Jan's Comics

Road warriors - page 10

2013-05-11T04:20:53Z

Road warriors - page 9

2013-05-01T04:29:50Z

Road warriors - page 8

2013-04-21T08:27:36Z

Road warriors - page 7

2013-04-11T04:10:46Z

Road warriors - page 6

2013-04-01T04:05:02Z

Road warriors - page 5

2013-03-25T04:00:17Z

Road warriors - page 4

2013-03-18T04:21:45Z

Road warriors - page 3

2013-03-11T06:20:22Z

Website upgrade

2013-03-05T14:14:33Z

I've upgraded the code that runs tigerknight.com. Actually, 'overhauled' would be a better descriptor, but here's what's new:

The site no longer needs your Paypal account's email to process payments. Situations where the payment goes through but the subscriptions don't get added are now a thing of the past.

The comment form has been simplified, to show whether you're logged in.

The S&S contributor pictures have been moved into their own page.

The new S&S subscription page has replaced the online store.

The list of changes looks short, but under the hood, a lot has happened. I now have a better foundation to add new features (avatars, comment editing) down the line.

Cloc.pl statistics, for the curious:
Old codebase:

-------------------------------------------------------------------------------
Language                     files          blank        comment           code
-------------------------------------------------------------------------------
PHP                            125            302           1358           9101
HTML                            36            159              7           1865
CSS                              2             88              2            451
Javascript                       3              4              2             65
-------------------------------------------------------------------------------
SUM:                           166            553           1369          11482
-------------------------------------------------------------------------------

New codebase:

-------------------------------------------------------------------------------
Language                     files          blank        comment           code
-------------------------------------------------------------------------------
PHP                            108            623           1332           5615
CSS                              3            136             20            754
HTML                             5              0              1             85
Javascript                       2              3              2             32
-------------------------------------------------------------------------------
SUM:                           118            762           1355           6486
-------------------------------------------------------------------------------

That's 44% less code, yet with more functionality (and maintainability) :) Much of it resulted from trying to keep the code DRY.

Deploying the new code to the website was frustratingly funny today. First, I develop my code in two places -- at home, and at work. I brought to work with me, by accident, an incomplete export of the database (it had all the data, but was missing the schema). So I had to run back home with a USB drive and re-export the database, then run back to work. And by run, I mean ride a bus across town and get stuck in traffic jams.

Anyway, once I got back, I uploaded the new files, and the site began throwing errors. I was wondering why the script wouldn't run and was complaining about missing classes. Missing classes mean my autoload function wasn't working, and I was pulling my hair out figuring why that was the case, when it worked perfectly on my development machine. Then it dawned on me:

I miscapitalized a single "c" for one of the file names.

Forehead slap. Code was being uploaded to a Linux server. I developed on Windows. Windows doesn't care about filename capitalization. Linux does. Derp.

Of all the possible scenarios that I predicted would happen, I would have never imagined this to be the one. XD

Road warriors - page 2

2013-03-02T08:04:06Z

Recovering data from an almost dead hard drive

2013-02-27T16:17:36Z

Last weekend, my cousin handed me a hard disk drive which was malfunctioning and needed data recovery. Just for my future reference (and in case it helps anyone else in the same situation), I've documented my process below.

Hardware:

80 GB Maxtor hard disk drive, IDE, in a generic IDE->USB enclosure

Symptoms:

When the USB cable is plugged into a Windows machine, the hard drive begins making loud clicking noises. The access (busy) LED lights up and stays lit for ~45 seconds until the read or write attempt times out or otherwise fails.

Windows cannot recognize the drive. It shows up in Disk Management, but with no partitions

Failed recovery attempts:

Booting the computer with Clonezilla and attempting to image the drive fails

Booting up in Parted Magic, gparted gets stuck trying to read the partition information (access LED turns on solid; drive begins clicking)

Spinrite sees the disk, and can read from it (albeit very slowly.. ~1 MB/sec). Left to run on a level 2 repair operation, it comes upon a bad sector and gets stuck (access LED turns on solid; drive begins clicking, Spinrite freezes)

Diagnosis:

At this point, I knew the drive had bad sectors. The data is still there, but I needed to image the drive while skipping over the bad sectors.

Successful recovery process:

Booted up the computer with SystemRescueCd. Hard drive attached while booting

Figure out where the drives are connected

$ fdisk -l

Drive to store recovered data is on /dev/sdb1, while malfunctioning drive has several partitions -- /dev/sdc1, /dev/sdc5, /dev/sdc3

Mounted a partition on the recovery drive (/dev/sdb1) that's formatted to exFAT

$ mkdir /mnt/recovery

$ mount /dev/sdb1 /mnt/recovery

Made a folder on the recovery drive to store images

$ mkdir /mnt/recovery/recovered

$ cd /mnt/recovery/recovered

Used ddrescue to attempt recovery of each partition

$ ddrescue -d /dev/sdc1 sdc1.img sdc1.log

$ ddrescue -d /dev/sdc5 sdc5.img sdc5.log

$ ddrescue -d /dev/sdc3 sdc3.img sdc3.log

Process takes a few hours. It will get stuck trying to read bad sectors, but will skip over them after some time.

For sectors where it really gets stuck, hit Ctrl+C to abort

Skip over that section (e.g. the 6 GB mark) and continue recovery

$ ddrescue -d -i6G /dev/sdc3 sdc3.img sdc3.log

Go back to the skipped section and rerun ddrescue, but with a sane timeout (3 minutes)

$ ddrescue -d -T3m /dev/sdc3 sdc3.img sdc3.log

ddrescue should have a pretty good image now. Attempt to mount image (it's NTFS)...

$ mount -t ntfs -o ro,loop sdc1.img

... results in an error. First clusters are probably corrupt. Unmount everything and power down the machine

Boot into Windows with the drive containing the recovered images. Install ImDisk, which allows corrupt disk images to be mounted

Mount the image using ImDisk (I mounted it to drive R:). Run chkdsk /f on the image

> chkdsk /f r:

chkdsk manages to fix the MFT, volume bitmap, and other structural metadata. Open up the drive to see if it works

> r:

> dir

Success! Repeat process on the other two images. Then copy the data out using any method

Road warriors - page 1

2013-02-23T05:00:19Z

The next party crashers pic

2013-02-15T07:07:37Z

Want to get into the next S&S contributors' picture? Instead of instigating a mad rush for openings, I am trying out an entry form. I will try to accommodate as many contributors as possible. Some spots are still limited, but you will be able to get in by raffle instead of a first-come first-served basis. Anyhoo, enough babbling... here's the form! Indicate your interest by 28-Feb; the raffle will take place shortly thereafter.

Reader tournament - round 3

2013-02-01T07:51:19Z

The end

2013-01-27T05:00:16Z

Yearbook

2013-01-13T05:01:01Z

Bananas

2013-01-12T05:40:40Z

Cookout

2013-01-06T05:01:00Z

Shipping

2012-12-30T05:00:08Z

Higher math

2012-12-23T05:01:04Z

Lakeside retreat - 4

2012-12-22T05:00:27Z

Cling

2012-12-16T05:00:45Z

Lakeside retreat - 3

2012-12-15T05:00:37Z

Job fair

2012-12-09T05:00:48Z