wally’s blog

Is it ironic that this Youtube video is refused (in the UK) due to “copyright”?

(It’s the IT Crowd piracy parody)

Disclaimer: I’m certainly no expert in the area of data recovery, and I will not take any responsibility for damage of you, your or anyone else’s hardware/data caused by following these ideas. This entry is nothing more than a diary of what I did for my and other people’s future interest or reference.

Recently I received a Windows XP machine which “wouldn’t start up”. The BIOS would simply output “A error reading disk drive” after the POST. Note this isn’t the same error as “Insert system disk” etc.

I proceeded to check the obvious, trying every boot order and boot option in the BIOS, all without avail. I then booted Knoppix and attempted to mount the data partition (what was C: in Windows), without success.

Important note: I made two stupid mistakes here. Firstly, I tried to initially mount it in Konqueror, and secondly I tried to mount it read/write. If you’re suspicious of a device’s health, mount it readonly and mount it via the terminal. It’ll possibly save your data and save you some wasted time.

I then tried to mount it from a terminal, but I received a variety of errors regarding DMA read failures. I proceeded to disable DMA (K Menu/KNOPPIX/Utilities/Harddisk/CD/DVD DMA Acceleration) and tried again. Still read errors.

At this point I plugged in an external 500GB hard drive (USB) and started an ntfsclone with the command

ntfsclone -o /mnt/sda1/hda1.img -s --rescue /dev/hda1

Whether due to the computer’s USB speed speed or (more likely) the HD’s state, it took about 9 hours to copy 10GB of data off. It unearthed several hundred unreadable sectors in the process.

It’s important to note at this point (if you’re not familiar with the NTFS tools) that the “-s” switch for ntfsclone causes it only copy parts of the filesystem which are actually in use. That means it does not do the same as:

dd if=/dev/hda1 of=/mnt/sda1/hda1.img conv=noerror,sync bs=8k

I’ll probably go down the same route as before, using my Autopsy script to get off what good sectors I can. But the above commands and flags might just make some else’s life a little easier.

It’s been on my mind for some time now … possibly due to my work (where I am interacting with “technical” people who even themselves struggle) - there must surely be a “near perfect” home router design, something which anyone who has used the Internet on their home computer can install and configure.

I currently believe that the “perfect” design must be a simple design, and have a simple interface.  It must rely on doing everything it can, itself.  It must not ask any questions that don’t need answering.  It must only include features that the home user would want - no VPN, no VLANs, no PPoE, and so on.

So in my mind the device looks something like a WRT54GL.  It has 1 “Internet” port, and 4 “other” ports.  It has wireless antenna.  And it has a power LED.

When you plug the device in, it offers DHCP on every interface - that’s WiFi, LAN and WAN.  The user opens “the Internet” and trys to browse, perhaps the documentation suggests going to “router.lan”.  Dependent on which interface they’ve connected through, a webpage will be displayed welcoming them to their “network”.  If they’ve plugged themselves into the WAN port - the page will ask them to connect using a different pot, because this is for the “Internet”.  For those using WiFI - a congratulations your Wireless network is working!

Once they’ve “OK’ed” the WiFi welcome or changed to a LAN port, a Wizard will walk them through setting a wireless network name and setting a passphrase.  Then it’ll try and connect to the Internet, using DHCP on the WAN port (the DHCP server is now disabled on that interface).

The router hosts it’s own DNS server - directing “router.lan” to itself.  Until the Wizard is finished, all DNS requests will point to itself, enforcing the user to finish before getting online.  The Wizard will ask for a “computer name”, allowing the user to setup DNS records for each device on their network.  Every computer’s MAC is saved for DHCP - ensuring each device gets the same IP every time.

The network would be in the 10.0.0.x range, possibly using a /16 mask to allow more than 253 devices to ever be connected (remember we assign an IP permenantly to each device).  This plus lifetime IP assignments should make it easier for non-techies to operate their network … “ten dot zero dot zero dot five” I believe is  a lot easier than 192.168.0.5.  And everything keeps the same IP, so little Jimmy’s netbook is always going to be 10.0.0.8.  With the “computer names” (DNS entries - possibly automagical via NetBIOS for the most part) make connecting between devices even easier still, i.e. wii.lan and netbook.lan etc.  We also minimize the likelyhood of conflicting with our modem device.

Finally there are “additional” options - not included in the wizard - allowing “Advert blocking” and “Website logging” etc.

The router should attempt DHCPing the WAN port - and only in the situation that DHCP fails should it provide the option to enter some details manually.  Home users don’t (and shouldn’t) have to care about their LAN DHCP pool or DMZs.  The router should offer some advice, allowing you to select what “kind” of device various clients are - for example “10.0.0.14 (XBOX 360) connected 12m 34s” - allowing the router to offer suggested port forwards (or just doing it with basic consent?), and potentially performing a portmap scan for “servers” to help the user understand what things they might want to allow (like RDP or HTTP).

Finally some access restrictions might be good, giving the option to stop access late at night, and of course some enforced QoS.

So the router is going to make a lot of assumptions - but all in the name of simplicty.

Kamikazi developers anyone?

On that note: I be feeling lucky!

If you’re an Essex University student (i.e. you have a valid Essex login) and you’re tired of having to enter campus for access to the Apple Higher Education store, you’ll probably be pleased to know you can do it from home.

If you’re running Linux/Unix (including OSX), you’ve already got everything you need.  Windows users will need to grab Putty and an X server.

Users of real operating systems will just have to open a terminal and enter:

ssh username@unix4.essex.ac.uk -X
Password: [enter password]
$ firefox

Windows users will still be downloading and installing the needed tools at this point, when they’re eventually ready, they’ll have to run Putty, enter “unix4.essex.ac.uk as the host, enter their username and enable X forwarding. Hopefully, if your X server is running, typing “firefox” at the prompt (after your password) will bring a firefox window to your desktop.

Now this firefox window will be slow, because it’s running on the University’s server. So be patient and use it only for what you need. As it’s running at the University, it’ll have no problem accessing the Apple Higher Education store.

Seems Google is indexing yet more information from the net, now you can search for movie times. Woo!

So I promised some updates… well here goes. I’ve been spending the last couple of weeks working through a rather (and increasingly) large list of “to-dos” down the church, everything from replacing maintained light strips to cleaning the windows, and not forgetting producing hard backups of all the server filesystems. It’s good fun to do, (well I’m doing it voluntarily so duh!), satisfying to say the least, but those no-brainer moments really do hurt the head.

I’ve also got a side project on at the moment; dad pointed out a few weeks back that the church’s new website has no hit counter. Now personally I thought hit counters were a bit 1990-ish, but then I am just one. I gave it some lengthy thought and decided that a hit counter is a nice idea, if rather than just being abstractly placed at the footer of each page (and while doing so displaying absolutely nothing useful*) we could have a “stats” page which shows some pretty graphs of how much “load” the website has been under during the past 24 hours. Enter RRDTool!

So far I’ve not had too much success, mostly because I’ve only been working on it for a few minutes at a time and partly because I have absolutely no interest in reading the manuals (I am a bloke after all). This week I’ll have to sit down and have a good read into it, and then try to knock up some scripts to feed it some data, rather than the other way around.

The thought process continued, and I started thinking about other “interesting” statistics that I could display. Some of them that came to mind in particular were:

• Bandwidth usage
• System temperature
• CPU utilization
• Mail server load
• Heating graph (remember hedwig controls the church heating)
• MySQL load

Well the list could go on and on, but I’ll cut it short there. Hypothetically these shouldn’t be too hard to implement, MRTG already does a good job of several, and some hacky perl scripts to filter out the number of hits in Apache, Squid and Courier’s log files should do three others. MySQL and the heating program will need a bit more thinking about however.

Once again I’ve gone off onto some geeky ramble, so I’ll leave you with some totally non-techy stuff to smile over. We’ve pretty much finished the conservatory off at home, Michael, Janice, Miriam and Daniel set off toward Brazil once more this morning (I’ve been up since 3.20am…), Suzanne’s Micra is so getting go-faster-stripes and the flat I’m (DV) moving into this upcoming academic year has just gotten a new bath, fridge and oven; with a ceramic top! (the oven that is). Seems like I might be getting the same luxury apartment as before

Been an interesting one this year… I’ll blog more later

Lets just say 60+mph winds are not fun when camping.

For anyone following this blog (is there anyone? lol), the moving of the Debian install worked without hitches (other than those covered in the blog), I did have to turn off DMA access to the replacement HD however, but I think that was purely because it was a really old school drive.

The Squid proxy is up and running, it took me long enough, mostly due to my complete hate of having to read documents (I really cannot read…), you can visit for the time being at http://www.plfc.org.uk:81/, once I’m convinced Apache redirects etc work properly (behind) it, then I’ll move it to the standard port 80. Watch that space. At the bottom of all pages the time is printed, just in case you’re interested to see when the cache was made.

And finally, my computer is back and running after a near death experience. Phew. It’d been randomly shutting off for sometime now, particularly on hot days. Naturally I assumed it was the CPU’s temperature, but 68 degrees C didn’t seem that high to me (considering many people claim to be running at 80-90). I did however remove the heat sink, give it a thorough clean (amazing how much fluff builds up in there!) and got some good ol’ (unbranded) heat paste / grease from Steve (Computer Scene over in Mildenhall. Go and say hi from me.) To give you some idea, the machine was (kind of ) powering off at 68 degrees C and idling at 64. I’m now at 50 degrees idle and no more than 54 when letting Nexus bots blow each other up. Problem solved I think

Update: came in this morning, much much cooler as I apparently left a window open by mistake… the temperature’s been near idling (with a GL screensaver) at 42 degrees C. Mmm mm!