Cloud Computing Trials – Launching Your First EC2 Instance
Okay, so after you have completed the signup process (which was fun because I got called by a very friendly female computer asking for my verification PIN) you will probably want to launch an EC2 instance. Here’s what to do:
- Click “Launch Instance” in the EC2 Dashboard
- Pick one of the starred Amazon Linux AMIs
- Set the instance type to “Micro”, click “Continue”
- Don’t worry about the options you have, simply click “Continue” again :)
- Now you can tag your instance, which isn’t really necessary if you are going to have just one instance
- Next, you can generate a key pair – do this if you know what a key pair is, otherwise you can skip this
- The next step is to set some firewall rules, I didn’t care about that for the first shot and just hit “Continue”
- You reach a review page where you can perform a final confirmation of the chosen settings. Hit “Launch” to start the instance.
After the instance has started, you can connect to it via SSH using the public DNS name and your previously generated key pair.
Okay, this was less than 5 minutes of work – quite fast, compared to the process of a normal dedicated/virtual server setup! Nice!
Cloud Computing – Experiments Kickoff
I’ve recently been confronted with cloud computing at work, which at first didn’t make me happy because you inevitably face constraints unknown from the world of dedicated or virtual root servers. For example, you might sooner or later realize that a good part of cloud computing is about creating images that can be cloned in order to quickly launch a deliberate number of additional instances running your application, which facilitates everything around scaling. You might also realize that if you launch an instance using such an image, this instance will of course not have any “saved state”, e.g. there is no long-term storage by default, which can get really annoying if you try to do Continuous Integration with cloud services. There are solutions to those problems, however.
I realized today that Amazon offers a free EC2 Linux Micro Instance for one year, so I decided to give it a try. I was able to sign in with my regular Amazon shopping account, which was a very positive surprise to me. I’m now going to feed my secret credit card data to Amazon in order to create my EC2 instance.
I intend to play around with the various services Amazon offers, including Elastic Beanstalk, which appears to be very interesting for Java web application experiments “in the cloud”, so that maybe I can get a better understanding of the advantages cloud computing has to offer.
For detailed information on the free services, see the Amazon AWS website – no worries, I’m not getting paid for posting the link :)
How to use Eclipse and Mercurial for local development
Mercurial is one of the most hyped version control systems right now, and it’s perfectly suited for quick repository setup on local development machines. I’ll tell you how to create an Eclipse Java project along with a local Mercurial repository. I assume that you have Eclipse and the Mercurial plugin already installed – this shouldn’t be too hard to accomplish ;)
It’s really absolutely simple:
- Create a new Java project
- Right-click your project, pick Team -> Share Project
- Select Mercurial
- Click Finish
This will basically do nothing else than running “hg init” inside your project folder in the workspace.
That’s it!
I’ll get back to this topic when I have finished experiments with multiple projects in one Mercurial repository – will see how that works out.
Update: don’t try to put multiple projects into a single Mercurial repository. It’s not designed for doing so. ;)
Abstract art created by JIRA
I use JIRA for some private tasks like organizing my master studies, which works quite well. Just for fun and profit, I’ve also used issue burndown charts with a sliding end date (e.g. exercise deadlines etc.), which was always very helpful. Looking at the complete chart, you can see quite well at which time new exercises came in and when I sent them out :-) The cross-cutting number of 15 open issues represents one block of exercises that I didn’t manage to work on because I didn’t have time to do so. The steep drop of the chart at the end marks the day the exams were over – why should I still do the remaining exercises at that point ;-)
Here’s the piece of abstract art generated by JIRA:
VServer fragged and repaired
Following an idea to try something, I decided to upgrade my Linux virtual root server to Ubuntu Lucid yesterday. Bad idea, because after the upgrade, the server didn’t start any more :) Luckily, reinstalling a vserver is easy if you are a HostEurope customer (which I am), so I reinstalled the entire server, pushed the backups back, and now everything is fine again. I will now retry the experiment I wanted to do, but without touching the operating system :)
Cinema 4D vs. Intel Graphics Chips
So I had this daunting task of fixing a broken Cinema 4D installation – being given the usual excuse “no, I didn’t change anything, but now it’s not working any more!” :) I quickly realized that Cinema was crashing in relation to ntdll.dll, wow, that’s about one of the most “inner” Windows DLLs. I tried some stuff related to OpenGL and other settings, all to no avail – I could run the program, but only in 256-color-mode. Not that nice to work with.
To keep things short and simple: the problem wasn’t Apple Quicktime, the problem wasn’t an outdated graphics driver, the problem wasn’t caused by missing software updates, the problem wasn’t Windows 7 – the problem was that the Intel graphics drivers were installed. After an hour of fiddling around with lots of different approaches, I simply wiped the Intel drivers off the machine, and now Cinema runs happily. Oh, it was an Intel 965 Express chipset, which is certainly not suitable for 3D applications at all, but who cares – now, the prog runs just fine, and I’ve made the world a happier place – at least a little achievement ;)
So, anyone out there willing to use Maxon Cinema 4D on an Intel (onboard) graphics chip: go get a decent graphics card!
Erasing old data, the easy way
So I got this new shiny laptop for work, which happens in cycles of 3 years – and now, I need to exterminate all traces of what I did on the old laptop before it goes back to the hardware pool or, even worse, out to someone who doesn’t belong to the company.
Up to today, I’ve always been horrified by the amount of work that was involved in erasing a HD properly – get a Linux ISO, burn it, boot, shred – it was always pretty stupid to waste a CD and lots of time. BUT: I learned how to speed this up today.
There’s a nifty tool called UNetbootin that allows you to quickly create a bootable USB stick. You can pick neat distros like Mint 9 Live (and even Mint back to version 3 or so), lots of Ubuntu editions, DamnSmallLinux, a couple of antivir rescue images, boot loader images, and so on – then, you only need to click a button and your USB stick will be prepared with the selected image. Cool.
I picked Mint 9 Live for my task because a) I wanted to see how fast it would be, running from a USB stick, and b) I wanted some eyecandy for this emotionally draining task of erasing beloved data.
I recommend using a reasonably “new” USB stick, because old ones are often very slow and/or buggy. So now I have this really nice Mint Live USB stick, and the old laptop is shred’ding away /dev/sda – as soon as it’s done, I’ll create a new partition, format it a couple of times using different fs types, and then return the laptop to the company’s hardware pool. :)
JDK6 installation without running the installer
As a Java developer, I often need to install JDKs, and often I need them to exist in parallel in different versions. At least for Windows, it’s only possible to download installers as EXEs that will install both the JDK and the JRE on your computer. Each time they run, they will put java.exe and a couple of other files into the windows installation directory in order to provide globally callable Java executables. WTF. I believe that this is not only a convenience mechanism for the end user, but also a way to force people to accept license agreements etc. by clicking a button. I do always accept those agreements, although I regard the fact that I’ve clicked a button in an application a really fragile agreement between me and the publishing company. So, I performed some experiments and it is of course possible to get all the required files without really running the installer. I can assemble my JDK and use it, bypassing the installer which messes with my system. So far so good. It’s quite a bit of work to do so, but it works in the end. I won’t publish the steps here.
Now for the question: dear Oracle, Sun, or whatever you may wish to be called – why can’t you simply put a checkbox on your website, a checkbox that allows me to agree to your terms and conditions, and then give me a ZIP that I can simply extract and use? Please! I know that this was possible in the past. Come on, it can’t be that hard, can it?
Comments welcome.
Red Dead Redemption 100%
As announced here, it was my declared goal to reach as much game completion as possible. I’ve reached the 100% mark today, so I think this is the maximum ;) Nice game.
Sun, or Oracle, JDK installer fail
In case you run into trouble because you used the Windows x64 JDK6 installer and you can’t uninstall, here’s what I did: I uninstalled the JRE first, which made it impossible to uninstall the JDK, all I got was an error 1723. Looking into the windows event log, I saw that the JDK uninstaller tried to load a regutils.dll from JAVA_HOME/jre/bin/ – which was not there any more because I had already uninstalled the JRE. I copied the JRE folder back to the installation location and then I was able to uninstall the JDK as well.
Don’t get confused if you use Google because most forum postings will tell you to update the Windows Installer etc. etc., this is probably senseless, especially if you – like me – have a completely up to date Windows 7 on your box.