier requests). I will note here also that there is a hard limit of 25 origin points, and 25 destination points. So if you have 100 origins all going to a single destination, even though the request falls within 100 results, this requires four requests. To combat these problems, I want to describe one way to lessen the impact of the request limitations, and offer some details about how I use the Distance Matrix service.

…