Also, installed mailhide. Not sure what it acts like, haven’t actually put any email addresses here.

Now to figure out how to import posts from other WordPresses…

I’m going to merge this site with my other Linux site. There’s not much on either of them — certainly not enough to justify having two separate sites for them. Maybe being all in the one place will make me more likely to post to either of them (aside from anything else, it’ll mean I won’t be trying to bend posts to fit the particular site better). This site is likely to become a mix of me talking about real stuff (like computers, or ideas) and meaningless ramblings (like what I cooked for dinner, or how hungry I am… hmm.. that wasn’t intended to all be about food…).

Also, if traffic ever ramps up, I might put some ads on the site. No point in putting them there now though with the 20 hits per month that it’s currently getting.

Stop it, just please stop it. As well as everything that guy just argued, a “rock star” programmer to me means someone who’ll take horrible shortcuts and code nasty hacks into already dodgy systems (if the company is iffy enough to seek out a “rock star” programmer, they’ll be dodgy systems) to meet stupidly short deadlines. While completely disregarding proper planning and teamwork. Sure, they’ll look amazing to management in the short term (“wow! you got that working this quickly?? But those other shmucks told us it’d take twice as long!”), but in the long term, other developers on the team will want to knock their heads off the nearest whiteboard as they end up compensating for said rock star’s shitty, hacky should-never-have-been-given-life code.

Git confuses the hell out of me. I get how it works (use Mercurial regularly at work, which is the same DVCS idea), but github does seem to make certain things much easier (setting up a remote repo, worrying about SSH keys, that sort of thing), and gives you an idiot’s guide to the complicated bits (initialising the local repository, setting the remote master).

It doesn’t seem to be as feature filled as Bitbucket, but it’s possible that I just haven’t gotten around to prodding at it enough. I can’t find ticket or wiki support anywhere. Support for multiple “organisations” is nice though (you can associate your account with multiple roles, and not get your personal stuff mixed up with your work stuff, for example), may end up using it some day.

Another nice feature is the github web page thing that you get. Need to look more into what can be done with this, but it’s a nice little perk.

To make from scratch:

• Cook some pasta like normal until it’s just slightly harder then you’d normally eat it. Very slightly. Pour the water off but keep a cupful of it, and throw the cupful of water and the pasta back into the pot.
• Grate a large handful of parmesan cheese.
• Empty about 3/4 of a small jar of pesto into the pot (I’m not sure how big the small jars are – they’re about 2 inches tall).
• Put the heat back on and stir it in. Throw on the cheese and stir that in.
• When the water has mostly boiled off and everything’s mixed fully, it’s ready.

Goes great with garlic bread. Never thought I’d put cooking guides up here.

I just tried to buy a copy of an album online, because it’s fairly decent and I know the guy who wrote it (I’d mention the name of it, but I don’t want it to come up in a Google search for “lying cheating thieving bullies”). Unfortunately, PayPal have still got my account blocked (apparently my moderately generic name showed up “on a list” – shocking). Also, they won’t allow my card to be used without logging in. So I can’t buy anything through PayPal. I also can’t close my account with them. Not until I send them some photo ID, some household bills at my billing address that aren’t phone bills/bank statements, and probably the rights to my first born child. Ignoring the fact that I don’t actually have bills going to my billing address that aren’t phone bills or bank statements, who do they think they are?

This isn’t the first tale of woe I’ve heard bout them. There’s an entire community of people who are in the same boat (Google for it), and I’ve heard accounts of them (or rather, their hired goons) showing up on peoples’ doorsteps demanding debt repayments because they (PayPal) made a clerical error.

Someday I hope to have something worth selling to the internet. It won’t be through PayPal (or their parent, Ebay, which I used frequently up until this happened). They’re nothing more then a bunch of thieving, badly regulated bullies.

With that, I decided to cut my losses and go with dirvish instead. An advantage of dirvish over Time Machine is that I can easily navigate the backup images on the server, where as Time Machine used a sparse disk image format, which Linux didn’t like dealing with.

In addition to the steps yesterday, I did a couple more things – added an SSH key to root’s authorized_keys2 file on the laptop. OS X is one of the few operating systems that seems to disable root logins by default; a sensible measure, but one I have to undo. The SSH configuration is stored in /etc/sshd_config (not /etc/ssh/sshd_config as it usually would be). Also, the root user doesn’t quite exist on OS X – everything would normally operate through sudo. Root’s home directory is hidden in /var/root/, not /root as usual. I also added the CheckHostIP no line to /var/root/.ssh/config.

Following the instructions here and taking hints from Andrew Harford’s guide here, I set up a vault on my desktop that will back up /Users on my laptop. I’d do the whole thing, but it seems a little pointless given that I can’t really restore a full system image from it anyway (there’ll be extended file attributes that Linux can’t handle… rsync can via -X but Linux can’t). Also, anything outside of /Users can probably be re-downloaded from a vendor easily enough. I might back up /opt at some stage, if only because getting everything back up and running from MacPorts is an enormous pain in the neck, and ports wouldn’t make huge use of OS X specific file system features anyway.

The only other thing of note that I changed from either of those guides is that I specified the rsync-client option in the vault’s default.conf file. I set it to MacPorts rsync, which is newer then the Apple-supplied rsync (v2.6.9 versus v3.0.7) and doesn’t wait around as long building a file list as its predecessor.

rsync-client: /opt/local/bin/rsync

It’s currently initializing the vault, I’m expecting this to take most of the night (going over wireless).

In the morning (or whenever I get a second to look over it) I’ll look into hooking in Andrew’s expiration script, it makes more sense then using dirvish’s regular expiration mechanisms for a laptop.

3.8G down, 107G to go..

I decided to finally look into the mDNS (multicast DNS) thing – I know my laptop (running OS X) has it set up out of the box because Apple’s file sharing uses it. I think they call it Rendezvous. Linux supports it through Avahi daemon and libnss-mdns, which I know is enabled on my desktop (I set up netatalk advertised via Avahi a while ago on it). After a few minutes of googling, I figured it out – just need to append “.local” to a hostname to do a lookup using mDNS instead of DNS.

[andrew@Strongbad ~]$ ssh xenu.local
The authenticity of host 'xenu.local (fe80::250:8dff:fe95:4ae7%en1)' can't be established.
RSA key fingerprint is dc:bc:96:3e:77:e0:51:b9:74:37:3b:4b:06:f4:a5:17.
Are you sure you want to continue connecting (yes/no)? yes
Warning: Permanently added 'xenu.local,fe80::250:8dff:fe95:4ae7%en1' (RSA) to the list of known hosts.
Password:

Whahey, it worked \o/ Interestingly, it uses the IPv6 automagical link local address, not the v4 addresses on the network. There’s a slight problem with this though. The IP address was used in the host key line that was added to known_hosts. If the IP (or in this case, the interface I’m connected via) changes it’ll splurt out an error, and I’m planning to set up dirvish over SSH (want to back my laptop up to my desktop), so this will need to work in an automated sort of way. To get around this, I can make SSH ignore the IP address when checking host keys by adding the following to ~/.ssh/config:

CheckHostIP no

This would be a problem if anything I used relied on IP addresses as a security measure, but they don’t. Host keys associated with host names are all that matter. I remove the line that was added to known_hosts last time, and try connecting again using the mDNS name:

[andrew@Strongbad ~]$ ssh xenu.local
The authenticity of host 'xenu.local (fe80::250:8dff:fe95:4ae7%en1)' can't be established.
RSA key fingerprint is dc:bc:96:3e:77:e0:51:b9:74:37:3b:4b:06:f4:a5:17.
Are you sure you want to continue connecting (yes/no)? yes
Warning: Permanently added 'xenu.local' (RSA) to the list of known hosts.

It worked. Now to get dirvish to work. This lack of backups is making me extremely nervous (Time Machine failed me miserably last time I needed it).

It turns out that it only works for really small images too. Apparently larger images (I tested the script with 20MB images, today I had a 20GB image) are compressed in some strange way. Today, after a Windows VM decided to eat four hours of work on me, I wanted to try using Sleuthkit to recover the data, so wanted to get the raw disk image. After examining the image with xxd (and Sleuthkit throwing errors) I concluded that something wasn’t right, and then realised that the 20GB disk image was only a 10GB file. Something was not right.

It turns out that the correct way to do this is to use the VBoxManage tool:

VBoxManage clonehd --format RAW original_file.vdi new_image.img

This trick was found at this guy’s blog (thanks!). It took an age and made my laptop cry while it was working, but in the end Sleuthkit saw the filesystem and I could continue my (eventually futile) quest to recover the data. (Really, if you actually wanted Windows to delete a file that cleanly, it wouldn’t have done it. There was no sign of it on the image. Not even searching the entire thing with strings and grep returned any trace of it. Screw you Windows.)