Archive for the 'Computers' Category

Upgrade

Tuesday, April 28th, 2015

Having learnt my lesson about staying up to date so as to not get stuck with a hard-to-upgrade install, I set about upgrading my Debian “server” to Jessie. How about that for ease?

Add a few lines to a config file (etc/apt/sources.list, to tell it where the updates are), apt-get update, apt-get upgrade (a cautious intermediate upgrade), apt-get dist-upgrade, answer a few questions, and off it goes, with 1 reboot (for a kernel change), and all services working with short interruptions only, and remotely over ssh, and only a couple of hundred MB download to have the latest release- running on antique hardware.

That’s how an upgrade should work.

A Bad Apple

Thursday, February 12th, 2015

Andy, Ross and BrownhillsBob may be expecting me to have a pop at Apple here (as is my way), but I’m not going to, or at least only a brief whinge, with the main target (again) being idiots that claim to know a product, but don’t.

Those of you with long memories may remember this lengthy rant. A short swipe at OSX, and a big load rant at fucking Symantec (as a colleague commented, is there *any* company Symantec have bought and not fucked up the product?), and a big rant at fuckwits who don’t understand what they are being paid to do.

The OSX server mentioned in that rant failed. To be completely fair, it’s worked completely reliably for six years now, which is impressive. So I’m not going to complain, and it was clearly hardware that was bost.

A few attempts by colleagues and myself to resurrect it failed, so we called the support company (sadly the same fuckwits from the story back in 2009). They wander in, (bringing a manual, which sets off alarm bells- I’d expect a field engineer to not need it….) say the server’s not supported by Fruitco any more, that parts are a nightmare, briefly try (and fail to get) Target Disk Mode, (which, I note, doesn’t work with disks attached to a hardware RAID card, so wouldn’t have helped), shrug a bit, say that our diagnosis of a buggered RAID card might be right or maybe it might be the logic board (as there’s little more than those 2 fucking boards in it, this is hardly advanced diagnosis, and leave.

At this point, I begin to wonder what we’re paying the fuckers for, and I start restoring the files to the only place we have a Backup Exec agent and 1TB of spare storage: a Windows Server 2003 box. Most of the data restores, but some recent work is lost as it didn’t make the tapes (the Mac workstations being too old for Time Machine), and some initially didn’t restore due to file naming incompatibilities (take it from me, anyone using mixed operating systems (our backup is Windows-based) should read this, and this: most of the restrictions are with Windows, but you never know what OS you may be sharing files with. I personally think it all went downhill once spaces were allowed in filenames :-), and here’s my brief whinge: I know the limitation is Windows, but allowing “:” and “\” in a filename is just fucking wrong, and supporting your hardware a bit longer would be nice.

Now then, what to do? The users are (mostly) working again. First of all, the original support co is ditched. We call another supplier, and the difference is incredible: engineer arrives, asks all the right questions, listens to what diagnostic steps we’ve tried, sounds like he knows what he’s talking about, diagnoses a logic board failure, and offers to take the server back to the workshop to attempt recovery of the recent data for a very modest fee. Guess who’ll be getting the support contract, and potentially an order for new machines in a while?

It also makes me think I should have taken better note of the warning signs six years ago: these people claim to be supporting us (and originally claimed to know the product, but, as is so often the case, don’t. I’m glad to say that I didn’t arrange their involvement.

Debian Again

Thursday, February 5th, 2015

The “server” I built back in 2008 suffered a nasty accident at my own fair hands: a cack-handed attempt to patch the GHOST vulnerability, compounded by a previous, similarly cack-handed (though succesfull) approach to patching bash previously taught me a hard lesson in package managment on Debian-based Linux: Don’t fuck with the rules, and when you see a confirmation that says you must type

Yes, do as I say!

or the operation stops should be taken as n indicator it’s about to break. It did.

The handy bit was that the system stayed up, but I couldn’t start or stop services. This wasn’t going to survive a restart, so I got the data and config files off with SFTP, IMAP’d the email into a folder on mylaptop, and shut it down. Then work got in the way and I had to rely on gmail for a few days.

Rebuild was a bit fraught: Ubuntu Server no longer comes with a suitable kernel for the very old Celeron laptop i use as a low-power server, and I thought I’d have to use a desktop variant, but returning to Debian provided a install CD that worked, and after a few false starts with postfix config, it’s up and going again. The lesson learnt here is to be cautious (don’t break dependencies), but not too cautious (if I’d kept Ubuntu up to date, the 2008 build would have been upgraded by now, and \i wouldn’t have had to bodge the patches).

Volumio

Wednesday, January 14th, 2015

I already mentioned my plans for a media streaming server with my Raspberry Pi, and finally got round to it: a friend donated an external disk enclosure that took a pair of 1TB SATA disks, and presented them as a 1TB RAID 1 volume over USB. A cheap USB hub, a case for the Pi, a £2.49 USB wifi dongle, and a quick download of Volumio (a modified Raspbian image) and all the bits are present, fitting them together was pretty simple, and I have a working media server with great sound quality, that uses little power and is completely hidden from view: all the hardware worked, with the only tweaking being a quick edit of /etc/network/interfaces to set a fixed IP on the wireless network.

Volumio is cool: it’s like IPCop in that it’s an open-source appliance based on Linux with a web interface to configure it and use it, but you can delve “under the hood” with ssh. It uses the mpd server, and presents itself on the network via SCP or a SAMBA (Windows network) share for uploads, and advertises on Airplay or DLNA. You can control it with a wide range of clients for all sorts of devices as well as the web interface, and it just found my DAC with no tinkering, and the sound from a FLAC file is as good as the original CD, even with the Pi’s limited horsepower.

I have a good amount of ripping to do…..

Into the Digital Age

Sunday, December 28th, 2014

Regular readers will know that I’ve got some prejudices about audio: for years I didn’t have an MP3 player, eventually relenting, although I still don’t do actual MP3s, and buy music almost exclusively on CD, though it has to be said, my views on downloadable music 10 years ago are starting to be disproved: MP3 at a highish bitrate (which is more practical with increasing bandwidth and storage) is good enough for most people, on most systems, at most times, and lossless formats are becoming more common, especially since Fruitco introduced their lossless format (though of course, they should have introduced it as an open format…).

One thing* has kept me away from using a computer to play music in the house: the analogue outputs of most consumer PCs (and I’m including Fruitco in this) hardware is a bit ropey- but then again, it was never intended for high-quality audio.

Enter the DACMagic. It’s a proper (though very small) hifi component with TOSLink, S/P DIF and USB inputs, and it’ll do the high sample rates that may not be neccesary, but more importantly, it’s a decent DAC chip with the compromise pushed a bit towards quality, and some initial testing sounds as good as the CD with a FLAC file (and, pleasingly, the device was recognised and working within seconds on Ubuntu).

The plan now is a Raspberry Pi and Volumino: the Pi’s analogue audio output is particularly compromised (hardly surprising given it’s a £35 computer) and the ‘proper’ stereo doesn’t have HDMI. There are cheaper ways to get better Pi audio with a Wolfson DAC, but as a bonus, the DACMagic’s inputs can link to my existing CD player too; a respectable but budget Marantz, and also, it comes in a nice black case that looks decent next to the other gear: initial comparisons sound like the DACMagic has improved sound here too, but that could be the confirmation bias- I’ve just bought it, so it /has/ to give an improvement :-).

*OK, two things. I’m an awful luddite, it would seem.

Surface Treatment

Saturday, September 6th, 2014

A few days ago, we got a Microsoft Surface Pro 3 at work. It’s not a bad machine: it’s a Laplet: a hybrid laptop/tablet, and it works well, if we excuse it for Windows 8- the hardware is nice, thin, light, and i7 versions are quick, so it’s a good fit for the very mobile staff that
will be using it.

I remain convinced that Win 8 is a bastardisation of touch-screen tablet OS and a desktop OS that feels like an unholy marriage, though I’m hating it less as I get used to it.

What really creates a whinge is this little stroke of genius, which caused a support call and much fannying around testing chargers this week.

You can see the product launch meeting now:

Dilbert

Yes, Microsoft launched a device, launched a dock for it at the same time (we got the dock a day or two after the device itself), and managed to make the two not work together at launch. Cue a large loss of faith in what should be a good product.

*facepalm*.

You see this a lot with technology, and come to that, with poorly managed processes outside of tech:

1. Decide on arbitary launch date and fix everything to that.
2. Skimp on the preparation/testing, or ignore the problems.
3. Wonder why it’s all gone wrong.

The result is pretty much as you’d expect; you look inept…

Dirty Boy

Tuesday, July 29th, 2014

I’ve gone on here before about how web filtering is wrong and doesn’t work properly, and how the bigger the scale, the harder it is.

We’ve also seen that, according to an Ofcom report (PDF, 1.1MB) customers have greeted the filters with rejection.

That’s quite gratifying, I think. People are being actively prompted to allow censorship, and are rejecting it. Of course, that the tech required is now in place will make it easier to do more packet inspection should law (or other means) request it…

Here’s the Open Rights Group‘s take on it, the approach is humourous, but the message is serious.

If you think this won’t happen, try the Scunthorpe Problem for size.

I’m personally of the opinion that an ISP should do one thing: provide the infrastructure to route packets to the internet, and maybe a few basic services (like DNS, SMTP etc). You might note that the sponsors of that video refuse to offer a filtered connection, something they’re to be congratulated on.

If, like me, you want to defend an open, uncensored Internet with reasonably privacy, then consider joining the Open Rights Group or the Electronic Frontier Foundation. Note that ORG is a UK organisation, EFF is US-based.

Loosely Comnnected

Saturday, July 26th, 2014

This is quite a wierd one: some time ago at a company I work for sometimes, a colleague tried to replace some old 15″ LCD monitors with shiny new 19″ ones, to be confronted by extereme flickering. I had a look, tried the monitors with my laptop, and got a flicker-free picture. I made sure the leads weren’t too close to mains cable, but no change.

We assumed some incompatibility with the (elderly) PCs, and another colleague changed the PCs recently. In the course of doing so, he discovered the real cause. A power lead- just a normal BS1363IEC C13 (colloquially known as a kettle lead), but, tellingly, with a rewireable BS1363 plug, not a moulded one. Remove the lead, problem stops. This lead was connecting one of the PCs that was working perfectly well, and flicker-free with the 15″ monitor.

I looked at the lead the next day:

The culprit: a badly fitted plug.

The culprit: a badly fitted plug.

and it seemed kind of OK at a glance, though that neutral lead should have been cut shorter.

What did turn out to be wrong was every terminal was loose: loose enough to turn by hand, so I presume that the intermittent connection caused enough noise to upset the new monitor, but not the old one. Disturbingly, this lead had passed a current PAT test, when potentially it’s a fire hazard: loose connections can overheat.

I don’t know if the connections had worked loose (which is one reason why connections in screw terminals should not be tinned with solder) or just sloppily fitted in the first case. The plug did rattle when shaken, but it would do that even with tight terminals, as the pins have a bit of play in the housing. Full marks to my colleague for spotting an obscure fault.

WYSINWYG

Sunday, June 22nd, 2014

Most people are familiar with WYSIWYG- What You See Is What You Get- a computer user interace that displays things in a format that fairly accurately displays on screen what the final output will be, so that (as a simple example) rather than a bit of code:


<b>this text is bold</b>

you get

this text is bold

I’ve spent several hours of my life recently trying to find out why a program I have to use daily was refusing to email people. Here’s the UI:

Skeuomorphic twice over? A web app emulates a phone emulating a slider switch, badly.

Skeumorphic twice over? A web app emulates a phone emulating a slider switch, badly.

.

You’ll notice the option to email two people, controlled by sliders. These are a skeuomorph: soemthing that icorporates design features of something it emulates- in this case, a slide switch. In fact, it’s a double skeuomorph: it’s a web interface impersonationg a smartphone impersonating a slide switch.

I’ve got two problems with it. Firstly it’s unneccesary frippery and animation, and secondly, it plain doesn’t work. It’s distinctly What You See Is Not What You Get. I’ll grant you there’s a certain amount of PEBCAK here on my part, but the control is broken.

If you click on the left-hand side of the control, and swipe accross, like you might with the control it imitates on a smartphone, the control changes to YES. It does the same if you click the right-hand end, or if you click and drag, keeping within the boundaries of the control. The difference is that if you use the first method, like I did, it shows YES, but registers NO to the back-end software, and gives you a several-hour troubleshooting session to work out why the email didn’t send. An older version of the software has a simple check-box here, and I suspect that this is a simple case of layering a bit of wankery over the top for effect…

Opinions differ on skeuomorphs: some consider them to be problematic, and some think them great, for the same reason: they imitate familiar technology, and so either make people confused, or make then feel at home. This one definitely left me confused.

Openretch

Wednesday, April 16th, 2014

Can anyone tell me how BT’s line providing division, Openretch Openreach survive?

The company I work for provides network services to varying people and organisations, and this means dealing with telecoms providers, and that almost certainly means the loose collective of fuckwits we know and love.

So far this week, I’ve had:

1) In response to a circuit order in a building that is partially let out (and where the BT duct enters via the let out area, and is a retail shop) “can we come tomorrow”.

2) In response to an order made 3 months ago, in a central Birmingham hotel, with the clear stipulation “you must make an advance appointment, the circuit is required by 17/4/14″, an “engineer” arrives today, has to wait a few minutes and is told “no, sorry, the room is in use, come back tomorrow”, and says “no”.

Said engineer was told “sorry, you have to. This is required by 1pm tomorrow, and was ordered 3 months ago. It cost a metric shitload of cash, and you haven’t called in advance, like we told you to, and you do this *every time* we request this. The room is available afer 5:30, or anytime tomorrow”

*shaking of head*

This barely describes the quantity of fucks the engineer didn’t give. “Resourcing”, he said. “not gonna happen”. “we’re only supposed to wait 15 minutes, and I’ve been 20″. The guy was, to be fair, a master of fuck not giving.

and not a single fuck was given.

Behold!

He departed. I called my colleague, among whom’s many talents are shouting at BT (and personally, I think it would be worth his salary just for that). He did so. Our circuit should be active tomorrow AM. We have a reference and everything. I will not hold my breath.

Honestly, if they weren’t still a virtual monopoly, they’d be fucked. I’m very thankful to the abilities of my colleagues, and still wondering what shape BT would be in had they not inherited a state-owned monopoly, but thanking my lucky stars I don’t have do deal with Cable and Hopeless Wireless any more, because they had sufficent sense to disappear.

[edit]

The engineer (the same one) came back the next day, and it worked….


This blog is protected by Spam Karma 2: 38953 Spams eaten and counting...