Category Archives: General rambling

Fedora 8 + fakeraid + second SATA controller = boom

Some more fun on Fedora… I’ve just upgraded my desktop PC to Fedora 8. Now, this machine has some oddities — namely, two separate SATA controllers (a VIA that’s built into the chipset and a Promise FasTrak that’s a separate chip on the motherboard). The problem with this arrangement is that the scan order depends on which module is loaded first…
So if the VIA controller is scanned first, /dev/sda ends up being the 80GB swapdrive, and /dev/sdb and /dev/sdc are the two RAID1’d drives on the Promise fakeraid. If the Promise controller is scanned first, reverse the order…

So the first time I boot the machine, I see something like this:

device-mapper: table: 253:0: mirror: Device lookup failure

And the machine crashes hard.

To fix this, disconnect the drive from the VIA controller and boot into Rescue mode. Chroot to /mnt/sysimage, then type “su -” to set up the shell. Then rebuild the initrd:

mkinitrd -v -f /boot/initrd-VERSION.img VERSION

Where VERSION is the current running kernel version (cat /proc/version will tell you this -- in my case it was 2.6.23.9-85.fc8). Once this completes, reboot and make sure F8 boots.

Once you're sure F8 is bootable (run through Firstboot before you reboot!), try reconnecting the other SATA disc. With a bit of luck, the machine should remain bootable. Voila, problem solved.

DHL? DH-hell!

Oh I do so love DHL.

The other day I ordered a Tektronix TDS2024B storage oscilloscope. It’s a nice little thing, nice enough that it had to be special ordered in. So I got a call on Monday from the distributor – “Hi, we got it in a few weeks earlier than expected, do you want us to send it now?”

My reply was a rather predictable “Hell yeah!”

So fast forward to today. The DHL truckmonkey turned up, shoved a card through the door and *RAN* back to his van.

I kid you not. Ran. This guy really didn’t want to deliver my parcel.

Calls to customer service go either unanswered, or they hang up on you. Hello, fuck off, *click*. Such nice customer service. Even arranging a redelivery is an exercise in utter futility. Fifteen minutes on the phone and I finally get put through to the automated redelivery thing.

“Your delivery has been rescheduled for.. Wednesday. *click*”

Seriously, there is no way in hell I’m ever going to do business with any company that deals with DH-hell. These assclowns can go to hell.

Had I been able to speak to an actual person at DHL and get this mess straightened out, this post would probably have been a “they screwed up but fixed it”. But no, by making themselves completely uncontactable, they’ve joined the elite ranks of my ShitList, along with good’ol’ ANC (who left £600 worth of computer equipment at a house two and a half miles away then denied all knowledge and destroying the tracking data).

DHL: The New Name For Securicor-Omega — Same Shit, Different Name.

File under: “Why do I bother?”

Another day, another dead hard drive. And another IBM, too, why am I not surprised.
This piece of crap is a 20GB Deskstar 34GXP that had an offline copy of my magazine article tracker installed. Basically, I rigged up some scripts to read the contents of the Elektor CDs and dump them into a MySQL database, with the PDFs and whatnot stored on disc.

You’d think after three HDD failures I’d learn to back up my data… Maybe now I’ve got TrueImage reinstalled I’ll set it to do automatic incremental dumps to the NAS…

Fine! I’ll build my *own* kernel…

Well, my PVR box is set up. It’s running Mythdora 4, with some customisations:

1. **Patched, older kernel** – the stock Fedora Core 6 2.6.18 kernel with the GRML kernel patches (from [GRML’s Mercurial VCS](http://hg.grml.org/grml-kernel/archive/3abf902cffd4.tar.bz2) in the /2.6.18 patches; you want 400?_net-r8169*.patch)
2. **HVR3000 driver modules** – Steve Toth’s HVR3000 driver kit, from [LinuxTV.org’s Mercurial VCS](http://www.linuxtv.org/hg/~stoth/hvr3000) with one small patch — in v4l/config-compat.h, make the line `#include ` read `#include ` instead. The analogue tuner *still* doesn’t work in MythTV though. But I don’t care because DVB seems to work fine.
3. **New ATI FGLRX drivers** – The latest version of the ATI FireGL/Radeon For Linux driver kit. Mythdora’s driver is *old*.
4. **Rebuilt IVTV and LIRC RPMs** – the standard AtRPMs RPMS, rebuilt for the new kernel
5. **A few bootscript patches** – to get *cx88_dvb* to load on boot, and to set up a few symlinks in /dev to allow Myth to access the DVB-S satellite tuner

RPMs coming soon, right after I get all the mess sorted out. The HVR3000 is a real bastard to get working on Linux, that’s for sure, and the analogue tuner seems to misbehave a lot (MythTV certainly didn’t find any channels)…

“More at ten…”

Yes, I am going to write a HOWTO. Not a good one, but a HOWTO anyway.

Another job done..

I’ve just spent three frickin’ hours up a ladder. Two in reasonable light, one in near pitch darkness.

What was I doing?

Rewiring the satellite dish.

I’m covered in various lacerations from all manner of prickly, sharp, pointy objects (mostly plants) and I have no thumbprint left on my right thumb, courtesy of a pair of screw-fit F connectors.

But the Sky dish works again. Well, sort of at least. 75% signal strength, 50% signal quality (it used to be 75/75). But that can be fixed by going up the ladder again tomorrow night and realigning the dish *properly* — the elevation is out a degree or so, but the azimuth is right on, going by my satfinder’s frantic beeping.

More fun for tomorrow at least.

… And nature makes a better opportunist.

Once again, one of my great plans has been foiled.

By a rodent.

For the past nine years, the house has been fitted with a satellite receiver of some make or description. First it was a Sky analogue system on the ASTRA-1 “19.2 East” constellation, then a SkyDigital system on the ASTRA-2 “28.2 East” constellation. After a failure put down to what I assumed was a dish alignment issue, the Digibox was disconnected and shelved.

So this morning I figured I’d shove a DVB-S card in my PC and see if the dish really was misaligned. Up the ladder I went, spanner in hand, ready to loosen the bolts and realign it.

A quarter-inch to the left. “Nothing”. A quarter-inch to the right. “Nothing”. A finger-push on the bottom. “Signal zero, quality zero.”

So I figured I’d eliminate the dish – I disconnected the cable at both ends and hooked a 9V battery to it on the house-side of the cable. Up the ladder I went, multimeter in tow. 2.3V. When it worked. Cable shorted inside… cable reading 470k outside.

All this, of course, pointed to a damaged cable. The cable is at ground level. I put two and two together, got five, and figured “what if a rat’s chewed at it?” Seeing as we’ve had rats in the garden before (though not near the house – mostly around the compost heap at the back), and that the cable was near to the ground, I figured it couldn’t hurt to check. I pulled out my cheap-as-chips LED torch, and climbed down from the ladder, tracing the cable along the wall. Half-way between the dish and the inlet point, I found what I least wanted to.

Four bitemarks in the cable.

One just grazing the surface, two half-way through the inner core of the wire, and one massive bite right through the cable. Said cable was as black as pitch (the core is usually white). So it’s pretty safe to assume that it’s not only ratbitten, but also pretty soaked inside as well. Which explains the zero signal report at the box.

So now I need 20 metres of decent CAI-approved satellite cable (read: CT100 or WF100, preferably the latter), new connectors, and a roll of self-amalgamating tape. And then I need to go up the ladder, re-wire the dish, and re-align it. Thankfully, only in the horizontal direction. Hmm, maybe I could point it at one of the HOTBIRD satellites instead. No! Must.. have.. BBC HD……

And the rats won’t win a second time. Because the cable’s going to run across, then down. Which leaves it open to avian predators. Though there are considerably less avians with a taste for PVC-sheathed metal than there are ground- and sewer-dwelling rodents with a taste for the same.

Of course, I could have grabbed my soldering iron and spliced some new cable in, which would have got the dish working well enough to align it, but that would have been the easy way out. And no doubt there are plenty more chew marks further along the cable…

Rats. Nature’s ultimate opportunists. Though maybe with the possible exception of dogs…

New theme time

Yep, it’s /New Theme Time/. I was getting bored of the default WP theme, so I’ve replaced it with [“Gray Gets Green 1.0” by Fred Banuelos](http://themes.wordpress.net/columns/2-columns/987/gray-gets-green-10/). I’ve also upgraded to WordPress 2.2 (about time too) and [PHP Markdown Extra](http://www.michelf.com/projects/php-markdown/) 1.1.4. More toys, yay.

Yes, the Castlecore server fell over again today. I did an Apache/PHP upgrade and Suexec went mad. A force-rebuild seems to have fixed it. Now if only I could figure out why upcp keeps crashing…

I’ve sorted out my new HDD array too – a two-drive RAID-1 with a pair of 500GB SATA Seagate Barracudas. The Promise ATA-RAID controller (read: *fakeraid* controller) works pretty decently, and the machine is **much** quieter now.

*[fakeraid]: A RAID controller that, while providing RAID functionality, has no hardware-based RAID acceleration.

Fate is conspiring against me.

Yep, I’m pretty sure the fates are laughing at me as I write this. I just started doing some work on a project, and now my G: drive (the 20GB I use as an overspill drive) has packed in. The famous Click of Death in fact.

Here’s a little sidenote for all the people out there that thought “nobody got fired for buying IBM”. That may be true, but lots of people got REALLY PISSED OFF with IBM when the GXP series drives started failing in droves.

Thanks IBM, now my archive of EDN and Circuit Cellar articles has gone to the great bit bucket in the sky. I think I’m going to go buy a couple of 400GB Seagates and build a RAID1 array. Let’s see a single drive failure kill THAT!

Mr. McKittrick, after careful consideration…

… I’ve come to the conclusion that your stock management system SUCKS…

Yep, spent most of today tracking down components. Now I’m planning to rewrite the shoddy little spreadsheet that manages and tracks my stocks of electronic components into a nice little web-based thing that isn’t so shoddy.

The plan is to have something that tracks manufacturer details, supplier details and the details of the part itself, including parametric data (e.g. value, voltage rating, polarity, category, pin spacing, OrCAD footprint, KiCAD footprint, etc.). That way I can see what I’ve got, when I got it, and where I got it, and also what it is, what it does, what its values are and where it’ll fit on a board.

Then maybe I’ll add something to track BOMs (bill of materials – aka parts lists/parts manifests) and component requirements, and what components are being used for (“RESISTOR 4k7: 4 units used for production of ‘USB mood light'”, “4 units returned to stock: USB mood light proto dismantled, returned to GOOD stock”)

Just a fun idea that might turn into something else. The web interface may well be view only, but I intend to use SOCI and wxWidgets to build a read/write mode client application that communicates directly with the DB server. The advantage being I can then combine that code with my Universal Multiplatform Makefile Template and create something that I can compile and run on Windows or Linux (or OSX if I ever go over to the Mac Side).

All adds to the fun though.

Redundancy at 20, such fun

So the axe has finally fallen. Looks like Chris Langley (ex-Dixons brain-damaged upper-management numpty that he is) has finally managed to run another formerly-successful business into the ground.

81 Jessops stores closing, including my own. Fairly new store, instore lab, great team, and we’re all for the chop, along with staff at three other nearby stores (one in Huddersfield and two in Leeds). Jammy gits in York get to keep all theirs.

What’s even worse is that the bastards at Head Office sent a press release to the BBC before they told the managers. I woke up on Thursday morning to the BBC Radio 2 news, second item was ‘Jessops are closing 81 stores with the loss of 550 jobs’. Great.

So my ‘research fund’ is basically gone until mid-July, when I start working for Echostar – writing software for set-top boxes no less. I just hope my colleagues can find something.. they’re a great team and it’s been fun working with them for the past 9 months.

The big problem is that having swallowed up all the independent brick-and-mortar photography dealers, there really isn’t anyone left besides Bradford Camera Exchange and Bligh Cameras. Well, there’s Janet Green’s in Halifax, but they’re even more expensive than Jessops (and their customer service leaves a LOT to be desired)!

“The definition of a fool is someone who does the same thing twice, and expects a different result the second time around”

I could probably say more, but at this point in time I really can’t be arsed. Suffice it to say I’m going to make the most of my staff discount while I’ve still got it – I hear Huddersfield have some Fuji Reala that isn’t selling…