Wow. Wait. What?

I am totally outraged. Seriously.

This morning, I received a mailing from my insurer, KBC Verzekeringen. This fact on itself already struck me since I have never, ever given them permission to use my email address for any commercial mailing purposes. However, I had made the mistake a couple of months ago to mail my contact there with my personal email address instead of using some personalized alias on my spam mailbox like I usually do. I suppose this is where they got my address from.

What caused my rage and made me write this post, is the fact that they have sent this email to all of their customers using the ‘To’ field. Really. Yes.

I wonder how such a thing is possible in a company handling a lot of sensitive data.

Furthermore, I have sent a polite though obviously displeased email in reply, to which I haven’t had any response yet. I would appreciate an apology. It’s the least they could do.

I’m seriously considering ragequitting my insurer.

I spent a good deal of last night repairing a MySQL master-slave setup for a customer of ours, running a fairly large website. One of their developers made the usual mistake of running a query much like “UPDATE table SET x = ‘This is a test’” and forgot the WHERE clause. This wouldn’t have been a big deal if it had been issued on their development environment, but this was on production.

Since this was a table with close to 60 million records, it took a while to execute on the slave, so we were informed in time and could prevent the query from propagating to the slave by killing it, forcing a rollback and stopping the slave replication. Luckily, this saved us the hassle of restoring from backups. We successfully dumped the table on the slave and updated the master again with the correct entries still on the slave using “SET SESSION SQL_LOG_BIN=0″. The master was saved and in good shape.

However, this left the slave without replication, and our worst nightmare came true. The Query Of Death was executed more than once on the master and thus was all over the binlogs that would be replicated. This meant that it was not as simple as skipping the next query on the slave and start replication again from there. The only thing I could think of was replaying the binlogs on the slave with the –safe-updates argument passed to MySQL, but that would take hours. We consulted MySQL to see if there would be an easier way, unfortunately it turns out there was not. So there I was, painfully executing log by log, anxiously monitoring every query that was executed. Finally at 4 o’clock everything was back up and running. The best thing? The website’s users didn’t notice a thing, apart from seeing ”This is a test” in some places for a while. Since the slave is used for full-text indexing with Sphinx as well, the searches were a bit off too, but nothing major.

I just wish there would have been an easier way… A query blacklist? Something?
If anyone has tips on how this could have been solved in an elegant way, please do tell.

yet. It’s FOSDEM!

For most of the staff, FOSDEM will begin tomorrow. We’ll start preparations at the ULB around 11AM, and hope to be finished around 10PM.

See you this weekend!

Two weeks ago, we had a FOSDEM staff meeting and we decided to make work of registering the FOSDEM domain names for the most common TLD’s. We noticed that the fosdem.com domain was registered by OpenDomain. I contacted them, a bit hesitating about their intentions, but I almost immediately received the kindest reply from Ric Johnson. He wanted to transfer the fosdem.com domain name to the FOSDEM organization, no strings attached! Thank you Ric!
If you want to know more about OpenDomain, please visit their website at http://opendomain.org.

The fosdem.be domain name was registered by Wouter Coekaerts, known for his contributions to the irssi IRC client. Jan VB contacted him, and he immediately agreed to transfer it to us. Also a big thank you to you, Wouter!

FOSDEM was very lucky that these domains were registered by people with good intentions. They could have been registered by a domain squatter, and we would probably have lost those domains forever. Such things teach you a valuable lesson. Please make sure that, if you start a project, you immediately keep the most important TLD’s from being registered by squatters.

Never even tought that this would work…

Notez-bien: eth0 was manually configured.

rush:~# ifconfig
eth0      Link encap:Ethernet  HWaddr 00:10:A7:0F:AB:43
          inet addr:192.168.92.7  Bcast:192.168.92.255  Mask:255.255.255.0

eth1      Link encap:Ethernet  HWaddr 00:10:DC:DD:75:DE
          inet addr:192.168.92.3  Bcast:192.168.92.255  Mask:255.255.255.0
rush:~# vi /etc/network/interfaces
 -- change eth1's IP from DHCP to static .7
 -- create configuration for eth0 with DHCP
 -- make sure your DHCP server hands out the old address of eth1 to eth0
rush:/etc/network# ifdown eth1; ifconfig eth0 down; ifup eth0; ifup eth1;
Internet Software Consortium DHCP Client 2.0pl5
Copyright 1995, 1996, 1997, 1998, 1999 The Internet Software Consortium.
... yaddieyaddieya
bound to 192.168.92.3 -- renewal in 2000 seconds.
Interface "eth0" is already enabled.
rush:/etc/network# ifconfig
eth0      Link encap:Ethernet  HWaddr 00:10:A7:0F:AB:43
          inet addr:192.168.92.3  Bcast:192.168.92.255  Mask:255.255.255.0

eth1      Link encap:Ethernet  HWaddr 00:10:DC:DD:75:DE
          inet addr:192.168.92.6  Bcast:192.168.92.255  Mask:255.255.255.0

How cool is that?

It seems like ages ago (well, actually it is) that irssi got an update.
Grab it while it’s hot! http://www.irssi.org

Oh and aparently they are looking for a logo designer, so if you have some graphics skills, maybe you could help those guys out? (hint, hint)