Diary of a geek

September 2004
Mon Tue Wed Thu Fri Sat Sun

Andrew Pollock


Other people's blogs


RSS feed

Contact me

JavaScript required

Thursday, 30 September 2004

Redhat gets on the conference bandwagon

The only thing I'm surprised about is how long it took them to get on the bandwagon. I have to wonder how sustainable this many Linux/FOSS conferences are...

[03:14] [lca] [permalink]

Monday, 27 September 2004

Dude, don't reinvent the wheel...

(Unless you've got nothing better to do of course, or it's fun to do it anyway)

Mike, take a look at smstools, it may well be just the ticket...

[01:29] [tech] [permalink]

Sunday, 26 September 2004

Hartley Challenge

Since I've scratched the coding itch, it's only gotten worse. Sarah is taking part in the Hartley Ability Cycle Challenge in November, which is a fundraising activity. I knocked up a webpage for keeping track of how much money they've raised.

It took a bit of lateral thinking to get JpGraph to produce the goods, but it wasn't that difficult in the end. JpGraph is truly awesome stuff. It's just a shame it's got a dodgey dual license that precludes its inclusion (well a more current version) in Debian.

[04:23] [code] [permalink]

Saturday, 25 September 2004

Engagement central

Dunno what it is about brothers getting engaged, but on Friday night I got an SMS from my brother saying he'd popped the question to his girlfriend. Last night, Sarah's brother popped the question to his girlfriend, and then turned his son's first birthday party into an impromptu wedding for good measure.

[16:48] [life] [permalink]

Thursday, 23 September 2004

September meeting

Tonight Richard Alexander from Cougar Computers gave a talk about a bid Cougar has made for an RFP to provide a cluster for APAC.

The solution Cougar has put forward is AMD Opteron based, and it was interesting to get an overview of how the Opteron SMP architecture works.

Over pizza, Michael asked if I was planning on open sourcing my CLUG meeting scheduler as a "LUG in a box". I hadn't thought about it, but I have written it reasonably modularly, so if I polished it up a bit more and made a few more things more readily configurable, I guess I could release it. I've never considered anything I've written to be of a high enough standard, or of enough general applicability to warrant getting released, but hey, maybe this will be different.

[14:45] [clug] [permalink]

Wednesday, 22 September 2004

Pentagon plane crash conspiracy theory?

Sarah pointed out this "documentary", which raises some interesting points about the plane that crashed into the Pentagon on September 11, that I hadn't really given any thought to (probably because the WTC was more of an attention grabber).

It seems to be a plug for this website.

[22:47] [opinion] [permalink]

Tuesday, 21 September 2004

Running multiple instances of Gallery

Gallery rocks, 'nuff said.

I only became aware of Gallery at LCA '04, when Rasmus mentioned it in his PHP tutorial. I immediately chucked it on my webserver and threw out the dodgey thumbnailing-on-the-fly PHP code I'd written, and I've never looked back.

A little while before I reinstalled daedalus in the June holidays, my friend Alison, who is over in the UK at the moment, asked me for recommendations on online photo gallery websites. I told her she could use my Gallery after I'd reinstalled daedalus, but what I wanted to do was rather than have all her photos lumped in with mine, was to have a separate instance of Gallery with individual albums for all my friends, similar to what I'd seen done elsewhere.

Now I'd looked into running multiple instances a few times (I really didn't want to copy the static Gallery code to a second directory, because that would remove the benefits of security updates and whatnot, and was generally not the elegant solution I desired), but the FAQ didn't mention it, and Googling for such a generic term wasn't turning up much either.

So tonight I finally sat down and looked at the code, and it was remarkably simple. The way I wanted to do it was define an environment variable on a per virtual host basis, which specified where to look for the config.php file. I added an SetEnv directive to the vhost for the new Gallery instance, and only had to patch one file to add some additional logic. If I can get the Gallery upstream to accept the patch, it'll all be good. At the moment, a new Gallery package would clobber everything, which is exactly what I'm trying to avoid.

It's still not ideal though, as when you first rock up to the new instance, it creates a default user database, with no users, and I've got no idea how to add new users. I ended up copying the user database from my other instance to bootstrap things.

[04:09] [tech] [permalink]

m4d l33t meeting scheduler

It's been ages since I've set myself a little coding project, but recently when Michael asked if I wanted to take over organising CLUG meetings I immediately saw the potential for yet another automated system.

So I knocked up a very simple PostgreSQL database, a spanky looking PHP front-end (which I should probably apply some access controls to) and wrote a whiz bang Perl script to query the database and send emails (can you tell that I'm a bit chuffed with this thing yet?).

I'm coordinating speakers for the monthly CLUG (fourth Thursday) and CLUG Programmers' Special Interest Group (PSIG) (second Thursday) meetings, and so my funky Perl script checks the database on a Monday to see if there's anyone talking on that Thursday's meeting, and if there isn't, mails the CLUG list asking if anyone would like to talk. On a Tuesday, it announces that week's meeting, with details from the database if there are any.

Noteworthy Perl modules that have to get a mention are Date::Manip, which does so much cool, er well date manipulation, and a new one I found to add to my toolkit, Text::Template.

[00:40] [code] [permalink]

Monday, 20 September 2004

Further consolidation in the security industry

In the beginning, there was iSecure, and it was purchased by SecureNet shortly before I came on board. Shortly before I left to go back to University for a year, Betrusted bought SecureNet, and then bought SecureNet's main competitior, 90east. Now, Betrusted and some company called TruSecure have apparently merged and will become Cybertrust.

Quite the consolidation. Film at 11.

[21:47] [tech/security] [permalink]

Sudo bites

Well, it's really more of a PEBKAC issue, or proof that one should not attempt system administration after running a Fun Run.

I noticed that the broadcast address on daedalus was incorrect, so I added a broadcast line to my /etc/network/interfaces and went sudo ifdown eth0; ifup eth0 and immediately went "D'oh!". Not the best thing to do on a Sunday evening to a box a 1000 kilometres away in an unmanned (on the weekend) datacentre.

[03:13] [tech] [permalink]

Saturday, 18 September 2004

Canberra Times Fun Run

Today Sarah and I ran in the annual 10 kilometre Canberra Times Fun Run.

It's the third consecutive year I've run, and I managed to come in under an hour (59:31, a new personal best), which was my goal, so I'm pretty happy. I ran all the way, except between 6km and 7km, when I got a bad stitch and walked.

Sarah did well also, coming in at about 1 hour 11 minutes.

[23:42] [life] [permalink]

Wednesday, 15 September 2004

I don't think so, Tim

I've decided to cave in and buy the COMP2400 textbook, because the copy I borrowed from my flat mate is one edition behind, and I'm concerned the chapters don't line up with the lecture notes. Here goes another $89...

I checked on the bookshop's website to make sure that there were some in stock, and got this record back:

Course Code: COMP2400

Department of archaeology and anthropology, eh?

[17:10] [uni] [permalink]

Monday, 13 September 2004

New trick: hexediting with dd

Today's excuse to procrastinate on doing assignments was getting the USB-serial adapter (a Targus PA088 that my Dad gave me because he didn't need it) to work with MacOS X.

Google happened upon this page (some guy's blog), which mentioned the PA088, but was more a red-herring in that it sent me off down the wrong path, trying to hack a driver for another vendor's adapter to work with my adapter, which I presume was a totally different chipset.

There was a lot of user comments for this blog entry, and some of them were helpful, and I ended up finding a driver on the chipset vendor's driver page

The interesting thing that I learned in the whole process was how to make changes to a binary file using only printf and dd. I ended up having to make use of this, because the driver insisted on naming the device /dev/tty.USB Serial (yes, there's a space in there).

Using hexdump -C, I identified the offset in the driver file of the space in the string "USB Serial", and then did this: printf "\137" | dd conv=notrunc of=USBSerialDriver seek=26071 count=1 bs=1 and then lo and behold (well actually after a bit of unloading and reloading of the kernel extension) the device had a more sane name.

[01:23] [tech] [permalink]

Sunday, 12 September 2004


Joey's braindump motivated me to have another crack at trying to build d-i locally to test a patch I'd come up with.

So I had a grand old time fiddling around with dchroot, and successfully built d-i in a sid chroot, incorporating my netcfg udeb. I was a bit confused at first, because the resulting .iso and initrd didn't have a .udeb in sight, but in the case of netcfg, it seems that it gets installed in the initrd beforehand, so it wasn't an issue in this particular instance.

Anyway, the big excitment for the afternoon, other than actually building d-i (which isn't that exciting, because I've built it before) was that my netcfg patch actually worked, which means as long as the netcfg maintainers are happy with it, #262624 can be resolved.

[01:45] [debian] [permalink]

Saturday, 11 September 2004

Housewarmings, movies and running...

Last night Sarah and I went to Mel's new place for a bit of a housewarming. For some strange reason we spent more time in the cold back shed than the nice warm house, but hey... It was nice to see Ash again as well. The house is huge, and right across the road from Ash's school.

After that, we went to see The Village with Mark, who's in Canberra for 6 months on a contract. It was really good. If you're thinking it's going to be as suspenseful as Signs (I did because of the way the trailer portrayed it) don't avoid it on the grounds it's going to have you on the edge of your seat and scare the crap out of you. The trailer misrepresents the movie in my opinion. But that said, the movie was excellent, with one of M. Night Shyamalan's trademark twists.

This morning Sarah dragged me kicking and screaming along to a run around Mount Ainslie with her. It ended up being pretty good. We did a 10 km loop around the reserve, and I was pretty happy with how I held up. Good practise for next week's fun run. It ended up being a really nice day for a run. So this afternoon I have some laundry to do, some cleaning, and I need to make a start on one of the multitudes of assignments I have to do over the mid-semester break.

[18:33] [life] [permalink]

Friday, 10 September 2004

Rudimentary netcfg hacking

#262624 has been annoying me for a while. The other weekend when AJ was here, we spent one night pulling apart iwconfig and comparing how it sets the WEP key to how netcfg was doing it (and not doing it successfully it seems). The result was a very rudimentary C program that would unset or set a WEP key.

This afternoon, I finally sat down and fiddled around with the netcfg source to work up a patch, and submitted it.

One of the problems I have with doing d-i work is I lack sufficient d-i juju to actually rebuild the entire thing. I can build udebs, and I can build the d-i initrd component, but I'm buggered if I can build a new CD. This makes it impossible to test changes that I make, which really sucks. I've gone as far as trying to squeeze a modified udeb onto an existing ISO, but then the checksums don't match and the the installer barfs at me. It's rather frustrating... I think I need to build up my debian-cd zen.

[21:56] [debian] [permalink]

Thursday, 09 September 2004


So NASA's $264 million star dust gathering mission has come to earth with a dismal splat

What drives me nuts is if the thing costs so much money, and took so many years, and was so supposedly important, is why the heck didn't they put a bit more redundancy into the parachutes? What's the worst thing that could happen if say all three independent parachute systems decided to deploy? Gee, the headlines would say "Solar probe came to earth three times more slowly than expected".

Sheesh. I think the boffins at NASA are too busy being mental giants to think about the simple stuff...

[04:59] [opinion] [permalink]

Wednesday, 08 September 2004

Tee hee

The lab am I in is full of people frantically trying to finish that COMP1110 assignment. If I had a dollar for every time I heard "pointer mutation", "first fit" or "next fit", I wouldn't be an impoverished university student :-)

[22:32] [uni] [permalink]

Pointed haired management quote of the day

A co-worker showed me an email...

Who is Amanda and what is her role in the backups?

[05:41] [work] [permalink]

Monday, 06 September 2004

What a crap day

I had such a crap day today.

It started with a lousy night's sleep. Then I actually managed to extract myself from bed at 6am (for a change) and go to the gym. This left me in a totally shagged state for the rest of the day.

I got the results for the first assignment for COMP2400 back. I got 37 out of 60, after losing 10 marks because I put exit at the end of each SQL file. I'm rather pissed off about this, because the assignment specification didn't state how the assignment would be run for marking purposes. I did my testing by going sqlplus / @filename.sql and I wanted to be returned to a shell prompt after running the query. Of course, the bulk marking tried to run them all in batch, and mine exited after the first query, requiring the marker to edit all 10 files to comment out the exit, hence the mark deduction.

I'm also annoyed at myself because some of the queries weren't logically correct. I should have known better. I know SQL. I've been doing it for years. Granted, I don't usually join a table on itself 6 times to answer a question, but hey, I had higher expectations for my results.

Moving right along to Maths, the results were back (already) for the mid-semester. I wasn't expecting to do anything spectacular, but I also wasn't expecting to pass by 1 mark either. At least it was a pass I guess...

The one highlight of the day was the result for the Finance mid-semester: 21 out of 26, or 78%. Given I was contemplating not sitting the exam, I'm pleased with the outcome, and I'm pleased that I did.

So now I just have the COMP1110 mid-semester on Wednesday evening, and the COMP1110 assignment due on Friday as immediate concerns. I dare say the third COMP1110 assignment is going to be released before the break, and so that'll leave me with three assignments to work on over the mid-semester break. Oh joy, oh glee.

[01:59] [uni] [permalink]

Mysterious outage

Just after midday AEST, daedalus.andrew.net.au started acting weird. My SSH connection was unceremoniously closed, and subsequent attempts to reconnect resulted in the TCP connection being opened, then immediately closed before any SSH protocol negotiation took place. Sendmail was still responding, but Apache seemed to have died. I had to ask for the guys at the colo facility to kick it in the guts.

Afterwards, I couldn't find anything in the logs. There's pretty much a gap from 12:14 to 13:39 when it was rebooted. My immediate thoughts, prior to inspecting the logs, were that either Mhonarc or MIMEDefang had eaten up all the memory for some reason, and the kernel had started killing things indiscriminately, but I'd expect to see some general carnage in the logs to indicate this. So now I'm going to check the AIDE output...

AIDE seems to think there hasn't been anything untoward (of course, I should really get a copy of the AIDE database off the box as soon as it's created each day...)

chkrootkit was also happy with things.

So I think I'll have to chalk it up as an undetermined crash. :-|
I don't like instability on a box a 1000 kilometres away.

[01:44] [tech] [permalink]

Saturday, 04 September 2004


I think I've finally got the second part of my all time current favourite assignment working. This is great, because I want to spend the rest of the weekend and up to Wednesday studying for the COMP1110 mid-semester exam, and then use the extension to polish it a bit. Having the algorithm working what looks like correctly takes a load off...

[01:52] [uni] [permalink]

Friday, 03 September 2004

changelogs.debian.net breakage

I just realised that changelogs.debian.net had broken, and didn't think it was terribly good to discover this by accident, so I wrote a quick Nagios plugin to keep an eye on it for me. While I was at it, I rejigged the script that keeps the copy of sid's Sources file up to date. It now uses rsync, and shouldn't have to redownload the whole thing every night.

[23:19] [debian] [permalink]

Two down, one to go...

Sat the mid-semester for MATH1005 today.

I don't think I'll do anything too spectacular - pass or fail. Hopefully a mediocre pass. Given that I really didn't get much of an opportunity to put in a lot of study, that's all I can expect.

[02:11] [uni] [permalink]


I was driving home from the gym this morning when I heard on the radio about how our very own Miss Universe, Jennifer Hawkins, had a rather embarrassing wardrobe malfunction, stepping on her dress and pulling it off, revealing her G-string clad arse. Cracked me up.

[02:07] [humour] [permalink]

Thursday, 02 September 2004

No vaiostat for Sarge

I've decided to file a release critical bug (#269615) against vaiostat-source as it was recently brought to my attention that it wasn't building correctly against current 2.4 kernels. I confirmed this, and dropped upstream a line. He was rather chuffed that it was in Debian in the first place, but has been busy and a bit unresponsive regarding testing and fixing the problem.

[07:43] [debian] [permalink]


Got this email to my personal address from one of my COMP1110 lecturers tonight...

Subject: CVS archive too public

Uh, nice web site.

I was surprised to find this:


It would be nice if you could keep your assignments to yourself.
So much for security by obscurity. I'd be surprised if a student found it. Or I would have been until this.

[04:53] [uni] [permalink]

Holy crap!!!

AJ said he had good Google Juice, but jeez, he only added my blog to his blogroll on the weekend, and now my blog is the first hit when you Google for my name...


[04:46] [geek] [permalink]

Call For Papers spam

Tonight I sent mail to the SAGE-AU announce and Aussie-ISP mailing lists letting them know the CFP was open. I also emailed Carter Bullard of Argus fame to see if I could convince him to put a paper in the CFP.

[04:32] [lca] [permalink]

LCA DoSes Subway again...

It's always amusing watching the denial of service effect of catering for an LCA committee meeting with Subway. Steve phoned the order through for collection at 6pm. I rolled up a few minutes after 6pm to see a lengthy queue and all hands on deck madly trying to make 9 foot long subs as well service the queue. Another 10 minutes later and the order is fulfilled. At least they gave me 6 double-choc cookies as a consolation :-)

[04:29] [lca] [permalink]

Wednesday, 01 September 2004

Patience is a virtue...

This morning I had to go back to the eye hospital for a post-op checkup. They always run massively behind schedule, so I wasn't overly concerned about running 10 minutes late for my appointment. As it happened, this time, there was even a queue just to check in at reception, and the waiting room was bulging. For the first time a noticed a little sign at reception saying something along the lines of "we know we have long waits, please be patient, we often have emergencies come in, and have to slot them in, and we'll always spend as much time with patients as is required, regardless of the schedule etc etc". I don't have a problem with it too much. I'd rather see the doctor knowing they're going to spend the appropriate amount of time on me, rather than try and rush me in and out in 30 seconds because they're running 30 minutes behind schedule. I actually used the time to do some further work on that assignment, so the wait (about 30 minutes) for me, wasn't an issue at all.

There are always a lot of elderly people in the waiting room. I figure they must do a lot of cataract surgery or something. Today was no exception. There was some particularly old bloke in the waiting room, with a woman who I presumed was his daughter (she seemed too young to be his wife, but she was reasonably old herself). They were bitching and moaning at the top of their lungs (I think because he must have been a bit hard of hearing) about how they'd been there for an hour and twenty minutes. And the bitching and moaning was ongoing. I could have ignored it easier if it wasn't so loud. It really started to irritate me after 10 minutes. Fortunately I didn't have to wait that long, and fortunately they were called in to see their doctor not too much longer after they started irritating me. I don't see the point in getting all worked up about the doctor running late. It just makes you all bitter and twisted, and doesn't make the time pass any faster...

Oh, and I apparently have very near to 20-20 vision now. Yay me.

[23:41] [life] [permalink]