I have an HP C4580 printer, and it’s connected to my home network wirelessly. Wireless printers have to be the most unreliable things I have ever come across. Today was a typical day; attempting to print 4 photos, it took me 45 minutes. And I work in IT, so you’d think I stood a better chance than most of getting the damn thing to work. I honestly don’t know how “normal” people manage to cope with these things.

There were two main problems; first, there was some crazy, half deleted / cancelled job sitting in the print queue that just wouldn’t go away. Turns out, in Windows 7 (yep, I should have used my Mac, I know), you can get rid of stalled jobs by doing two things: restarting the printer and restarting the Print Spooler. The first I’m sure you can work out yourself. Just unplug it, for example, then plug it back in. The second one is only slightly trickier, but still ludicrously simple, in that you go to your Control Panel, then Administrative Tools, and then Services. Find “Print Spooler”, right click it and choose “Restart”. You might need to through this process twice (reboot printer, restart spooler) until it finally clears your jobs.

That should have been it, I’d thought. But it still didn’t work. This time, it was because my PC had managed to get all confused about the IP address of the printer. It is usually on my network at 192.168.2.200, but somehow it got 169.254.134.14 instead. I changed the IP address somewhere in the Printer settings on my computer, but it made no difference. So I just deleted the Printer entirely from my computer and added it back in, within Devices and Printers. After doing that, printing started working again.

It really shouldn’t be this difficult. Goodness knows what happened to cause this trail of devastation. Good luck if you too have a wireless printer.

Have been hitting my head off a brick wall for about an hour attempting to get Xemacs running on my windows machine via X11 from my linux vm. In other words, I have a linux vm running on VMware Esxi 5, and I wanted to run Xemacs (that is on the linux machine) on my Windows 7 box. I just kept on getting “X11 forwarding refused”.

Now, even getting to the point of seeing that error message was a PITA. I have been attempting to run XLaunch (part of Xming), and it’s pretty damn good, but it was crashing without giving me any output, so I had to run it from the CLI like so:

C:\Program Files\Putty>plink.exe -ssh -X -v user@vmhost xemacs21

When I ran that, I got the error message. I tried in vain to figure it out, and all it took was to edit the /etc/ssh/sshd_config file and add this in:

AddressFamily inet

…and restarting the ssh service after, obviously. And it worked. I shall give my head a rest now.

So I host various web sites, including this blog, with 1and1. Over time, I’m becoming more and more likely to move away from them, because I’d prefer to host on a VPS that I control and am not therefore at their whim.

Today is a day when I feel more strongly about this.

One of the sites I host is a simple Perl CGI based site hosting photos of my kids since my eldest was born over 11 years ago. Today (or at some point in the past 2 days), 1and1 upgraded the host and the site broke.

First, the upgrade. They went from Debian Etch to Debian Squeeze. Yep, they skipped Lenny. However, I’m glad they were to Squeeze, because Etch is fucking ancient, and Lenny is no longer the stable distribution. Quite why they didn’t upgrade to Lenny 2 years ago is anyone’s guess.

The site stopped working, I deduced after 10 minutes of adding debug logging, because the environment variable REMOTE_USER is sometimes called REDIRECT_REMOTE_USER. 1and1 sent out a notification beforehand, and in it, they said this:

If you are using very old scripts or software, please note that after
a redirect by Apache, certain internal variables, such as ‘REMOTE_USER’,
will henceforth only be available as ‘REDIRECT_REMOTE_USER’.

Now, I have to chuckle at this. They are blaming my “very old scripts” despite the fact that it is 1and1 who have been running a very old operating system, which therefore exposes environment variables.

Sigh.

So, if you have a bust website and you are using htaccess, as I am, and things are no longer working, it might be because of this. I thought it worth mentioning because it might save one random person a headache.

Started having a look at ICEfaces today. Seems ludicrously awesome and complicated at the same time. A good start.

Attempted to run through the easyAjaxPush tutorial, which is here:

http://wiki.icefaces.org/display/ICE/Easy+Ajax+Push

All went sweetly until I got to the first attempt to deploy, and the buttons didn’t appear. I decided to push on and see if they were there by the end, and they were not. However, it transpired that the fix is simple; the XHTML page isn’t being processed as a Faces Servlet. To fix it, all you need to do is add an url-pattern of *.xhtml, such that you have a servlet mapping similar to this:

<servlet-mapping>
<servlet-name>Faces Servlet</servlet-name>
<url-pattern>/faces/*</url-pattern>
<url-pattern>/icefaces/*</url-pattern>
<url-pattern>*.xhtml</url-pattern>
</servlet-mapping>

Note that this only involves adding one line to your web.xml file, don’t add the whole lot, as you should already have a servlet-mapping element for Faces Servlet there to begin with.

Once that was there, the example worked beautifully. It’s a bit scary how simple it is, tbh.

[ Later ]

OK, I’ve since discovered that, instead of adding this element:

<url-pattern>*.xhtml</url-pattern>

…you should instead add this element:

<url-pattern>*.jsf</url-pattern>

Now, this confuses me at the moment, because there isn’t a file with a .jsf extension, but something is doing that conversion for you. In other words, if you have foo.xhtml, you access it by going to foo.jsf and so long as you add in that url-pattern I just mentioned, the JSF tags won’t be ignored.

I figured out earlier this evening how to clone a vm in virtualbox. Then an hour later, they release a new major version (4.1) where you can do this in a ridiculously simple way. Here’s some brief notes on how to do this, without needing to go anywhere near the GUI (not that the GUI is bad, but it’s just ultra-fast in the CLI). Oh aye, and this is done by cloning a guest instance that is running Debian Squeeze. Some of the info is specific to Linux and/or Debian.

First, you need to get Guest Additions installed on your instance. It’s easy; just go to the docs to see how to do it. Doing this will allow you to find out the IP address later.

Next, you want to boot the instance and delete the following file:

$ sudo rm /etc/udev/rules.d/70-persistent-net.rules

This saves your machine getting confused about the network interface of your machine, as otherwise it’ll start without eth0 and with eth1 or eth2 (or higher; you get the idea) instead, and it won’t be configured.

Next, on the server itself, run this to clone it:

$ VBoxManage clonevm "Squeeze" --name Clone --register

…where “Squeeze” is the name of the original instance, and Clone is the name of, you guessed it, the clone we’re about to create. This might take a while depending on the size of your hard disk in your Squeeze instance, and how fast your disk is on your server. Go get some 15k disks; they are awesome.

Next up, we want to get the instance booted, so again from the server, run this:

$ VBoxManage startvm Clone -type headless

It’ll take a minute or two to boot up. Go make a cup of tea. Finally, to get the IP address of the instance once it’s booted, nail this command:

$ VBoxManage guestproperty get Clone "/VirtualBox/GuestInfo/Net/0/V4/IP" | awk '{split($0,a," "); print a[2]}'

That’ll return an IP address, and you can now SSH on to the host (assuming you have openssh-server installed on it, of course; if not, this was kinda pointless).

Two days in and virtualbox still impresses me for a variety of reasons. Slightly annoyingly, a new version (4.1) was released today, just 2 days after I downloaded it at home. Still, it’s a good sign that they are continuing to work on it, unlike vmware server (grrr).

It’s late, and I really should get to bed, but I just wanted to say one thing: virtualbox is truly in OMG territory. I can totally see myself moving to this from VMware which, tbh, is a total PITA these days. Virtualbox just seems to, you know, work. Straight away. With no crap. Sweet.

Gerrit is a web based code review system, and I decided I’d have a look at it. The install process is not easy, and not documented in such a way that I’d consider simple.

[Note: I later discovered the document that caused my confusion, which is http://en.wikibooks.org/wiki/Git/Gerrit_Code_Review – note that it, at this time, doesn’t contain anything about creating the project on the gerrit host!)

First off, installing the software (v2.2.1) to use MySQL just didn’t work; some unknown problem happened and the stack trace wasn’t helpful. I took the cowards way out and use the embedded H2 database instead, which gave me no such problems.

Next, the questions asked during install were not particularly obvious, if you didn’t know what it was really asking. For example, “Location of Git repositories” – what’s it asking here? It transpires that this is the location that Gerrit wants to store its own copies of the git repositories; or at least, that’s what I think it means. I ended up letting it use the default value which is simply “git”, because then it stores them within its own directory structure.

After this, I figured the documentation would tell me how to set up a new project. I’m not sure what I was originally looking at, but the first thing you need to do is to create a project repository by doing something like this:

ssh -p 29418 review.example.com gerrit create-project –name new/project

One of the guides I used did not have that in the instructions. This was a fail. As in most things, once you know how to do it, it’s easy, but on the basis that the instructions I was first using didn’t make any mention of needing to create a project outwith the web interface, I was stumped.

Now, at this point, I ended up causing myself no end of problems by thrashing around trying to get it to work. I ended up deleting something that I really shouldn’t have deleted, while attempting to get the git repository that I had initially stated in the setup (/usr/share/gerrit) to work. Bad move. Don’t go in and delete stuff in a terminal as otherwise you’ll confused the crap out of the web interface which then doesn’t cope well with things disappearing. So my bad there, but the lack of clear docs on how to get started didn’t help.

After the success of managing to get the new project created, it was relatively plain sailing. You need to create a branch within the web interface, I’ve yet to see any documentation about this but I didn’t look too hard as I figured it out from the error messages when attempting “git push” and from looking at the web app itself. I’ve yet to figure out how the workflow goes when making comments on a code change, but the UI at this point looks pretty neat, so I shall give it more of a chance from this point on (I’d given up an hour ago, judging gerrit to be pish, before my brain refused to be defeated and I came back and got it working).

So, ultimately, if you want to try it out, here’s my tips:

1. when installing, just use the H2 db, especially if you’re just trying it out
2. when asked for the location of git repositories, just let it put them in the default location
3. don’t go changing things in the filesystem after you’ve installed it; a schoolboy error, admittedly, but the tool doesn’t appear defensive enough to say “Oi! you’ve broken something here, mate!”
4. you need to ‘create-project’ from the command line before you can “git push” anything
5. you need to create the branch in the web ui before you can “git push” changes on a branch
6. be prepared for the ui to be green. I hate green with a vengeance.

Other than that, it looks good enough to continue investigations…

Is it just me, or is en.wikipedia.org currently down? I can’t even get an IP address for it atm. Hey ho.