tonywhitmore.co.uk  

Episode 13 of the Ubuntu UK Podcast has been released. In this show:

• A interview with Billy Cina from Canonical about training and certification, and have details of how you can get a discount on Ubuntu training.
• Sarcastic News.
• Trademark issues with Ubuntu.
• An interview with Andy Stanford-Clark, Master Inventor from IBM.
• We discuss Virtual Private servers and their purpose.
• We announce the winner of the Canonical store voucher from the last episode and start a new competition to give away a virtual private server.

Both interviews in this show are really interesting and well worth listening to, particularly if you’d like 10% off Ubuntu training!

This episode is a few days late as real life has intervened on all our counts. I would say it won’t happen again but it might so I won’t. We also forgot to mention that Alan isn’t in this episode because he was on holiday at the time of recording.

Three or four of my friends on Facebook are actually couples, where two people share the same account. My instinct is that this seems a counter-intuitive approach to social networking for several reasons. But I could well be wrong and would welcome other people’s thoughts.

1. It makes it harder to find people. When people have a shared account they often end up with a hybrid name which is less likely to be found in a search, or catch your eye as a name you recognise in the “People you might know” list. Add in change of name brought about by marriage then chances of recognition are even further reduced. It is possible to enter maiden names as part of a profile, causing your account to show up if people search for you by your maiden name, but not many seem to do this and it undoubtedly works better with only one name being tinkered with.
2. It feels weird being friends with someone you’ve never met. Sometimes I know, and consider myself friends with, both halves of the relationship. Sometimes I don’t. If I’ve never met your partner, seeing status updates etc. from them is probably not a huge amount of interest to me and I can’t tell which content added to your profile is from you. On the other hand, if you have separate accounts, I have met your partner and decide I would like to keep up with them, I can add them as a friend too.
3. It negates the networking part of social networking; being able to define your relationships with other people. I met him once, I worked with her, I spent three weeks mountain climbing with him, I ‘ve been married to her for 45 years, that sort of thing. You can define yourself as being in a relationship with someone and identify who that someone is. You can still leave people viewing your profile with no doubt, if that is your wish, that you are firmly coupled up, without having a shared account. With a shared account, your timeline or employment data, if you have entered any, will be a mish-mash summation of both your experiences. If I was at school, say, with one half of a couple with a shared account, the logical inference from the network is that I was also at school with the other person’s schoolmates. We can’t have logical incongruities on social networking sites!
4. What happens if a couple with a shared account were to split up? Not that I would wish it upon my friends, of course! They’ll be faced with reconstrucing the digital manifestation of their social network. Friendships will have to be requested and accepted. Photos will have to be re-tagged, maybe re-uploaded. It’s easy to imagine arguments over who gets custody of the Facebook account. With separate accounts, you can just “cancel” your relationship and inform everyone of the fact in one fell swoop!
5. How do you tag correctly a photo with only one half of the couple in? How do you tag a photo with them both in, but in different parts of the frame?

So, why do people have shared accounts? Indeed, is it the result of a specific choice or just the default position for some couples that share everything? Is it so that both halves of the couple are reassured that they know what their oppos are up to on the treacherous and fickle Interwubs? Are there any up sides? Have I got the wrong end of the stick?

I followed the official HOW-TO on installing from a USB stick. I have a local mirror of the Ubuntu repositories so wanted to use them as a source for downloading the packages, not the archives out there on the ‘net. It also meant I didn’t have to download and copy an ISO on to the USB stick. The first thing I had to do was add main/debian-installer restricted/debian-installer to my mirror, as these little repositories hold the code needed by the installer. Once I did that I was able to work through the installer as normal, manually specifying my mirror at the appropriate point. The “Ubuntu Desktop” meta-package failed to install due to dependency problem, but the Xubuntu Desktop package installed OK. The installer detected the existing (original) installation on the first hard disk partition and added it to the GRUB list.

Happily the new installation booted without needing to add the “acpi=off” argument to the kernel. I did notice some errors related to the PnP BIOS but the system still booted. However, I suffered from the same bug as Alan with the video driver. I discovered that switching to the console (Ctrl+Alt+F1) and back to X (Alt+F7) cleared the display. What I got then was the upper left 640×480 pixels of an 800×600 display. But I could log in and start applications etc. It’s an interesting and disappointing regression though. I have thus far failed to get anything more out of the display than that. Removing xorg.conf and allowing the GUI configuration tool to start was probably a good idea, except that the font on the tool was too small to read. By miles and miles. The icons and other furniture were normally sized though. In order to get round that I removed xorg.conf again and rebooted without the VGA cable attached. This caused the failsafe X configuration tool (xfailsafedialog) to run with readable text. Win. Eventually after several attempts I got a reasonable XFCE desktop running, however I totally failed to get X running in widescreen with the geode driver.

In fact, for ages I failed to get it to produce anything more than 800×600 on my LCD TV, despite numerous attempts to configure X to do more. Along the way I discovered that the tiny fonts issue is X using a high DPI setting for fonts by default. By opening Applications->Settings->Settings Manager->User Interface, selecting 75dpi and restarting X I got nicely legible text, even at the default 9pt font used by XFCE. Eventually, through a combination of copying xorg.conf from my MythTV box, using the X configuration tool and blind luck, I found a setup which produced 1024×768. Then 1280×1024. No widescreen resolutions though, I guess the chip or driver just doesn’t support that yet. There was a weird bug at the higher resolutions though that the mouse wouldn’t click on the “Applications” menu properly. It worked fine controlled remotely using x11vnc though.

Generally I’m finding this unit quite quirky. It doesn’t power off properly when shut down, and hangs on a reboot. It has occasionally done the “white screen with no POST” thing I talked about in my first report on this box. It’s doesn’t seem reliable enough that I’d want to use it in on a remote site or install it in an inaccessible place (without full remote control over the power and a remote console connection, anyway. ) That’s a great shame as its ideally specified to be an appliance.

The laptop-style PSU has an output of 12V pulling up to 3.3A, which suggests it could be powered from an in-car charger socket, making it ideal for an in-car entertainment system. In operation the unit is virtually silent. You can hear intense hard disk activity and there’s a slight electronic “zing” from the device, but to all intents and purposes it’s silent.

The first time I powered it on, a white screen appeared and stayed there. No sign of HDD activity, or indeed any other sort of activity. After a power cycle, I saw a familiar BIOS POST screen, followed by GRUB and the Xubuntu 7.04 splash screen. Booting was not very fast, about 90 seconds to the login screen with a further 20 seconds to the fully loaded Xubuntu desktop.

I logged in as the default non-root user. There was a volume entitled “888M Volume” on the desktop which I was unable to access via the icon. This leads me to wonder whether people are expected to use the root account by default as Windows users tend to do with the Administrator account. Alongside Firefox 2.0.0.3 was Opera 9.21. Rather weirdly, these were the only two applications given their own shortcut on the taskbar. It seems odd to include two web browsers anyway, but to give both equally high profile could lead to confusion. In the trash can was a shortcut to Opera, originally in /home/david/Desktop/. Presumably David was the guy who made the image up for these units. There was also a group in /etc/group called “david”. The output of lspci and /proc/cpuinfo is:
00:01.0 Host bridge: National Semiconductor Corporation Geode GX2 Host Bridge (rev 21)
00:01.1 VGA compatible controller: National Semiconductor Corporation Geode GX2 Graphics Processor
00:0d.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL-8139/8139C/8139C+ (rev 10)
00:0e.0 USB Controller: VIA Technologies, Inc. VT82xxxxx UHCI USB 1.1 Controller (rev 61)
00:0e.1 USB Controller: VIA Technologies, Inc. VT82xxxxx UHCI USB 1.1 Controller (rev 61)
00:0e.2 USB Controller: VIA Technologies, Inc. USB 2.0 (rev 63)
00:0f.0 ISA bridge: National Semiconductor Corporation CS5535 ISA bridge (rev 13)
00:0f.2 IDE interface: National Semiconductor Corporation CS5535 IDE
00:0f.3 Multimedia audio controller: National Semiconductor Corporation CS5535 Audio
00:0f.4 USB Controller: National Semiconductor Corporation CS5535 USB (rev 06)
00:0f.5 USB Controller: National Semiconductor Corporation CS5535 USB (rev 06)
processor : 0
vendor_id : Geode by NSC
cpu family : 5
model : 5
model name : Geode(TM) Integrated Processor by National Semi
stepping : 2
cpu MHz : 398.450
cache size : 32 KB
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 1
wp : yes
flags : fpu de pse tsc msr cx8 pge cmov mmx mmxext 3dnowext 3dnow
up
bogomips : 798.74
clflush size : 32
I had connected my unit to my LCD TV, capable of running at 1920 x 1080. The onboard graphics card automatically configured the display to run at 1280×1024 which is pretty good. The screen configuration dialogue offered VGA, SVGA and XGA resolutions. There were three small black rectangles across the top of the screen which disappeared when I moved the mouse over them, but I’m not sure if this is a fault with the drivers, desktop environment, graphics card or panel.

main, restricted, universe and multiverse repositories were all enabled. There were 136 updates to be applied to the new system. It would have been nice if all available updates had been applied as part of the build process. The 80GB HDD was partitioned in to two ext3 ~35GB partitions, one mounted as root and 5% used, the other mounted at /scratch and essentially empty. A slightly weird mounting system it has to be said, especially as the default non-root user didn’t have permissions to write to /scratch. It would have made more sense to make the second partition /home where users might make use of the storage, rather than /scratch which many people may not find and work out how to use it to store data on.

It was about this point that I worked out the reason for the mysterious “888M Volume” icon. There was a swap partition on the hard disk, but when I looked at the output of top, it showed no swap enabled. Running “swapon -a” returned an error and I needed to run “mkswap” on the partition before it registered as swap. So it looks like something in the imaging process didn’t set up the swap partition properly. It wasn’t getting enabled but was getting detected by HAL/dbus and showing up as a volume on the desktop. Weird.

Then I rebooted and it kernel paniced. So I tried with the old kernel, that paniced. I tried with the “noacpi” option suggested by Alan and it still paniced. So I’ve broken it within one evening of playing with it. Eventually, I worked out that the incantation needed to be “acpi=off”, after which I was back in the game.

I then tried to play back an OGG Theora file, to test the unit’s suitability as a media front-end. The default media player is XMMS which doesn’t play Theora files. So I installed Xine and although the internal speaker played back the soundtrack to the video for a few seconds, the picture was green and purple and just plain not right. It did persevere with attempting playback but I put it out of its misery once I was able.

The second 35GB seemed like a useful place to install other operating systems whilst leaving the default installation intact. The onboard NIC doesn’t support PXE booting, so the options really were for a USB CD-ROM drive (which I don’t have) or a USB flash stick thing. Fortunately I do have one of those. There was even a very useful official HOW-TO on getting the USB stick set up correctly. The USB stick had to be in one of the rear USB sockets, not the ones on the front. There were at least three different installations I want to try on it. Xubuntu Hardy, Ubuntu Hardy (GNOME) and Ubuntu Hardy Server. My successes or otherwise will form another post.

This is my significantly late review of LUG Radio Live UK 2008.

Having taken the Friday off work, the drive up the country was fairly leisurely, stopping to pick up bits of equipment on the way. Having had walkie-talkies at the US event the gents were adamant they wanted them for the UK, mainly so they could look important and pretend to be spies. Having arrived at the Novotel to find the power out in the room (not good when you have several video cameras to keep charged up) and a £4 car parking fee we headed over to Jono’s for a chat with him and Aq about the plans for the weekend and pizza. Then back to the hotel for a shower and off to the Hog’s Head for the traditional Friday night piss-up. It was great to see so many members of the LUG Radio community there. It was impossible to talk to everyone, such was the size of the crowd, but importantly all the crew heads who had been annoying each other on conference calls for weeks finally got to chat without the “satellite delay.” And Neuro, mrben, Dave2, oojah and many more! It was also good to see new faces there, not least Ciemon who seemed to settle straight into the LRL vibe. Brilliantly Emma Jane, ultra-lovely lady who we hung out with at the US event was there, having flown from Canada to speak in the UK. I also chatted to Matt Garrett for ages, discussing social etiquette in geek circles. I avoided having more than a handful of drinks (which is quite enough for me these days anyway), unlike some people who spent the rest of the weekend suffering.

The Saturday morning get-in was much more straight forward than last year. It really helps when you know where the plugs are, what equipment you can expect to encounter and what you need to bring with you. One of the highlights for me was having such a great AV crew. It was four times the size of the crew last year, and that meant that everyone got some time off to enjoy the event, as well as ensuring we got coverage of all the talks. It really made the whole weekend go much more smoothly. Barring a power outage when we tripped a circuit and found out that the caretaker wouldn’t be in to reset it until Monday, it all went pretty smoothly. So much so that Jono had time to think up stupid ideas like the “Furries for Justice” protest during the Second Life talk.

As ever I didn’t get to see any talks from start to finish, but I saw bits of Emma and Bruno as well as most of the Gong-a-thong, featuring more of mrben than I’d ever wanted to see. I gave my little contribution, a re-written version of the “We happy few” speech from Henry V on the end of LUG Radio. It got a few laughs which was good, but I’m not sure how many people in the audience knew the speech that was parodied. The live show, which was supposed to be the last ever LUG Radio episode, turned out not to be. After many loving kickings from the community the gents relented and we will be running another LRL event next year. Nothing is planned yet of course, but it will be a lot of work to make the event as popular without the regular podcast episodes to provide publicity. I think it’s really great news though because this year it seemed like LRL had really grown-up. Without being more mature. I’m sure next year’s event will rock.

The next best bit of the live show was that the gents gave me the coveted title of “Community Hero.” It’s a real honour to be given this title, even if I unluckily won the year there was no prize to accompany it. It also shows, I think, how open the LUG Radio community is. Like all communities from which friendships have grown, accusations of cliquishness are sometimes pointed at the LUG Radio community. But I’m not part of WolvesLUG or the Scottish Mcfia, I didn’t have any special “in”s. I just listened to the show and sent a couple of e-mails. I joined the IRC channel and the forums, admittedly early on when it was easier to get to know people. I went to the first LRL event and noticed that the presenters spent quite a lot of time helping speakers get their laptops working with their projector. I thought they probably should be doing other things and afterwards collared Aq and suggested that I could help out doing that if they held another event. They did, and I did. It went from there. I just volunteered for stuff and showed that I could do it to a reasonable degree of competence. It’s a great community and I’m proud to be part of it, and especially proud that the gents would think I’m “hero” material. Now I just have to get the conga line off Xalior.

After the show was the after party, but that’s a post for another day.

In a recent episode of the Ubuntu UK Podcast we reviewed then gave away a mini-PC from Viglen, the MPC-L. We also told you how could get hold of the units for just under £80, which is significantly less than the advertised price. I’m not going to tell you how, if you want to know you’ll have to listen to the show.

Anyway, I ordered one about a week and a half ago and it turned up today. First impressions? Well, Viglen probably aren’t going to be winning any rewards for environmental packaging any time soon. The unit is almost exactly the same width and depth as a CD case, but was shipped in a standard size PC box. I presume this is largely because they included a keyboard, but the amount of empty space and air pocket packaging showed that the space was not being effeciently used. Spot the 50p on the top of the box (and the actual unit below) for scale.

On top of the box was a black Viglen-branded USB keyboard and a Microsoft-branded USB optiocal mouse. This mouse feels like it’s the cheaper end of the Microsoft USB optical mouse range, but at the price there’s no reason to complain.

See what I mean about the packaging? Inside the small white box is the bubble-wrapped MPC-L unit, along with a laptop-style PSU. Oh, and unlike the review unit there was an instruction sheet giving the default username and password. The specification sheet enclosed in the box lists 512MB RAM, 80GB HDD (not the standard 40GB) and 12 months on-site maintenance, 7 day response. And a “Xubuntu Operating System pre-installed.” All for £79. Yey!

Although I’d seen, even dismantled, the review unit having one of my own is quite pleasing really. The unit is so physically small that it lends itself to all sorts of cool ideas. Indeed we had lots of interesting suggestions as answers to the competition. I’m going to try and run it as a MythTV front end just to see if it can cope with it, but other devious thoughts have already formed in my mind. I’m quite tempted to buy a second one.

So if you haven’t ordered one yet, download the show and find out how to order yours. Any geek worth his or her salt will find something interesting to do with this little box! Now I have to go and turn it on…

Episode 12 of the Ubuntu UK Podcast is out! In this episode:-

• A great chat with Chris Jones about his project Terminator and his dreams of a robot future.
• Sarcastic News
• We discuss the new Ubuntu training courses and certification
• A great chat with Jeremy Allison from the SAMBA project.
• Some listener feedback.

• Competition:
• We announce the winner of the Viglen MPC-L and start a new competition to give away another Canonical Store voucher

This wraps up the content recorded at LUG Radio Live and sets up some exciting stuff for the next episode. Get it here.

Some of you may have read my previous post about the state of my ADSL service. Judging from the number of comments, quite a lot people read it. Some of the suggestions revolved around using the engineer’s socket on my master socket to test the connection. The master socket was too old to have such a socket, but when I took the faceplate off I noticed that it didn’t have a ring capacitor. That’s bad. Weirdly the slave socket did have one. So there was at least one somewhere in the circuit, just the wrong place. Could this be affecting the performance of my phone line? We’ve not had any work done on the phone line since moving in, so having double checked my wiring to make sure I wasn’t making an idiotic mistake, I logged a fault on BT’s website. Having worked my way through various “wizards” on their website I logged the job as a wiring fault. I was also able to select the half day in which I would like the engineer to call. (It would have been better to be able to select a smaller window, but that’s a relatively minor gripe.) Even better, there was an engineer available in a couple of days’ time.

Spin forward a couple of days and I got a phone call from a man called Stuart telling me he was on his way and would be there in ten minutes or so. When he arrived he recognised me as he’d been reading my website! Having explained the problem, he investigated the junction block on the front of the house (not likely to be a cause of resistance on the line, apparently) and the master socket. I was right, there was no ring capacitor, so he replaced it with a new NTE5 socket. He also checked the slave socket, cut out the ring capacitor and ensured the terminations were sound. Although the problem was logged as a phone fault, Stuart usually dealth with broadband faults and as part of his checks to ensure the line was working properly tested that the ADSL synchronisation speed with the exchange was good. The software on his laptop reported 1.4Mbps, which Stuart seemed to think was reasonable being so far from the exchange. I would be pretty happy with that speed if it was stable.

My own kit, once reconnected, was getting 1Mbps, so there’s still something causing a drop in speed. This is despite replacing the microfilters and even buying a new ADSL modem. I’ve ordered an interstitial plate (a.k.a. iplate) for the master socket to see if that improves things.

So thanks for your efforts, Stuart!

I have an 8Mbps ADSL service with my ISP, PlusNet (formerly Force9, now owned by BT.) Except I don’t. It’s never run anywhere near 8Mbps. In fact, I thought I was still on a 2Mbps service until I enquired a while ago about an upgrade, only to be told I was already experiencing the thrill of an 8Mbps service.

The promise of this fantastic service sadly has not matched my experience with PlusNet. I have never seen my connection exceeded 1.6Mbps down. It has, in fact, got slower over the last couple of years, to the extent that the heady days of 1.6Mbps downstream seem a dim and distant memory. Every couple of months I’d lose another 200kbps or so from my connection speed. Once in a while I’d open a support ticket to see what my ISP could do to help me about it. Bugger all, is apparently the answer. BT (who own PlusNet) won’t investigate bandwidth issues on 8Mbps lines until the speed is 400kbps or less. Let’s just reiterate that. BT consider an ADSL service that reaches 6% of the advertised capacity to be working properly.

Well, today I noticed my connection crawling slower than ever before. Upon checking my ADSL modem, it reported that it was connected at just 224kbps. Even by BT’s rather pathetic standards this is classified as a fault, so I headed to my ISP’s support pages to log yet another support ticket. They have revised their reporting procedure for speed faults and now I have to run BT’s speed tester. This is a website which one visits but requires Java to be installed. This presented a problem as I run a 64-bit version of Linux and Sun (who make Java) haven’t bothered to make a version of their plugin for 64-bit Linux yet.

Fortunately I was able to find a 32-bit install which I could use, however the java application hangs at 96%. It seems I am unlikely to be able to progress my support request without this test having been run (as BT log the results of the tests on their servers and use it to check whether a line fault should be investigated.) It’s also worth mentioning that the speed tester site is a “stress test” for bandwidth. It times how long it takes to download a given amount of data and uses that to calculate the performance of the line. This sort of test is quite silly as it is easily affected by other applications or network resources using some of the bandwidth. However, the connection speed reported by the modem is always accurate and represents the maximum theoretical throughput the line can give in that session. So really that should be enough proof that the line is too damn slow, especially when the theoretical maximum is so pitiful. Here’s a screenshot from the web interface for my ADSL modem which should clearly communicate the state of my connection to someone who knows what they’re talking about:

That’s right, I have more upstream bandwidth than I do downstream. I decided to plough on with my support request and started to work through the broadband fault checker on the PlusNet website. First of all it asked me which Operating System, checked for known faults on the line and asked me what make of ADSL equipment I am using. All fair enough. It then asked me whether my issue was that I couldn’t connect at all, or whether it was a speed issue. I happily selected “speed”. The next question was whether I had ever been able to connect to my ADSL service. As if I was going to be complaining about the speed of an ADSL service to which I had never connected.

Pressing on, I was asked what date the service had last been fully working. Well, by BT’s metrics presumably sometime in the last week when it dipped below their magic threshold of 5% performance. However I consider the issue to be the gradual and continued degredation of my ADSL service, which started, as near as I can remember, back in September 2006. So I entered 01/09/06 in the date box, using the DD/MM/YY format requested. Next question was “On what date did the problem start?” Well, by logical extension this must be the day after the service was fully working. Surely? Unless I’ve missed some fundamental nuance in the question. So, I entered 02/09/06 as the date the problem started. Fail.

It seems there are some rather large usability flaws in the fault reporter. At this point I was left wondering which dates would be acceptable to PlusNet. Clearly the logical answer isn’t valid. I progressed the month and year values through the ages until I found one which the reporter accepted. Apparently January 2007 is the earliest a problem could have occured. Well, sadly, PlusNet is wrong in this case. The next suggestion was that a virus may be causing the problem. Well, a virus, or more accurately a network worm, might well be responsible for consuming upstream bandwidth and causing performance issues. However a virus wouldn’t cause the line to connect at such a slow speed in the first place. Also, while it is a worthwhile question for Windows users, all the machines in the house run Linux where there is no real virus threat at the moment.

The next step was the tenth of an undisclosed number. It required 3 completed speed tests using the BT speed tester site. You’ll have read already how this failed to work for me, and how, given the session connected as such a slow speed, a stress test of this particular fault is not hugely relevant. So I plumped for the option which said I’d completed the tests, figuring that I can explain the failure to actually run them once I’d opened the ticket. Constant or intermittent fault? Constant. I was able to give my explanation of the BT speed tester site in the futher comments box.

I continue to be amazed at the reliance of companies like PlusNet and BT on shoddy applications like this. I’ve been PlusNet’s customer for 6 years and yet they won’t help me resolve this issue despite their own logs showing a gradually decreasing connection speed. In fact, they seem intent on putting barriers in their customers’ way by using things like Java applications which exclude sections of their customer base. You really, really would think they’d know better. Right now, I’d be prepared to move to any ISP who charges about the same as I’m currently paying and would work with me, BT and whoever else to get my service back to what it was two years ago, because PlusNet don’t seem able to do so.