Entrenched

I was debugging some SQL today when I noticed that I was seeing a query go through the PostgreSQL log a few more times than it should have been. We log all queries that take over 200ms so on the odd occasions some queries will show up. It's rare to see one show up more than a couple times, so this one just didn't feel right. As ignoring those hunches usually turns out for the worse, I had a look at the latest generated pgfouine output and I saw that it was showing up in there as well. Something was definitely wrong.

Well it seems the almighty Slashdot is itself vulnerable to Slashdotting.

We noticed it was returning a 500 Internal Service Error around lunchtime today and the noise on twitter confirmed others were seeing the same thing.

According to the chief network engineer Slashdot went down due to an massive amounts of traffic on their internal network, about 40 Gbit/sec across their core switches.

Service was restored after the problem switch was isolated, however the origin of the cause is yet to be identified.

Slashdot is a popular tech news website from which the term 'Slashdot Effect' was coined - where their thousands of readers all follow the same link in at once, bringing that site to it's knees.

I've struggled for a while trying to decided where I store the configuration of the web-based applications I'm developing. Do I store them on disk as files or in the database? And how can I do it quickly, cleanly, and in a way that won't lead to mistakes or catastrophic disaster?

Finally I have a technique I'm satisfied with. One that is safe, neat and low risk.

According to Monty Taylor, of the MySQL based Drizzle project, 'SQL is Dead'.

In a short presentation yesterday at LCA, Monty demonstrated why he thought SQL should take a back seat to more efficient forms of talking to relational databases.

I was surprised today at linux.conf.au at how many people were carrying around netbooks.

At breakfast someone at my table was demonstrating an application on one. And as I looked around the lecture hall during the Systems Administration Miniconf I spotted numerous ASUS Eee PCs, and Acer Aspire Ones.

I asked someone what the appeal was and they told me it's the incredible portability of the devices that makes them so attractive.

I find it interesting that a category of computers that barely existed at the start of 2008 is now emerging as a preferred choice amongst many technology professionals over the larger laptop, and I expect this year we'll see this strong grow continue.

But it gets me wondering: How many of those who regularly use a netbook also regularly use a laptop? And is there anything besides their portability that makes them desirable?

It's 2009 and what better way to start the year than with a conference.

This week I'm hanging at linux.conf.au. It's the first time in its ten year history that it's been held in Hobart. LCA is the leading conference for the Linux and Free and Open Source Software community down-under.

So for the rest of the week I'll be blogging from the conference and sharing my thoughts, observations and insights.

If you're wondering what else is going on you can follow the action on flickr, twitter, or around the blogosphere.

We've all heard about the clean feed. Can we actually do something to stop it?

I've recently realised that over the last couple of years, I've kept a much closer eye on what platforms websites are running. It's interesting to see parts of a site running on a different platform than the rest, or watching as a "redesign" suddenly comes with more than just cosmetic changes.

Recursive output buffering is a great tool in web-oriented programming, but it can go _really_ wrong.

No matter how often you use UNIX tools, once in a while you get caught out trying to put everything together. This happened this morning while I was setting up a cronjob and it didn't work as expected.