Part of my hesitation in moving away from FVWM and over to GNOME was having to give up a lot of my custom keyboard commands (my goal in life is to pretty much never use a mouse). GNOME Do helped to bridge much of that gap. After a few days, it became a pretty much essential part of my UI.

Problem is, although Fedora 11 brought a newer version of Do (0.8.1.3), the Do plugins aren’t yet in the Fedora 11 repositories. That doesn’t leave Do crippled, but I was missing some of the added functionality I had grown accustomed to.

I took some time at lunch to build Do and its plugins locally. It took a bit of effort since the installation instructions were geared towards Ubuntu, but I updated the Do wiki with notes specific to building on Fedora. Anyone who is missing the added features of the plugins can check out the build instructions over at the GNOME Do wiki to get the plugins installed for full Do awesomeness.

Repost: This happened back in Feburary 2009 and was posted as “Disaster Averted” mainly to annoy my students with the fact that class was *almost* cancelled that night. I’m reposting it with a more accurate title for searching purposes (read: I needed to find this again but couldn’t because of my annoying title).

When I got to campus today and turned on my laptop, I was greeted with an all black screen and just the word “GRUB”.

Shit.

Some quick background for my students… GRUB is one of the bootloaders available for Fedora. In short, a bootloader is what takes control from the BIOS at boot time and passes it to the operating system. No bootloader = no operating system. So here I found myself, on campus needing my laptop for class tonight, without anything to help me.

That last part needs some elaboration. Fedora added a really cool feature where the Fedora installation can be installed to a USB drive. While I was at Red Hat Summit last year, I got Fedora 9 installed to the USB drive they gave out as swag. That doesn’t really help much since I kept forgetting to swap out the USB drive on my keys with my newly pimped out Fedora install stick.

My forgetfulness came back to bite me, since when I needed it most, what should have been an extremely conveniently located live Fedora install is sitting in my office at home, a few feet away from half a dozen Live CDs of various Linux distros that also would have been helpful to have around.

That aside, this was actually sort of cool in a way. While studying for my RHCE, I found new and creative ways to destroy a Red Hat installation to practice for the exam. I had managed to run into this exact issue and knew how to fix it, I just needed some form of rescue CD to do it.

The first half of that was easy, I logged into a machine in the lab and downloaded the Fedora 10 ISO from fedoraproject.org (insert generic “Try that with Windows” comment here). Thankfully, the lab machines have CD burners. However, I needed to find a CD first. Having no luck digging around the lab, I went downstairs to the university IT department’s tech helpdesk.

“Are you a student?”

Ok, so it was a fair question to ask me. It’s the last class before Spring Break and I’m not feeling well, so I’m here in a Villanova sweatshirt, jeans, and sneakers.

“No, I’m a professor. I need the CD for class tonight.”
“Really?”
“I assure you, I’m not a student.”

Thankfully, one of my students last semester had introduced me to one of the IT guys in the room who recognized me and vouched for my employment. I can’t really fault them for asking; I’d get annoyed too if a student bust in the helpdesk asking for free stuff.

One burnt CD later and I was into the Live CD. A few commands after that, I rebooted sans CD and was up and running.

Which brings me to the meat of this post, how to reinstall GRUB from a Live CD. It’s actually pretty simple. After booting into rescue mode, run:

grub

That’ll bring you into the interactive GRUB shell (notice the prompt changes to “grub>”).

grub> find /grub/grub.conf
(hd0,0)
grub> root (hd0,0)
grub> setup (hd0)

The find command was just to verify the hard drive where the boot files are located. The setup command completed almost instantly, which actually had me a bit worried that it did nothing. But the output looked pretty convincing:

Checking if "/boot/grub/stage1" exists... no
Checking if "/grub/stage1" exists... yes
Checking if "/grub/stage2" exists... yes
Checking if "/grub/e2fs_stage1_5" exists... yes
Running "embed /grub/e2fs_stage1_5 (hd0)"... 16 sectors are embedded... succeeded
Running "install /grub/stage1 (hd0) (hd0)1+16 p (hd0,0)/grub/stage2 /grub/grub.conf"... succeeded
Done.

After that, it was as simple as rebooting without the CD. I’ll still need to look around to make sure nothing else is messed up, and even more disconcerting is what caused this in the first place. But for now, at least I’m not stranded on campus with a dead laptop.

In the meantime, hopefully now I’ll remember to put the damn bootable USB stick on my keys… (Update: I put it on my keys that night when I got home from class)

I can’t help but think of this xkcd comic which feels kinda relevant right now.

Happy Birthday Spacewalk

June 17th, 2009

Spacewalk 1 Year Birthday

Zeus pointed out that today marks the 1 year birthday of Spacewalk being announced. He mentions some of the technical accomplishments made during that time, but there are a few other aspects I find important:

  • Like he mentions, there have been 5 releases in that time. Even though the code is derived from Satellite, that’s a really impressive turn around for a new project, both from a technical standpoint (getting the build structure in place, adjusting to the new tools involved in being open) and from a management standpoint (moving to a model of rapid releases from much slower production releases is a bit shift in thinking).
  • The community is alive and kicking after just one year. It’s awesome to see users answering other users on the Spacewalk mailing list. The Freenode chat rooms (#spacewalk, #spacewalk-devel) are not just filled with Red Hat developers but rather users and other developers as well. We even have non-Red Hat contributors already.
  • The team hasn’t imploded in the process. I wasn’t around for the majority of the initial work to release it as open, but I have watched the team migrate to an open model. It’s not as easy as just telling everyone one day to start talking in Freenode instead of the internal IRC server. There were also a number of decisions to make regarding how to manage the relationship between Spacewalk and Satellite that could have been a lot uglier if the team wasn’t so solid.
  • There was a ton of activity and interest at the Spacewalk Campground at Red Hat Summit last year.

Check out more at Spacewalk wiki or come find us on Freenode at #spacewalk-devel.

I came to the realization that I have too many computers constantly running in my house. It’s not so much a “go green” thing to use less electricity as it is me being cheap and wanting to lower our bill. It’d also be nice to walk in a room and hear silence rather than whirring computer fans or have to clean dead spiders from multiple desktops. After reading an article in Linux Journal about hacking the MyBook (external network drive that runs linux under the covers), it looked like it could serve a lot of uses, starting off with being the house file server and backup system.

So far, I’m off to a good start. To begin with, the Linux Journal article is already somewhat obsolete as there’s no need to hack it anymore. The latest firmware flat out gives you the option to enable SSH on the device; once you’re in you can do pretty much whatever you want. One click in the web UI and one terminal later, I was logged directly into the system.

The MyBook runs BusyBox linux. I’ve never worked on a stripped down linux install like this, but other than missing a few commands I’ve come to love (I continue to type less; my brain just refuses to acknowledge that it’s not there) it’s pretty easy to navigate if you’re comfortable on linux.

Initial Setup


The instructions tell you to install the included software which only runs on Windows, at which point I promptly said f-that and tossed the guides. Once you plug the box into the network, it will use DHCP by default to get an IP. Use your preferred means for figuring out that IP, such as checking your router logs for the lease information. I went a different route and just port scanned my network looking for an IP I didn’t recognize.

-> nmap -sP 192.168.0.1/24
 
Starting Nmap 4.76 ( http://nmap.org ) at 2009-06-16 17:42 EDT
Host 192.168.0.1 appears to be up.
...
Host 192.168.0.112 appears to be up.
...

I snipped out the rest the results, but you get the idea. I only have a handful of non-static IPs internally so it wasn’t too hard to figure out which one was new. The box runs lighttpd and has a pretty slick (albeit a bit slow) interface to most of its guts, so after digging the included CD out of the trash to look up the default password, I was able to connect to it and start dorking around.

The MyBook comes with a lot of features, with the small caveat that I don’t actually need most of them. So the first thing I did was run around disabling things. It’s very cool that the device supports Samba and AFP, but frankly, I just don’t need them (I’ve had the wife running linux for years now, she rocks). I also killed FTP and the Mionet software. It wasn’t all slash and burn; I enabled SSH and NFS and made a new user for myself. Ultimately I’ll turn off the daemon that controls the slick white light on the front of it, but the novelty is still there and it looks damn cool at night (and really creeps my wife out since the light pulsates).

At that point, I was able to SSH into the box and (potentially) really get myself in trouble.

Enabling Cron


The file sharing aspect of the box is very handy, but part of wanting a device I could hack away at is to have it do more than just be a dumb file server. To start with, I want it to run out and backup my web sites (this blog included) nightly. I have a series of scripts on the system this is replacing that already do the task, it was just a matter of getting the MyBook to do them.

Cron is on the system, but it doesn’t work out of the box. I had to make one small change: creating the directory in which it’s going to store the schedules.

mkdir /var/spool/cron/crontabs

Here’s where I got lazy. I just set the permissions on the cron and crontabs directories to 777 to make my life easier. It’s far from secure, but I’ll work on that later. For now, I just wanted to get my non-root user to be able to schedule things.

The other change that needed to be made was to start the crond daemon. I didn’t find anything in the web UI to support this, but given the rate at which the firmware is exposing functionality I wouldn’t be surprised to find it there in the near future. I hacked up a pretty rudimentary init.d script so it would auto-start on reboot and was on my way.

SSH


Once I had the ability to schedule tasks, I needed the MyBook to be able to log into my web server. The ssh client is already installed, however it took a bit of hacking to get to a point where I could generate public/private keys to automate the login.

The main issue is that the home directory for all created users is simply defaulted to /shares. That directory serves as the root of all of shared files. Not that I’m expecting to create a ton of users, but I didn’t like the idea of a single .ssh directory for all users. I didn’t find any user mod tools on the install, so I just changed the /etc/passwd file to give my user a different (read: private) home directory. After that, it was as simple as running ssh-keygen like any other linux system (I won’t get into those details here, but it’s pretty easy to google around for generating SSH keys).

After installing the public key on my webserver, I was good to go for automatically logging in.

Rsync


I was happy to find rsync was installed and working out of box. Between that and scp, I was pretty set for my current backup needs.

Profit


The rest of my backup automation fell into place. I had to edit my scripts a bit since the MyBook doesn’t come with bash, but otherwise the rest of the process was painless. Nightly, my MyBook runs out to my webserver, backups a snapshot of the database, and copies over any changed files. All of this from a 6 inch by 6 inch little box on my desk. The desktop that previously did this has been powered down since the weekend. Awesome.

I’ve also copied over the “family” shared files to the MyBook, such as music and pictures. It was done simply through NFS so I won’t go into details here other than to say that there were no unexpected hiccups in the process.

There’s more I want to tinker with on the box. The web UI is in PHP and after a bit of looking around, I was able to find and start editing the web UI site. I’m thinking of adding my own section that will at very least report on the previous night’s backups. I’ll get to that eventually, for now I’m happy to have replaced one of the desktops in my house. :)

MyBook Bitch Slaps Desktop

Statement of Purpose

June 15th, 2009

A friend of mine just got accepted to a grad program at Cornell. He wrote up a good blog entry on the statement of purpose essay he had to write that might come in handy for any of my students who are applying for various grad and research programs.

The entry can be found here. Take a look around the rest of the site while you’re there, the dude is brilliant and posts about some interesting projects he’s into.

Mission Control

June 10th, 2009

In case you haven’t noticed, there is a pretty common theme among geeks: “Because I can.”

One of the things I was looking forward to in Fedora 11 is better support for dual monitors. I wasn’t disappointed, the UI is very intuitive and it just worked, allowing me to share my laptop screen onto a spare monitor (seriously, I have too much hardware just lying around). So my home office now looks like:

Mission Control

Something else to realize is that with my synergy setup, a single mouse/keyboard is used to control both systems, meaning I can move the mouse from the leftmost side of the laptop to the rightmost side of the right monitor, an act that currently requires me to pick the mouse up to get more space 6 times (yes, I actually counted).

Not sure how long I’ll try this setup out; I’m guessing I’ll have a wicked headache before lunch. But that doesn’t change the fact that my office looks like a NOC right now.

My work laptop is a Lenovo T61 (side note: easily the best laptop I’ve ever had). It comes with some cool extra features, such as a fingerprint scanner that until now has remained unused. After a bit of dorking around, I got it enabled in Fedora 11.

It was actually pretty simple. I just needed to install a few extra packages:

yum -y install fprintd fprintd-pam gdm-plugin-fingerprint

The next step is to tell Fedora to allow fingerprint logins. This is easily done through the GUI through System > Administration > Authentication. Under the Authentication tab, select “Enable Fingerprint Reader Support”. Keep in mind that since it’s a system-wide preference, you’ll need root to make that change.

Once the Fedora install itself will let you login with a fingerprint, you need to register your own. The easy way would be through the UI at System > Preferences > About Me; a button for Fingerprint Login should be at the top right of the resulting dialog. In my poking around I had actually registered (or “enrolled” in their terminology) my fingerprint through the terminal. There are a series of fprintd- executables that will let you bypass the UI. In this case, I used fprintd-enroll (I won’t talk through the steps, it’s all very self-explanatory) and then fprintd-verify to run a quick test to make sure it could read it back correctly.

That’s it. Log out, select your name in the list, and swipe away. Sure, I only lock my laptop when I’m at school. Yes, my password takes maybe 3 seconds to type in. But this wasn’t about making my life easier, it was about the cool factor. And yes, being able to log in by reading my fingerprint is damn freaking cool.