|
| Greg's diary |
| June 2008 |
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Sunday, 1 June 2008 | Dereel | Images for 1 June 2008 |
| Top of page | ||
| next day | ||
| last day |
Spent a lot of time with documentation today, both with web pages and “Beautiful Architecture”; I've managed to get the tables working, though no thanks to any of the tools I have.
Had just about finished that when Yvonne suggested we head to Mount Wallace with Chris to view a newly-imported Peruvian Paso stallion:
|
|||||||||||||||||||||
That kept us until evening, so didn't even get a chance to look round the garden today.
The CFA are currently going round distributing information and asking for donations. That's definitely a worthy cause, and this afternoon we were discussing giving a substantial sum.
Then I opened the package and found a CD:
|
|
||||||||
Note in particular the inscription:
|
|
||||||||
The message is incorrect, of course. It doesn't require just a PC; it also requires Microsoft, as the second line implies. It seems that our Government has decided that it's sufficient to limit the supply of vital survival information only to those people who use a specific operating system. Why? I'm thoroughly disgusted.
For me the answer is clear: formal complaints, and no donation. Instead I'll offer my services to provide a standards-compliant CD instead of one requiring proprietary software.
This CD was made by Vividas, whose standards compliance credentials are evident: I got 60 validation errors on their home page. Here's an excerpt from how their home page renders on my system:
|
|
||||||||
15 years later Vividas is history.
| Monday, 2 June 2008 | Dereel | Images for 2 June 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
My DocBook adventure is pretty much over—I suppose I should really go back and fix up the references—but I'm still interested in the problems it raised. Adam Witwer of the O'Reilly tools team recommended trying oXygen (also known as <oXygen/>), which is commercial software with a 30 day trial license. Downloaded that—over 50 MB of binary!—and tried installing it.
There are no FreeBSD binaries, of course, so I downloaded the Linux version, which didn't install on FreeBSD. Installed on ceeveear instead, which went about as well as you could expect. Here some screen shots:
|
|
||||||||
Aaargh! The horrible pathname with spaces in it! I still don't know why people do this, but it's clear that they don't write scripts.
|
|
||||||||
And here it wants to create a desktop icon. How can it do that without knowing what desktop? This machine doesn't even have a display. Not surprisingly, it failed, but I was able to start oXygen by running a shell script with the undocumented name ~/Oxygen/oxygen (watch those capitals, neither of them matching the name of the editor).
The documentation for the editor looks reasonable, once you get past the help menu, which offers to register the editor, but doesn't tell you where the manual is:
|
||||||||||||||||||||||
It's there, though, under the heading “Help...” (what does ... mean?). Some of the chapters have rather surprising content: “Getting started” tells you what kind of documents it can handle (isn't that part of the Key Features?), and “Editing Documents” appears to be only a description of how to select Unicode characters (in a manner reminiscent of my diatribe about OpenOffice, slide 26 and on):
|
|
||||||||
It's not until you click on the little book icon in the too-narrow left-hand pane that it opens up and shows you the rest of the contents; there's no way to even know they're there until you do; the individual pages have no link to the next.
|
|
||||||||
Played around with that for a while. At least it displays the real markup text and not just an approximation to the results. It also allows you to enter and save invalid text:
|
|
||||||||
I don't know if that's good or bad, but it doesn't do anything obviously better than nXML mode, which leaves me wondering why I bother. It's also worth noting that it highlights the wrong tag, the <para> tag, when it's clearly the <foobar> tag that's wrong. nXML mode gets this right:
|
|
||||||||
Still, it looks like a relatively well put-together piece of “modern” software. If you like pushing mice around, deciphering unrecognizable icons and reading inappropriately sized panes, this might be for you.
Panes are an interesting thing. They've been around for a long time, of course, since the days when “windowing” environments didn't have proper window managers. They're marginally better than tabs, since they show you a little of their content all the time, but they have to be a compromise between their own size requirements and the overall size of the window. The solution is obvious: make each of them a separate window which you can move around (ideally to a different display), or even iconify or close when you don't want it, but hardly anybody seems to do that. You'd have to think that Microsoft limits you to a single window, but that's hardly likely. oXygen does use multiple windows—help is in a separate window—but it insists on keeping the help window above the main window, even if you try to raise it, and it iconifies them both together.
High time to do some more brewing. Racked my last brew and found more of the same scaly infection floating on the beer. It must be in the fermenter, since I didn't use anything from the previous brew. Filled it up with water, bleach and a bit of hydrochloric acid (which releases free chlorine). After a day of that, I hope nothing will survive.
Crushing grain is still a pain! Today, for some reason, nothing would grip in the grain mill, and I was turning for ever. Finally gave up and tried it in the chopping attachment for the mixer, which worked much better, but I'm a little afraid it could lead to another stuck sparge.
Also spent some time correlating old photos. I still prefer to learn for myself rather than to take accepted frameworks for this sort of thing, but it does mean that over the last 10 years I've had several different ways of arranging things. It's not just a matter of moving them, though that's a problem in itself: they're on the web and could have links to them, which I'd like to be able to resolve. I'm currently using symlinks, but I wonder if some heuristics in the 404 document would help.
| Tuesday, 3 June 2008 | Dereel | Images for 3 June 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Lots of fermentation-related activity today: bread baking, making kimchi, and preparing for another brew of beer.
For that I needed to finish milling the grain, and it occurred to me that I've had an electric grain mill literally for decades—I bought it in the mid-1970s for grinding spices. Why did I then go out and spend lots of money on a hand mill? Maybe I was worried that it wouldn't be up to the task, or that the results wouldn't be good enough. Tried it out and found that neither was correct: it was faster than the hand mill was even when driven by an electric drill, and the resultant crush looks more even:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
This shows four different crushes: above centre Weyermann Pilsener Malt through the hand mill, below centre Weyermann Pilsener Malt through the electric mill, on the left, Powells “Munich” malt through the electric mill, on the right Powells “Munich” malt through the hand mill. In both cases, the electric grind looks more uniform. The difference in grind between the Powells' malt crushes is particularly evident, in particular the almost oat-like appearance of the husk in the hand-crushed grain (right):
|
|
|||||||||
|
|
|||||||||
This seems to be at least partially a characteristic of the Powells malt, though I don't understand why.
About the only thing that is different is that the grain comes out slightly warm, maybe 25°. How serious is that? It's been dried at higher temperatures, and then next thing that happens to it is that it is mashed at temperatures over 60°.
We're still waiting for more rain, and until recently there seemed to be nothing on the horizon:
Australian Government Bureau of Meteorology Western District Forecast Issued at 1531 on Monday the 2nd of June 2008 ... Nil. Ballarat Tonight and Tuesday Fine apart from morning fog patches. A partly cloudy day with a light wind tending moderate southeasterly. Min 3 Max 15 Precis: Fog then fine. Wednesday Fine. Min 4 Max 13 Thursday Fine. Min 6 Max 12 Friday Fine. Min 6 Max 12
“Fine”, of course, means “dry”. But then only 24 hours later we get:
Issued at 1531 on Tuesday the 3rd of June 2008 Ballarat Tonight and Wednesday Fog redeveloping this evening, then gradually clearing tomorrow. Cloudy conditions to follow with a shower or two developing. Light to moderate southeasterly winds. Min 4 Max 13 Precis: Fog, then a shower or two. Thursday Shower or two. Min 6 Max 14 Friday Shower or two. Min 6 Max 12 Saturday Shower or two. Min 5 Max 11
I wonder how long and how much computing power it will take before accurate weather forecasts are possible.
| Wednesday, 4 June 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Today I was supposed to brew again, and not too early—I'm almost out of beer. But since seeing the demo at Grain and Grape last month, I can't face brewing with my primitive setup. Today was nearly ready to start when I couldn't find the sparge ring for the mash tun, and so had to postpone. Yvonne came back from horse training shortly after midday, but by that time it was too late to start: it takes about 90 minutes to mash, 45 minutes to sparge, 2 hours to bring the wort to the boil, 1 hour boil, 20 minutes aeration, for a total of at least 5 hours.
Spent the afternoon tidying up the shed while Yvonne planted some stuff in the garden. Hopefully it wasn't too early; I'm concerned that we haven't prepared the soil well enough yet.
In the evening managed to damage my sparge ring by heating up the pot without any water in it; I had forgotten that I had decanted the water to another pot. There was surprisingly little damage, but the plastic washer in the disconnect melted, and I'm concerned about whether it will still be usable, both from a functional and from a taste point of view. Something (my subconscious?) seems to be telling me to upgrade my equipment.
| Thursday, 5 June 2008 | Dereel | Images for 5 June 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
I've decided not to brew any more until I have better brewing equipment. Started thinking about how to achieve that, with the intention to go into town and do some shopping for what I could find in Ballarat, That took so long, however, that I decided to put it off until tomorrow.
More work in the garden, transplanting more stuff. On investigation, the “black lilies” appear to be a kind of Dracunculus vulgaris. They grow from a large tuber. Replanted a few to see how they'd deal with it; it's not clear how much sun they like.
In the afternoon investigated how to recode TV recordings (stored as an MPEG-2 transport stream) as a program stream without commercials. I had already tried once without success, and today it wasn't much easier.
Based on my recollection of that experience, decided to use mencoder. I had already written a script that built Edit Decision Lists (EDLs) from the database, but somehow I missed one of the myriad parameters—and forgot to write down what I had done, so at a certain point I was back to the beginning. It goes to show the utility of writing down the details when keeping a diary.
This time started by RTFM'ing mplayer(1), which describes both mplayer and mencoder. It's not easy reading, and it's 7517 lines long, about 120 pages. When I did find the parameters, I ended up with a continual stream of messages like:
ERROR: scr 11.901, dts 6.560, pts 6.620
ERROR: scr 11.949, dts 0.000, pts 6.580
ERROR: scr 11.968, dts 0.000, pts 6.600
No help in the man pages, of course, and relatively few on Google. What I found was a suggestion to encode with the parameters -mpegopts format=dvd:tsaf. That's clearly intended for DVDs, and I was working with 720p streams, which are incompatible with DVD formats. Tried it anyway and, sure enough, the error messages went away. With the more logical -mpegopts format=meg2:tsaf, they reoccurred. I couldn't see any encoding errors in either case, so I suspect that the error message is spurious, and the suggestion is a bug workaround. Interestingly, the dvd parameter doesn't limit the resolution to DVD values.
That still didn't drop commercials. Looked at the description of the EDL, which was vague enough: three columns, meaning start time (in seconds), end time and action (skip or mute). The example shows white space as delimiters. But must the first column start at the beginning of the line? Must the space be spaces, or are tabs allowed? Who knows?
Well, I know now. The first column doesn't need to start at the beginning of the line, and I think that the white space can include tabs, though I left them out of the script once I had finished. That wasn't the problem at all, though specifying the time in seconds meant that I first had to find the frame rate before I could build the EDL, something that I had not thought of last time.
After a lot of experimentation, discovered that it was because the TS has time stamps that start at some arbitrary time in the past, not at the beginning of the recording. mencoder has a parameter -ignore-start, but it's documented to only work for avi files, but then lots of mplayer documentation suggest that the support for MPEG is even worse than is the case, so tried it anyway—and it didn't work. I was at least able to confirm that it worked if I did it twice, since the output of first recode is a program stream which starts at 0. But what a pain! It would all have been so much easier if there were an option to specify frames rather than times.
Decided that it wouldn't do any harm to keep the “commercials” in the stream, in case the heuristics for guessing the commercials were wrong. So now my recode just makes a program stream, and mplayer interprets the EDL on playing. That works, and if there is any problem, I can even skip back into the commercial: EDLs appear not to work in reverse.
So, to avoid having to search again next time: here's recode, a bash function, without some of the comments I have in the current version, which is still subject to change:
recode ()
{
# parms:
# -msglevel 1 Error messages and worse
# -oac copy Copy audio
# -ovc copy Copy video
# -of mpeg Output in MPEG-2 container
# -vf harddup Insist on duplicate frames (why?)
# -mpegopts format=dvd:tsaf No idea. We don't want DVDs, but it solves error messages
# format=mpeg2 should be the correct one, but it produces the
# messages
MENCODER_PARMS="-msglevel all=1 -oac copy -ovc copy -vf harddup -of mpeg -mpegopts \
format=dvd:tsaf -ignore-start"
mencoder $MENCODER_PARMS -o $1.recoded $1
STATUS=$?
if [ $STATUS -eq 0 ]; then
touch -r $1 $1.recoded
mv $1.recoded $1
else
echo '***' Failed, status $STATUS
fi
}
And here's gencutlist:
#!/bin/sh
# $Id: diary-jun2008.php,v 1.90 2023/07/16 22:48:04 grog Exp $
# Get cust list information from database for specific file
for i in $*; do
FILE=`basename $i`
FPS=`mpid $FILE | awk '{print $7}'`
echo $FILE is $FPS fps
mythcommflag -f $FILE --gencutlist
image=`echo $FILE | \
sed 's/\(....\)_\(....\)\(..\)\(..\)\(..\)\(..\)\(..\).*/chanid = \1 and starttime = "\2-\3-\4 \5:\6:\7"/'`
echo "select mark from recordedmarkup where $image and (type = 4 or type = 5) order by mark" \
| mysql -s mythconverg \
| awk -v FPS=$FPS -f ~mythtv/queries/makecutlist > $FILE.cutlist
done
This calls two scripts. mpid extracts parameters from an MPEG stream:
#!/bin/sh
# $Id: diary-jun2008.php,v 1.90 2023/07/16 22:48:04 grog Exp $
#
# Identify an MPEG stream.
# usage: mpid $*
if [ "$1" = "-v" ]; then # verbose: lots of stuff
shift
for i in $*; do
if [ -f $i ]; then # ignore non-files; this is only part of the story
echo -n "$i "
mplayer -vo null -ao null -frames 0 -identify $i 2>/dev/null | grep -v ^ID_
fi
done
else
for i in $*; do
if [ -f $i ]; then # ignore non-files; this is only part of the story
echo -n "$i "
mplayer -vo null -ao null -frames 0 -identify $i 2>/dev/null | grep ^VIDEO:
fi
done
fi
makecutlist then takes the cutlist extracted from the database and converts it into an EDL. Yes, now it's clear that the name isn't quite what is sounds like.
# Take individual frame numbers, one per line, and print in EDL format
# $Id: diary-jun2008.php,v 1.90 2023/07/16 22:48:04 grog Exp $
{
if (havestart == 0)
{
cutstart = $1 / FPS;
havestart = 1;
}
else
{
cutend = $1 / FPS;
commlength = cutend - cutstart;
printf ("%8.2f %8.2f 0\n", cutstart, cutend);
havestart = 0;
}
}
| Friday, 6 June 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Finally got round to go into town today, though I didn't achieve much. First to buy some sherry—why is real Spanish sherry so weak nowadays? It's a fortified wine, after all, and things like Tio Pepe used to be 17% v/v ethanol, but now all the Spanish sherries I can find are 15%, barely stronger than normal Australian wines. A little less ethanol isn't a bad thing in itself, but they also taste so insipid. I get the distinct impression that it has been watered down.
While I was there, into a bookshop to look for a book to give as a present; it had occurred to me that a cookbook would be appropriate, and since Yvonne is French, a French cookbook seemed appropriate.
I've ranted about cookbooks for a long time, in particular units and quantities. I was pleased to note that some of the books really did specify the units clearly. But the quantities? I found recipes for Paella valenciana in two Spanish cookbooks, one of which at least noted that beans are frequently included; but both specified, for a large paella, “a pinch of saffron”. That's a far cry from the 1 g that I use, and I get the impression that neither author understood the crucial importance of the spice in the dish.
But I was looking for French, and spent some time considering what dishes to check. pot-au-feu? boœuf bourgignon? They didn't seem to have them. They did have “duck à l'orange” [sic], so I checked that. Both of the recipes included mint, which in my experience is anathema to the French; I certainly can't imagine where you would buy the kind of mint that the recipe presumably implies. Where do people get these ideas from?
So no book. I did end up buying “The foodies' guide to Melbourne” and a book on pruning shrubs and trees; together they came to considerably more than I had planned to spend on the cook book, but both fill a real need. The former, hopefully, will lead us to better food supplies.
As a result of the long search for cookbooks, was late for everything else. To Celsius in the hope of getting a cheap sprinkler solenoid valve, but they have now completely closed down. To the home brew shop to discover that carbon dioxide cylinders cost between $270 and $350 (by contrast, LPG cylinders cost $30), and to BOC, who couldn't supply me with fittings for my gas stove, and was told that they would rent me a cylinder for about $117 a year, ostensibly because they needed testing on each refill. They would not sell me a cylinder. I'm left with the feeling that they're abusing their market position.
At Midland Irrigation finally found a sprinkler solenoid valve without difficulty, and not significantly more expensive than Celsius would have asked.
Back home, set to recoding my remaining MPEG TS files, not with complete success: on one file, at a specific position, mencoder SIGSEGVd. My guess is that this is the usual mplayer breakage; the file was pretty messed up.
More garden work. It's looking more like winter now, though the cannas are still blooming, and the first daffodils are coming.
| Saturday, 7 June 2008 | Dereel | Images for 7 June 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
A fair amount of correspondence today. Tom Maynard writes:
You've pushed one of my buttons once too many times. Regarding scientific names for plants the genus is capitalized, but the species is not. So your Dracunculus_Vulgaris <http://en.wikipedia.org/wiki/Dracunculus_vulgaris> should be "Dracunculus vulgaris".
Guilty as charged, and changed.
Matthias Schmidt writes:
I saw a screenshot on your website displaying mutt:
http://www.lemis.com/grog/Day/20050908/big/wantadilla:0.1.gif
The color scheme with the yellow background looks very nice, so I wanted to ask, if you could send me your muttrc?
Most of my configuration files are on my programs and hacks page, but the .muttrc contains too much personal stuff, so I don't publish it. The methods are described in the mutt manual, but since Matthias asked specific details, here's an excerpt from my .muttrc. It doesn't include all details, just enough to give the idea. The first colour is the foreground (text), and the second colour is the background. To add to the fun, I have remapped green to bright white in my xterm config, so where I write “green“ below, I get bright white:
# default colours
color hdrdefault brightblack white
color quoted blue white
color signature red white
color indicator white blue
color error brightred white
color status black yellow
color message red white
color attachment magenta white
color search black green # how to highlight search patterns in the pager
color header brightblue white ^(From):
color header red white ^(To):
color header red white ^([Cc]c):
color header magenta white ^(Subject):
color header cyan white ^Date:
color body red white "(ftp|https*)://[^ >,]+[^]) >,.]" # point out URLs
color body red white [-a-z_0-9.]+@[-a-z_0-9.]+ # e-mail addresses
# Highlight messages with specific texts
# Note that "green" is remapped to "bright white"
color index blue white FreeBSD
color index blue yellow netbsd
color index black yellow brew
Leonne, surname unknown, has taken a look at my mystery plants page and has a number of comments. She writes:
The grub is commomly called a Bardi grub. They are the larval stage of a moth. Good for fishing bait but not much else. It eats the feeder roots of plant and often leaves dead patches in lawns. There is a spray you can get from most gardening places to get rid of them.
The plants you have as asters are very hardy drought tolerant plants. The two yellow ones are different types of Gazinias and come in all shades from red to white. You often see them on the sides of roads and suburban nature strips. The middle variety self seeds and can take over a neglected patch of ground.
It seems that Leonne didn't follow the link here. Also, I believe I'm correct in guessing that all the flowers belong to the genus Asteraceae.
Mystery 1 is a common english country garden plant whose name escapes me at the moment. A nursery man should be instantly able to name it for you.
Mystery 2 is a type of Arum Lily. Does it smell really bad like rotting meat? Lilies of that colour often do.
Again, it seems that Leonne didn't follow the link here.Mystery 3 is Sparixia (not sure on spelling) very old fashioned hardy bulb that self seeds. very similar to Ixias (again not sure on spelling) My Grandmother had acres of them when I was a child and they never got watered and were mown off when they finished flowering.
Mystery 5 looks like a Pandora creeping vine.
I subtitled this one “creeper”, which is incorrect. This might have misled Leonne. In fact, they're rather like the Sparaxis.
Mystery 7 is a native oxalis non invasive like the yellow one. Best kept neglected. Phosphorus intolerant.
The Salvia looks like Pineapple sage. Do the leaves smell like pineapple when crushed? If so great in summer salads and cooking.
Once again, it seems that Leonne didn't follow the link here. I state there that the leaves smell like mint.
weed 1 looks like Pig weed. Australian native supposedly edible Indigenous food with medical properties.
In the afternoon, spent some time taking some extreme close-up photos:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
A number of things have become apparent:
I took these photos with the help of fluorescent light and without adjusting the white balance, so the colours are incorrect. In particular, the background should be neutral white. I thought of starting again, but it's a considerable effort, and it should be motivation to fix them with the GIMP.
Modern cameras have many advantages over those of 40 years ago, but the region of macro photography is not one of them. From my Pentax Spotmatic days I have a set of automatic extension tubes (i.e. they stop down the aperture automatically) and the bellows in the photo above, which also have a cable release connector to stop down the lens. Today extension tubes are devoid of any logic, so they're actually worse off than 40 years ago, and a quick look at eBay shows that most of the bellows on offer are of a similar age to mine; only 8 of the 287 items were advertised to be for digital cameras, and even then I'm not convinced that they were correctly listed.
With macro photography, more than anywhere else, good focussing is crucial. The viewfinders of modern digital SLRs no longer have microprisms or split image, and they typically have a much smaller field of view than the old film SLRs, all of which makes things more difficult. I can't believe that this is an unavoidable consequence of smaller sensors.
My suspicions about the cause of the flare on earlier macro exposures proved to be correct: it's due to the 50 mm f/1.8 Zuiko lens from the Olympus OM10. Here photos taken with the 50mm f/1.4 Super Takumar from my Spotmatic and then the Zuiko lens:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
More tidying up. I'm hoping the house will look tidy and normal by the first anniversary of our moving in next month. At least moved a whole lot of stuff into the shed. Finally the dining room table is clear! Here a photo from 9 September 2007 and now:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
We only ever used it once, then started using it as a repository for tools in early August last year.
| Sunday, 8 June 2008 | Dereel | Images for 8 June 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
I'm quite happy with the book I bought a couple of days ago, RHS Pruning & Training, by Christopher Brickell and David Joyce. It's the first book I've seen that makes the distinction between pruning (normal maintenance of a bush) and renovation (making good previous neglect). It's the latter that I want, and the book addresses it almost directly:
In a newly acquired garden, neglected roses, particularly bush roses, often look as if they are hardly worth keeping. ...
Today decided to apply the techniques—of which there are two: cut down 50% to 70%, and cut down to the ground—to a couple of rose bushes in the north bed. These photos are more obvious when enlarged (click on one of the photos):
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
We'll see how they fare. The second one is in the shade most of the time, so it's never going to be really happy.
Winter is here, sort of: we have autumn plants that only now are coming into bloom, such as the Strelitzia reginae, and also spring plants, such as a rather sleepy daffodil:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
It was supposed to rain today, but the weather bureau changed their minds again, so Yvonne decided it was the day to finally burn off the pile of wood we have accumulated over the last 9 months:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Somehow didn't get much else done.
While watching TV in the evening, we considered how long it would take to drive from San Francisco to Pasadena. Clearly most people would immediately think of California, and that was my intention. But to make it quite clear, I specified the journey as from: san francisco, usa to: pasadena.
I've had some strange results from Google Maps, but this one takes the cake. First it took me to Seattle, then by kayak to Sydney, then across country to Adelaide, which has a suburb called Pasadena. But why Seattle? Why a kayak? How can you get across the Pacific in a kayak in the 42 days they specify?
| Monday, 9 June 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
The fire's still going! Most of the wood burnt out in the first hour yesterday midday, but some of the larger trunks were still going this morning. It continued all day, and in the evening there were still some remainders.
We've been trying to second-guess the Bureau of Meteorology for some time now, and today we decided that despite the forecast of rain, we could probably go riding. Set off, and within about 500 metres it started to rain, so back again after one of our shortest rides ever. And, of course, the few drops that came down were all for the next several hours.
More work on the sprinklers, and finally installed the longest stretch yet, 73 metres at the north of the house. Now I just have the south side to think about.
| Tuesday, 10 June 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
The fire's finally out! And, to my surprise, there was almost nothing left—just a couple of charred pieces about 50 g in weight. That'll change, of course; we have more radical pruning to do.
Into town today to continue my search for various odds and ends. Finally found a cookbook as a present, “Cooking &Travelling in South-West France”, by Stephanie Alexander. I had not wanted to buy a book by a non-French person, but this one had the advantage that it includes ingredients (possibly substitute) that are available in Australia, along with a list of places to buy them (sadly, we know nearly all the ones in Melbourne). Unfortunately, to judge by the map on the inside cover, she seems to have only visited the northern part of Aquitaine, and not been further in Midi-Pyrénées than a quick visit to Toulouse. Still, the recipes don't look too bad.
Then to Elgas for another attempt to find a regulator and hose for my gas stove. Regulators are no problem, of course, but the connector on the stove was “metric” and thus non-standard—in this metric country, the choices I had were all in inches. Bought a regulator and hose, and on to Enzed, where they found some adaptors pretty quickly and more cheaply than I had feared.
Back home, and it worked. At least something that went roughly according to plan. I wonder if the regulator is correct, though; the flame doesn't seem as strong as I recall it, but it's been over 11 years.
| Wednesday, 11 June 2008 | Dereel | Images for 11 June 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Phone bill arrived today. Telstra has finally paid all the sums agreed on 6 March 2008, after only a little over three months, though the way it was reported amazes me (this will require a couple of mouse clicks to make legible):
|
||||||||||||||||||||||
As near as I can tell, the remaining sums were two monthly fees of $88.14 and excess usage charges of $8.24, less a partial refund of the fees of $33.51, leaving a balance to pay of $54.63. They split the sum into $29.94 and $24.69 across two different dispute references, neither of which I can find anywhere in the correspondence. These people never cease to amaze me. Are they stupid, malicious or both?
I've been dragging my heels over the shelves I started putting into Yvonne's office a month ago. The stumbling block was the cabling that had to go under the floor, where the redbacks live:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Finally got round to at least starting, but there's plenty more to do.
| Thursday, 12 June 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
One of the problems we had when we moved to Dereel was that we didn't really have anywhere to put the dogs. In Wantadilla we kept them in the laundry, and they had access to the outside through a cat door.
Things are different here: we have a much smaller laundry, where the dogs have been sleeping anyway, and we can't just let them outside, because the main road is right next to the house. Yvonne has been asking for a cat door anyway, meaning that we have to maintain the temporary “prison” in perpetuity, not to mention cutting a hole in the laundry door. Today I finally gave in and installed one. I'm still not happy.
Chris around for dinner.
As the next part of my new brewery setup, bought a couple of stockpots on eBay and paid with PayPal. I have a balance of about € 100 with them, about AUD 170, but the pots came to $250. According to the PayPal policy, which is almost impossible to find again on the web, they should have taken the balance as partial payment and debit the rest from my bank account. Instead they left my balance untouched and debited the total sum from my bank account. What a pain! Does this mean that I have to spend exactly the sum in my balance to get rid of it?
| Friday, 13 June 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Winter is gradually coming, and with it the problems I've already mentioned with my air-conditioning system are becoming more acute. Finally decided to do something about it and called Fujitsu on the number specified, 1 800 226, 347. No reply—I just got an answering machine, so I left a message.
There was another link, of course, giving a number in Sydney (no free call), +61 2 8822 2500. Called that; it took three attempts to get connected, and with a bit of trouble they gave me the phone number of Angow and Hutt in Ballarat.
In the meantime, Brian called back responding to the first message and told me that they were Fujitsu, and they didn't do air conditioners—that was Fujitsu General, a different company. But where's the link from that page? Where's the overview? He also gave me the number 1 300 882 201, which isn't mentioned anywhere on that page, nor in the phone book.
Called up Angow and Hatt, left a message and got a call back from Colin Angow, with whom I had already spoken last winter. Spent about 40 minutes on the phone explaining the situation, which he suspects could be an “application design” issue, although he couldn't say why. He didn't completely reject the idea that the problem was that the system was measuring the wrong variable, but he didn't seem to be very keen on believing it either. About the only concrete things that came out of it were:
I would have to pay $60 for him to come and look at it, not even refundable if it proved to be Fujitsu's fault.
It should be possible to move the position of the sensor elsewhere, though Fujitsu don't support this concept.
There's a DIP switch somewhere on the controller that sets a fixed offset of 4° between reverse air temperature and set temperature. It can be disabled.
He also claimed that the “Auto” position on the heating/cooling selection disabled the temperature sensor and set a fixed 20° temperature. I have difficulty believing this, but I suppose anything's possible.
We're off to Melbourne tomorrow for my uncle Max's 80th birthday, and we need to stay at a hotel. Spent some time searching the web for good prices. The party's at the Box Hill Golf Club, and the nearest places I could find on the web were in Whitehorse Road in Box Hill and the Chadstone Executive Motel in Oakleigh, both about 5 km away. Finally chose the latter, because it has a free breakfast. Tried to call them directly rather than going by Wotif, but they weren't in the phone book; accidentally called another hotel with a similar name and address, who were in the phone book but not in Wotif, and they gave me the phone number. It's surprising that any hotel would not be in the phone book, and interesting (but probably not surprising) that there are hotels not in Wotif, who must charge more money than they think it's worth. Certainly a limitation of the web.
Got a reply to my complaint about the PayPal payment telling me that I had done it wrong, and if I had wanted to use my credit card, I should have done so. Clearly my prejudice against PayPal is well founded: I never mentioned a credit card, and they completely ignored my message. Sent an appropriate response.
| Saturday, 14 June 2008 | Dereel –> Melbourne | Images for 14 June 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Up early this morning to head to Melbourne. Just out of the shower, heard the tell-tale beeping of the UPSs. On further investigation, found that a circuit breaker had tripped on the circuit that supplies all the computers—and it kept tripping. Spent a lot of time turning off individual components, without finding anything, and then turned them all back on again, even (accidentally) a 2 kW heater, and it carried on working.
Was just scratching my head about that when Yvonne came in and told me that the dam water pump had stopped working, and that she had left it off. This is the pump that I have been using to pump water from the tank into the horse trough, and which I had been meaning to use for my new brewery. Confirmed that yes, indeed, it was that pump. Damn, especially since we were due to leave for Melbourne, and Chris still needed to give the horses enough to drink. Connected up another hose to the sprinkler system, which is (currently) supplied only by the submersed pump in the bore.
That wasn't everything, unfortunately. In the office, discovered that ceeveear had gone down, and for some reason hadn't restarted mythbackend. While trying to do that, managed to hang X on eureka: it hadn't remounted the NFS file systems (probably the switch was down at the time), the .ssh directory was missing, and this strange bug with the password prompt, which causes all X-related processes to go into a sleep state, bit home. Had to restart X.
Did that, and also copied all my photos from my camera to the computer, just in case. Tried to remove the files from the camera, and the machine (boskoop, Apple) hung: for some reason it had tried to remount file systems from lagoon, which was currently down. Rebooted lagoon to umount the file systems, and finally I was done!
Also a letter from PayPal, ignoring all previous correspondence and asking for exact details. I don't know why; these people never read anything anyway. Sent it all off, with some more clarifications. I'd be astounded if anything came of it.
Despite all that, off not just on time, but a couple of minutes early. Made it to Yarraville without incident. According to the “Foodies guide to Melbourne”, not far from Grain and Grape there's a bakery called Hausfrau. With a name like that you'd expect German bread, so we went to take a look. What we found was a cafe with some bread on sale, but nothing which, as the German proverb says, ripped us from the stool.
On the Grain and Grape, where I spent far too much money on pumps, false bottoms and gas cylinders. At least I now have most of the things I need to set up my HERMS system; all I still need are some connections and a heating coil, which was out of stock.
It was still early, so on to the Queen Victoria Market, where we ended up spending more money. Then to a Korean restaurant round the corner in Victoria St, where we had quite a good lunch, then off to the Royal Melbourne Botanical Gardens.
What a mess people have made of Melbourne traffic! I couldn't work out a way through the spaghetti on the South Bank, so down King St. to Flinders St., and then tried to get across the river at the station, on the corner of Swanston St. But you can only turn right there between midnight and 6 am. On east, trying to find a way to turn round, but made it all the way to the corner of Spring St and Little Collins St before I could finally find my way around. It took something like half an hour to drive 3.9 km. Interestingly, Google Maps chose exactly my intended route to get there, happily ignoring the ban on right turns into St. Kilda Rd. But then, they also think there's a restaurant at the destination.
It's been over 11 years since I was last in the Botanical Gardens, and in those days I wasn't overly interested in gardening. This time I was left a little disappointed: the place is clearly in need of more funding. In particular, the signs are insufficient. It's difficult to find your way around, and many plants are not identified.
That was particularly interesting and irritating for us. We found at least four plants that we had been trying to identify. Here the comparisons of the ones that we identified:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
We knew that this one is a salvia, but only now do we know that it's a Salvia microphylla. It looks quite like another variety, Salvia elegans (“pineapple sage”), but that bush has longer and more pronounced flowers with black stems:
|
|||||||||||||||||||||||
There seems to be a fair amount of difference of opinion on Salvia microphylla. Some call it watermelon sage, others blackberry sage, and nothing I've found on the web mentions the mint-like smell of the leaves, which the sample in the Botanical Gardnes also had. But then, Salvia elegans is supposed to smell like pineapple, and I didn't smell that on the sample I saw yesterday. There's also a variety of opinion on the size of the plant, though all agree that it flowers almost continuously. There also seem to be different cultivars, one of which has red and white flowers. None of the links on the web look quite like ours.
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
This plant has not flowered since we moved in; last year there were no buds at all. We had thought it was a kind of magnolia, but clearly it's some kind of Camellia. We'll know more when it flowers.
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
This is Aloysia triphylla, or “Lemon Verbena”. Yvonne was sure from the start that this was some kind of Verbena, but I hadn't been too sure: all the Verbena photos I had seen looked very different. That's probably because this plant belongs to the relatively small genus Aloysia. The leaves do smell of lemon.
|
|||||||||||||||||||||||
This one is everywhere in the gardens (though, like at home, not in bloom right now), but I couldn't find a name for it.
Then off to look for our motel, the Chadstone Executive Motel in Oakleigh. Printed out a map from Google maps to help us find our way: the easiest way to get there seemed to be to leave the Monash Freeway at Huntingdale Road. Set off along the Freeway looking for it, but it wasn't marked, and it wasn't until we passed Springvale Road that I realized that something had gone wrong. Left the freeway and set off back along Wellington Road. According to the map, the motel was 1.6 km after Clayton Road. By this time it was dark, and there were no visible house numbers, so crept along for some time before we finally found the motel—4.8 km from Clayton Road, and fully 3.2 km from where Google Maps said it was! I suppose it's only to be expected that Google Maps gives this distance as 4.0 km.
To be on the safe side, called the Box Hill Golf Club, who confirmed my suspicion that the other end of the map was wrong too: instead of taking us to 202 Station St, in the middle of the golf course, it took us to the end of Penrose St, off the east end of the golf course.
While researching the problem, I found much more breakage. From the previous URL, remove the to: address, and “202 Station St” moves about 200 metres west.
Next, I entered the addresses exactly as before and got almost the correct address for Station St , but still the wrong address for the motel. I also accidentally entered “Station Road” instead of “Station St”. Station Road doesn't exist round there. The result started further north on Station St, and took me closer to the correct motel address, just on the wrong side of the road, which involved a change of route.
It seems impossible to save directions in Google “My Maps”. here are the correct directions, which don't have much to do with what Google told me.
While trying to put both sets of directions on a single map, I came across still more errors. Adding the correct destination address to the original directions took me from Penrose St to Dandenong Road and back to Station St, clearly not what I wanted. So I removed the intermediate destination, which also drove the error into hiding. The “correct” error shows that the origin is 1.8 km and 4 minutes from the correct origin, and the motel is 2.8 km and 5 minutes away from the place marked on the directions (I measured 3.2). More to the point though, this implies a completely different route from the one Google Maps gave me. I've tried to superimpose them on a single map, first the wrong route from Box Hill to Oakleigh, then the correct route from Oakleigh to Box Hill, but then Google Maps gets confused about the first origin.
I've complained about Google Maps before. Clearly they're offering a good service, but they need some way to accept and incorporate feedback. Errors of this magnitude are inexcusable.
Finally off to the golf club, which we found quite easily—then discovered that there was a motel in our price class literally across the road. It was neither on the web nor obvious in the phone book (though I hadn't looked very carefully). Max had called us in the morning to tell me about the place, but I had thought it was the other place on the Maroondah highway.
The moral? The web has a long way to go before you can rely on it for daily business. Bad maps, incomplete hotel listings, lots of pain.
It's funny being back in Melbourne at a family reunion. It's been nearly 50 years since I last lived in Melbourne, and in that time I've been all over the world; now it's like coming home. It's made all the more obvious by the fact that during this time, Max and his immediate family have lived in almost exactly the same place (Hawthorn and surroundings) that my grandparents moved to in the early 1930s.
Max's birthday party was fun. His daughters Chris and Helen put on a skit about delivering him to an old folks' home (where the supervisor, played by Chris, ultimately refused to admit him):
|
|||||||||||||||||||||||
In the process they recapitulated his life and what a good bloke he was. The line I liked best was during an “assessment” of his mental capabilities. They showed a photo of his wife Margaret, saying “He loves to debate. In particular, he's been debating over and over again with this particular woman for over 60 years”.
|
|||||||||||||||||||||||
| Sunday, 15 June 2008 | Melbourne –> Dereel | Images for 15 June 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Woken at about 2 am by a car theft alarm going off outside the window. After about 10 minutes, while I was still wondering whether I should go outside and forcibly silence it, somebody finally turned it off. One more reminder why I hate big cities.
Like all hotels in Australia, we had to vacate the room by 10 am, so down to the Prahran Market, again for the first time in 11 years, to take a look around.
First to the Essential Ingredient, where we found whole foie gras for the first time outside France—packed in plastic. I didn't see a price, which was probably just as well; just about everything else had astronomical prices, including cooking pots costing up to $500. Decided that this was the yuppie side of town, and that Essential Ingredient catered for people who have more money than understanding.
The market itself is much smaller than the Queen Victoria Market, but it did have the advantage of being open on Sunday, and the general quality of the food is better. It's a pity it's so difficult to get to.
Decided against a Yum Cha lunch and then headed off back home, getting there in plenty of time for lunch. Found that the restart of X yesterday had also taken down my satellite connection monitoring collector. I must find a better way of ensuring that it keeps running.
Also a letter from PayPal, explaining why my credit balance was not used for the last transaction: it was in the wrong currency, and PayPal doesn't automatically change currencies, at least in the opinion of “Robert”. The only problem is, he's just plain wrong. All of the previous payments to eBay since March were processed in this way. The only difference in this case was that the balance wasn't enough to pay for the entire transaction.
Sent a stiffly worded reply asking people to first read my message and check the facts before making this sort of claim, and got a pretty quick response from “Liza”, who told me:
Upon reviewing your account, I can see that the balance on your PayPal account is in US Dollars and not in Australian Dollars that cause the system not to use the balance as you have sent payment in Australian Dollars as no Australian Dollars on your PayPal account the US Dollars is not used.
The grammar is original in more senses than one. Not only did she repeat “Robert”s nonsense, she got the currency wrong—my credit balance is in Euros. Another stiff message in reply, along with URLs and screen shots of the previous transactions. I never cease to be amazed how stupid these responses are—I don't think I have ever had a sensible reply from PayPal.
| Monday, 16 June 2008 | Dereel | Images for 16 June 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
One of the insights I gained in the 1960s while keeping a diary on paper was that “a page a day” did not reflect real life. One result, even in such interesting times as the Asia trip, was that I wrote far too much about routine events. I've since accepted the fact that, depending on the day, diary entries can vary from a line a two to several pages.
Last Saturday was an example of the latter, and today I spent most of the day documenting what I did over the previous two days. What a pain Google Maps are! Also spent some time putting together a toy greenhouse that I bought at ALDI yesterday.
I'm still waiting for a follow-up from PayPal, but I did get a mail message (with single line paragraphs, of course, which I have fixed here):
On 06/14/2008 I replied to the email message you sent regarding your PayPal account. As part of PayPal's commitment to excellence, I want to make sure I met your needs in my response. Would you please take a minute to answer a few questions to let me know how I did?
...
Thank you for your help!
PayPal Customer Support
Commitment to excellence, indeed! I didn't even get enough information to identify which of the four people who mishandled my case was meant. Followed up the survey, of course, and while I was at it, documented the whole sorry affair. As I said,
I have equally little faith in this customer satisfaction survey. I suspect that, like every other communication I have had with PayPal (this is the third such incident), it will disappear into a black hole.
| Tuesday, 17 June 2008 | Dereel | Images for 17 June 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
The stock pots I bought last Thursday have already arrived, so off to give some (positive) feedback to the vendor. In the process, a window popped up:
Buyers, you can no longer receive negative or neutral Feedback from sellers. You should leave honest and accurate Feedback without the fear of receiving ngiative or neutral ratings.
Revisiting this page 15 years later, it's not clear whether the typo above is eBay's or my fault.
That's excellent news. It stops sellers from giving negative feedback out of spite, like Cameta Camera did.
The size of the pots is interesting for a number of reasons: they're supposed to be 100 litres and 50 litres, but the measurements published on the web suggested that they're over 10% less. In fact, the volume is pretty accurate—if you include the volume of the pot itself: they've measured the volume based on the external dimensions, which seems to be standard malpractice. This gives me real volumes of 97 and 48.5 litres respectively.
The other thing of interest is the weight of the pot, since it enters into the thermal calculations. One option is to measure the volume and calculate it. For the 100 litre pot I get the following details:
| Thickness | 0.4 cm | |||
| Outside diameter | 53.0 cm | |||
| Inside diameter | 52.2 cm | |||
| Height | 45.3 cm | |||
| Area base | 2206.2 cm² | (outside diameter² / 4 * π) | ||
| Volume base | 882.5 cm³ | (area * thickness) | ||
| External volume | 99940 cm³ | (area * height) | ||
| Internal volume | 96946 cm³ | (inside diameter² / 4 * π * height) | ||
| Volume of wall | 2994 cm³ | (external volume - internal volume) | ||
| Total volume of pot | 3876 cm³ | (volume of wall + volume of base) | ||
| Weight of pot | 10465 g | (total volume * 2.7, density of aluminium) |
The problem is, the pot only weighs 5.8 kg, including the handles which I didn't account for here. Take them away and it's reasonable to assume that I have calculated exactly double the real weight. The same applied for the other pot. But how? I can't see anything wrong in my calculations, and though my measurements of diameter and height might be out by a millimetre or two, I measured the thickness with vernier calipers, and they agree with the specifications. Put it to people with time on their hands, the ones on IRC, and they all came up with the same answer within the constraints of rounding.
So what's the answer? Hollow pot? Unlikely. Tapering thickness? It would need to taper to 0 to explain this discrepancy, clearly not the answer. Different alloy? It would have to have a density of 1.35 or so, and alloys that light are expensive. The pots weren't. I'm still puzzled.
PayPal didn't answer my last message, so I send out another one. Incredibly, despite my request to first read the message before inventing a response, I got not one but two repeats of the claim that I had to convert currency manually, one of them pointing to a URL which contained the text:
Buyers:
- Pay for purchases in your selected currency.
- Payment is automatically converted to your desired currency.
- No need to hold a balance in another currency to send a payment.
I can't believe how incredibly incompetent these people are. Is it possible that this is an exception to the adage “never ascribe to malice that which can be adequately explained by incompetence”? More details in the PayPal page.
Message from Michael (the one with no surname) today, recommending Openstreetmap as a replacement for Google Maps:
Probably the best long term option is to get involved with openstreetmap where you can build maps from a blank page, (especially in your area: http://www.openstreetmap.org/?lat=-37.8167&lon=143.7667&zoom=12&layers=B00FT) It'll be a while before openstreetmap starts offering directions, but at least the map looks like it will be reasonably accurate.
Box Hill Golf Club looks ok on openstreetmap: http://www.openstreetmap.org/?lat=-37.83918&lon=145.12436&zoom=16&layers=B00FT You'd have to do your own navigating, but at least it's obvious which road to enter by... Openstreetmap doesn't have street numbers though, and it also doesn't have any information about the no right turn into StKilda Rd.
It sounds a little like a Wikipedia for maps. It would be nice for it to be successful, but I suspect the model will be self-limiting.
I've had an account with LinkedIn for some time now, mainly because people keep asking me to “link” to them. The first to invite me was Nathan Cochrane of The Age, at which point it was so inactive that I resigned and then later rejoined. It still hasn't shown any useful function, but today, after accepting another invitation, I decided to finally download a photo, the same one I put on my home page. That shows yet another case of typical web breakage.
The photo is http://www.lemis.com/grog/Photos/20020202/small/grog-in-office.jpeg. Here's what happened when I tried to download it:
|
|
||||||||
Instead of checking what the file was, it looked at the name. That's bad enough—you can put anything in a file with any name—but it doesn't understand what JPEG means. Maybe it thinks that the people weren't that expert after all.
So I copied the file to /var/tmp/stupid-linkedin-breakage.jpg, which it happily downloaded. Then it gave me an opportunity to “edit” (i.e. crop) it:
|
|
||||||||
The “edit” window is much smaller than the photo, and there's no way to resize it. I can barely make out the outline of the top of my skull, and managed to at least get myself in the cropped area. But the result is only 80 pixels square, so it's completely unrecognizable anyway.
We bought a rabbit at the Prahran Market on Sunday, and today Yvonne made her favourite rabbit dish, lapin à la moutarde. Unfortunately, we made the mistake of using (mild) English mustard, which caused the sauce to curdle:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
It's really difficult to find good foodstuffs here.
| Wednesday, 18 June 2008 | Dereel | Images for 18 June 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Didn't do much during the day. Did a bit of work in the garden and finished the toy greenhouse, which is covered in a clear PVC foil. I suspect it won't last long.
24 hours after my last message, still no response, possibly because there was nothing to misunderstand. Send out another one, effectively with the same demand, just formulated slightly differently. In view of the apparent inability of PayPal personnel to follow texts of more than a couple of lines, followed it up with 5 messages, each containing one paragraph of the message.
I very quickly got a reply which astounded me. It made no incorrect assumptions and seemed to have understood the issue, and it brought a plausible explanation: automatic currency conversions only work if they involve the entire sum. On this occasion my credit balance was in Euros and the bank account was in AUD. It also included a phone number to call, which I'll certainly try next time I have problems: 1800 073 263.
| Thursday, 19 June 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Winter is making itself felt, at least in mood. Still no frost, but there was a surprising amount of mist (called “fog” in Australia) which lifted only slowly, and the whole atmosphere was as dreary as Germany in November. Perversely, spent a bit of time in the garden, but not too much.
One of my latest ideas, with Chris Yeardley's support, is to do proper web hosting for people who can't afford the astronomical prices that “professional” companies such as Vividas charge. The idea is to offer a really cheap basic page (describe who you are, and present it in a form that will interest people) for peanuts, and offer value-added services.
That's nothing new, of course, but we're planning to do it right—and nowadays that seems to be something unusual. By chance I ran into an example of what we don't want to do this afternoon:
|
|
||||||||
That's not hotfrog's fault, of course (well, maybe the emblem at the bottom, which is completely illegible on my 3 MP screen). This page is free, and the information is entered by the user, incorrectly and incompletely: the street name is Beaufort Road, not Beauford, as anybody in the area would notice. But part of our service would be sanity checking.
Warren Toomey recently moved oztivo.net (not to be confused with oztivo.org, which appears to be Godaddy's attempt to cash in on the name OzTivo) to a virtual system in the USA, which runs on Linux. He's quite happy with it so far, but I'd really prefer FreeBSD. That's not chauvinism, just practicality: it's easier for me. Today sent out a message asking for FreeBSD-based hosting, and found a surprising number of offers. Now I need to find out what we need.
What do we need? For the small customers, neither bandwidth nor storage is the issue. For me, at least storage is. Discovered that I am using over 20 GB on www.lemis.com, also known as ozlabs.org, though much of that is wasted space. Still, I currently use over 11 GB for the single web site, and this proves to be in the order of magnitude that the smaller offerings provide. On the other hand, the bandwidth doesn't seem to be the issue; for $20 per month you can get 200 GB of network traffic, but only 12 GB of total storage. Am I underestimating the amount of traffic? In any case, a lot of research to do.
| Friday, 20 June 2008 | Dereel | Images for 20 June 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
There's plenty of work piled up, though nothing really urgent, and somehow I didn't feel like starting anything in particular. Then Edwin Groothuis appeared on IRC and talked about a person who had accidentally had a 40 GB Microsoft image copied onto a 1 TB RAID array on a FreeBSD server. That space was gone, of course, but he wanted the rest back. There were 7 file systems on it, so most should still be intact, but of course the very first thing that gets overwritten is the BIOS partition table.
Dragged out a recovery program that I wrote in August 2002. Of course, it had suffered from bit rot—in particular, for some reason I had copied the relevant kernel header files from that version of FreeBSD, perhaps because the person for whom I wrote it didn't have them installed on his machine—and it didn't handle UFS-2.
Spent the rest of the day working on that, discovering more and more about the history of UFS in the process. Many of the fields in the superblock are no longer in use, and some that appear to contain something don't agree with their comments, notably:
int32_t fs_sblkno; /* offset of super-block in filesys */
int32_t fs_cblkno; /* offset of cyl-block in filesys */
In the file systems I observed, every fs_sblkno was set to 40, and every fs_cblkno was set to 48. The only way I could find the difference between a primary and a secondary super block was because only the primary superblock gets the name of the last mount point stored at fs_fsmnt.
The magic numbers are interesting too:
#define FS_UFS1_MAGIC 0x011954 /* UFS1 fast filesystem magic number */
#define FS_UFS2_MAGIC 0x19540119 /* UFS2 fast filesystem magic number */
As Jürgen Krause observed years ago, that looks like somebody's birthday. But whose? Kirk? Eric? Either would fit. A little research reveals an indirect hint:
Expand timestamp to 64-bits. Current epoch expires on MKM's 84th birthday
That's when the time stamp overflows out of 31 bits:
=== grog@dereel (/dev/ttyp4) ~ 391 -> date -r 2147483647
Tue Jan 19 14:14:07 EST 2038
So yes, it's Kirk's birthday.
By the end of the day, had some improvements. Plenty more to be done, though. The latest version is here.
We're still low on hay—thanks again, Paul Ludovici—and Yvonne set off again to pick some up. Chris Yeardley helped, and came back for dinner. More web site discussions. One thing that came out of that was that Chris has a different (but not incompatible) idea of what we want to do with the web site. She's more interested in designing complete sites for horse people.
The current matter with PayPal is really over, but today I got yet another message, noting that part of the problem had been that I didn't get a choice of how to pay for the item, and promising to forward the matter to the “technical department” for resolution. I don't think that implies that I'll hear back from them, but it's good do see some follow-through.
| Saturday, 21 June 2008 | Dereel | Images for 21 June 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Spent a fair amount of time talking about disk recovery software today, more from the theoretical point of view: the person who needed the help didn't show up again. Maybe he can afford weekends.
Mail on the topic from James Andrewartha:
In the past I've used TestDisk to recover partitions after deleting the partition table. It doesn't seem to have a huge amount of support for FreeBSD though, but the code might be worth a look.
We have yet more mushrooms:
|
|||||||||||||||||||||||
Callum Gibson reported on IRC that he had to go under the house to fix an antenna cable, giving me a nudge to do the same for my network cables. How I hate it! I did get two of the three connected, but each time I had to go out to put more cable through the floor, and after the second one I couldn't face it any more. I don't know why it's so distasteful—at least in winter I have the advantage that there are no redbacks.
| Sunday, 22 June 2008 | Dereel | Images for 22 June 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Finally overcame my revulsion and finished installing network cables under the house. It's not really that bad, but I really needed to force myself to do it. After that, of course, I had to terminate the cables, something I've never done before with Ethernet.
What a pain! I don't know whether it was my combination of Cat-6 cable and the Cat-5 connectors that I got from Chris, or the individual components, or just my lack of ability, but it took forever to get all the wires to go all the way into the correct slots. Crimping wasn't an issue, on the other hand. Finally decided that the easiest way was to straighten out the wires in the correct layout and then cut the ends off even—the lengths tended to be a little different after straightening out. Got the two important cables up and running, then called it a day.
It's interesting, though, to note various opinions I got on the subject of wiring. At the one end of the scale: “It doesn't matter which wire you put where, as long as they're wired 1:1 at each end”; clearly that's suboptimal, since all wires come in pairs. At the other end was the recommendation to keep the pairs twisted as far as possible, one of the reasons why I had so much trouble getting the wires into the connector. In the middle was a discussion about the exact sequence: for some reason, the recommended layout is one pair at each end of the 8 pin connector, one exactly in the middle (pins 4 and 5), and the other one taking up the remaining pins. But then there are two different layouts, T568A and T568B, which differ only by the colours of two pairs. Why? I had suspected crosstalk issues, and the Wikipedia page mentions them too, but that doesn't make sense for 100BASE-T if you only swap the two pairs that carry the signal.
| Monday, 23 June 2008 | Dereel | Images for 23 June 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
The bloke with the trashed disk popped up again today; in the meantime he had been trying recovery with scan_ffs, which is significantly less verbose than recover. With its help he found most of his partitions and was able to access some, but not all, of the data. In particular he had the following issue when running fsck -n on /var:
THE FOLLOWING DISK SECTORS COULD NOT BE READ: 209628224, 209628225, 209628226, 209628227,
/dev/amrd0s1e: INCOMPLETE LABEL: type 4.2BSD fsize 2048, frag 8, cpg 0, size 52428800
That looks like an I/O error, so we took a look at the console messages:
g_vfs_done():amrd0s1e[READ(offset=46631632896, length=16384)]error = 5
Error 5? That's:
=== grog@dereel (/dev/ttypj) ~ 4 -> grep "\b5\b" /usr/include/sys/errno.h
* @(#)errno.h 8.5 (Berkeley) 1/21/94
#define EIO 5 /* Input/output error */
How do you get an I/O error on a RAID system? One possibility is if you try to go off the end of the device; but the RAID device was about 1 TB, and the (sector) address was well within that range. Then it dawned on me: he was talking to /dev/amrd0s1e, a slice. The label looked like this:
# size offset fstype [fsize bsize bps/cpg]
e: 52428800 100663296 4.2BSD 2048 16384 0 # /var
f: 52428800 310378496 4.2BSD 2048 16384 0 # /var/mail
So the slice had the sector addresses from 100663296 to 153092095, and here the disk data structures were pointing to sector 209628224, which was in the empty space between partition e and partition f. I had already wondered about these spaces; this was a fairly clear indication that they were part of the partition, and thus the partition size was wrong. umounted the file system and changed the size to:
e: 209715200 100663296 4.2BSD 2048 16384 0 # /var
That used up all the space to the beginning of partition f. Bingo! All the data was back.
Is this a bug in scan_ffs? Hard to say. I know that recover doesn't currently try to guess the partition size; my approach has been to look at the partition table offsets and work out the sizes manually. Anyway, nice to know that we got back almost all the data.
More work in the garden. Our Olearias have been flowering almost non-stop since they got water:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
There was also a second one of a different kind, which bloomed yellow, but which looked old and not very happy, so we pulled it out. That didn't stop it, of course, and recently a number of shoots have been coming out of the bed, so today we transplanted them:
|
|||||||||||||||||||||||
There's another photo of where we planted them, but it really needs to be enlarged significantly to show the stakes. The two stakes at the left and the one at the right are the yellow ones; the other one is a white one which we transplanted a week or two ago.
Also put some plants in our toy greenhouse:
|
|||||||||||||||||||||||
Toy or not, it was noticeably warmer inside, even though the sun hadn't been shining. We'll have to keep an eye on things when the weather gets warmer.
The kangaroos are back:
|
|||||||||||||||||||||||
This was only one of five, apparently sent by the others (who remained about 50 m away) to case the joint. He wandered off when I came closer. I wonder if this presages more damage in the garden.
Spent the rest of the day finishing the rearrangement of the lounge room, which had been waiting on the underfloor cabling. I'm getting more proficient at terminating network cables now; the trick seems to be to ensure that all the wires are absolutely straight at the end, so that the slide nicely into the holes in the connector.
By evening we had got rid of most of the over-floor cables. We still have a telephone cable that I need to puzzle about, and I don't want to put the loudspeaker cables under the floor until I know exactly where they go. But things are looking a lot tidier.
| Tuesday, 24 June 2008 | Dereel | Images for 24 June 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Somehow spent most of the day in the office, including writing up stuff about the garden and following up on my options for web hosting. Also hung some pictures—I've set myself a soft goal of making the house look like we've completed moving in by the first anniversary. That was more frustrating than I expected—why are picture frames such a mess?
| Wednesday, 25 June 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Another day with little to show. A bit of work in the garden; finally mowed the lawn. I fear our ride-on mower is not long for this world. The simplest problem is that the battery is pretty dead, but the steering also looks worn out, and various bolts are missing. I wonder what the condition of the engine is—it seems to have much less traction than when we got it.
It's been very windy, and our toy greenhouse looks like it won't hold out too well. One of the joints came apart; possibly I should tie the corners together so that that can't happen.
Also pruning a Salvia microphylla. I've read various ideas about how to prune Salvias, some of which suggest that they don't respond well, but in the end had to do it anyway to get at the grass that was growing from between the roots. To my surprise, it wasn't one salvia, but many: it seems to create underground roots. Also, even the established canes had new leaves coming out close to the base, so I pruned back to there. I strongly suspect I'll have no trouble. Pulled out a couple of the underground shoots, and planted one; I suspect it'll do fine too.
| Thursday, 26 June 2008 | Dereel | Images for 26 June 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
Into Ballarat today to do a lot of shopping. My electric razor has died—yes, I do have one, but it doesn't get much use, and it survived nearly 20 years. Looking for a new one was more complicated than I thought: the prices are more than I expected, and most don't even have a beard trimmer. I suppose the days of the three-day beard are over, and people are clean-shaven again.
Also to Enzed to buy components for my HERMS system, and pretty quickly found everything I needed at a price I didn't need. It wasn't until I was on my way home that doubts set in; it looks as if I'll have to replace some of the stuff with other alternatives. Why is it that this is so complicated?
One thing that I did find was a 2200 W electric heating element for the HLT, embedded in a hot water jug. This kind of elements normally costs in the order of $70 to $80; embedded in the water jug, it cost $7.80. Now I just have to find how to get it out and into the side of the HLT, another unresolved issue.
Chris Yeardley along in the evening, and we noted that she's currently only second on the list of 898,000 hits for the Google search for +Chris +Lilac. Did what we could to improve matters:
|
|||||||||||||||||||||||
| Friday, 27 June 2008 | Dereel | Images for 27 June 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
My uncle Max sent me an Email with an attachment of type video/x-ms-wmv; clearly a Microsoft format, but at least the MIME type was wrong. And I was, once again, left with this problem that I had to configure my mutt to display it.
But how? What's the name of the file. mimetypes? Went looking and found a total of 2100 files with names containing “/mimetypes”. It was in /usr/local, so much I recalled, but looking around didn't help much. Finally, as so often before, decided to ktrace the mutt process and see what it was reading. A good thing, too: why put the information in mimetypes when it would fit just as well (and with less chance of being found) in /usr/local/etc/mailcap?
The problem was, I had entries in mailcap:
audio/*; showaudio %s
audio/basic; showaudio %s; compose=audiocompose %s; edit=audiocompose %s
video/*: mplayer %s
video/x-ms-wmv: mplayer %s
The last line is completely superfluous, since video/* should also include it. But why didn't either of them work? Tried with an MP3 attachment, of Aligi Voltan playing “my” Savary jeune of 1847, and that worked fine, suggesting that the configuration info was in a different file.
Back to ktrace: where did it get the configuration information from? To my surprise, it was) in /usr/local/etc/mailcap again. So why didn't I find it? Two reasons:
I had used the wrong syntax. Most configuration files of this nature separate the fields with a colon (:); this one uses a semicolon (;).
The software completely ignored the error, so that I couldn't find out about it easily. How I hate software that does that! And it seems to be on the increase (“don't scare the user”, with the implication “confuse the hell out of him instead”).
With this simple patch, including removing the redundant entry, things worked again:
--- /usr/local/etc/mailcap 2006/08/27 15:31:12 1.6
+++ /usr/local/etc/mailcap 2008/06/28 02:15:53
@@ -27,8 +27,7 @@
-video/*: mplayer %s
-video/x-ms-wmv: mplayer %s
+video/*; mplayer %s
We've decided to try a virtual server from RootBSD, and today I got the configuration details. Getting into the system was straightforward enough, but there was nothing in the way of ports installed. Installed bash pretty quickly, but Emacs was another matter. On the one hand, the tarball downloaded in record time:
=> Attempting to fetch from http://ftp.gnu.org/gnu/emacs/.
emacs-22.2.tar.gz 100% of 36 MB 3402 kBps 00m00s
That's a far cry from the speeds I can get here. But building took forever, over 3 hours (at least partially due to the kitchen-sink nature of Emacs, which installed about 60 other ports), and at least partially because it was the second thing to get installed. But clearly my virtual CPUs (yes, there are two of them) aren't the fastest.
While doing that, also finally copied some MPEGs from teevee to DVD. I've been meaning to burn “real” DVDs, but getting the Makefile together has been a lower priority, so for the time being I'm just burning ISO-9600 file systems. Freed up about about 40 GB of disk space, with more due tomorrow.
The kangaroos are more and more evident. These ones are more timid than usual and bounced off when I came closer:
|
|||||||||||||||||||||||
They'd be OK if they stayed on that side of the fence; hopefully they will.
| Saturday, 28 June 2008 | Dereel | Images for 28 June 2008 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
A little work in the garden, connecting the bore pump directly to the horse trough with a 1 inch pipe. To my surprise, the flow was so strong that it stalled the pump; we'll have to avoid opening the valve fully until we have proper fittings at the end.
On with the installation on the new web server today, and things progressed pretty well. Over to Chris' place in the afternoon, where things didn't progress as well: somehow the ssh configuration on her Apple is messed up, and I need to work out why rather than just redo the configuration.
The real issue, though, is the DNS hosting. Chris is with Ausweb, who live up to the normal standard of DNS registrars. After logging in, she found a link to “Modify DNS”, which looked promising—until we discovered that Ausweb thinks that “DNS” stands for “Domain Name Server” and not “Domain Name Service”. And that's all they offer: how to add or remove a name server. No option to maintain the zone, which is currently served only by Ausweb.
It seems that DNS has been renamed to “Domain Name System”, a renaming with which I'm not the only one to disagree, as the Wikipedia talk page shows.
So we tried to add w3.lemis.com as a name server. That failed with a message I haven't seen before:
What's that? It happened for all lemis.com addresses, but not for others, so possibly it's something to do with GANDI. Possibly the intention is to require my authorization before declaring one of my systems as a name server for another domain, but I can't find anything relevant on GANDI's broken web site, and as long as I can add any A record I want, it doesn't help much anyway.
In the end gave up and declared ozlabs.org as a name server and went home and entered a ticket with Ausweb. By the time I had entered that, the new name server was visible, so set up DNS to make it a slave to the Ausweb name servers. That didn't work either:
Entered a ticket for that one, too, and got mail confirmation with some suggestions to use the “knowledgebase”:
|
|
||||||||
Following those links brought a surprisingly empty suggestion:
|
|
||||||||
This is the only relevant part of the complete display. As I suspected, the message was sent as multipart/alternative, and presuambly intended to be read as HTML, so took a look at that. To my surprise, that was even worse. The links had no text:
|
|
||||||||
Looking at the (marvellously indented) HTML shows that that's what they specified:
<tr>
<td><font size="2" face="Verdana, Arial, Helvetica, sans-serif"><a h
ref="http://ausweb.com.au/helpdesk/index.php?_a=knowledgebase&_j=questiondetails&_i=&ticke
tid2=UWO-20351&ticketkey2=a0 d57a2c&doclose=1">.</a> <a href="http://ausweb.com.au/helpdes
k/index.php?_a=knowledgebase&_j=questiondetails&_i=&ticketid2=UWO-20351&ticketkey2=a0 d57a
2c&doclose=1"></a></font></td>
</tr>
<tr>
<td><font size="2" face="Verdana, Arial, Helvetica, sans-serif"><a h
ref="http://ausweb.com.au/helpdesk/index.php?_a=knowledgebase&_j=questiondetails&_i=&ticke
tid2=UWO-20351&ticketkey2=a0 d57a2c&doclose=1">.</a> <a href="http://ausweb.com.au/helpdes
k/index.php?_a=knowledgebase&_j=questiondetails&_i=&ticketid2=UWO-20351&ticketkey2=a0 d57a
2c&doclose=1"></a></font></td>
</tr>
This display is preformatted and wrapped at 90 characters so that it doesn't go off the edge of the window. This is sub-optimal, but I can't find a better way to do it: if I just set the font and put in explicit line breaks, I lose the indentation. If I change the blanks to , I lose the ability to wrap. If you have ideas about how to use the screen width without losing indentation, I'd like to hear from you.
I suppose the messing around with font sizes (making them smaller) is par for the course. But there was only a single dot to represent the link, so it didn't make much difference. And, of course, I got the same empty document. No wonder it was rated 0.
Chris is used to using CPanel to maintain her web site. I didn't find it in the Ports Collection, for a good reason: it's commercial. Not only that, but it costs Real Money, well over $500 per license. So we set off looking for alternatives. So far, with the help of Callum Gibson, we've found a number of leads to follow up: there's stuff at the end of the Wikipedia article, also a discussion at http://lordmatt.co.uk/item/966/ , and there's a thing with the FreeBSD friendly name Penguinator. That'll be for Chris to examine.
Despite the name server problems, spent some time setting up virtual hosts, using names like narrawin.lemis.com and pasocentral.lemis.com. Rather than refer to my previous httpd.conf files, I read the comments and set them up accordingly. The result?
Why? The entries all looked right. Spent some time trawling the web, finding the usual mixture of misinformation and different problem, before I finally found the clue: if you use virtual hosts, you must specify the NameVirtualHost directive. There's nothing obvious in the comments to tell you:
+++ httpd.conf 2008/06/29 04:56:02
@@ -1025,7 +1025,7 @@
#
# Use name-based virtual hosting.
#
-#NameVirtualHost *:80
+NameVirtualHost *:80
| Sunday, 29 June 2008 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
It's not overly cold (day temperatures round 8°), but the wind is biting, and this badly insulated house does nothing to make things more comfortable. An ideal day to work indoors.
Mail from Stephen Rothwell today, reporting the zone transfer errors for narrawin.com. Nothing from Ausweb, of course. Seems they sleep at weekends.
It occurred to me that Ausweb weren't involved in the DNS for pasocentral.org, so set to to play around with that. That was more pain than I had expected, certainly not helped by the slow satellite connection. After setting up the web server to point to our new server, I reloaded the name server, and nothing changed.
That's usual a serial number problem, but what we had was:
named.conf:
zone "pasocentral.org" {
type master;
file "db.pasocentral.org";
db.pasocentral.org:
pasocentral.org. IN SOA ns.pasocentral.org. grog.lemis.com. (
2008062901 ; serial
nslookup:
=== grog@ozlabs (/dev/pts/2) ~/etc/bind 23 -> nslookup -q=any pasocentral.org
Server: 127.0.0.1
Address: 127.0.0.1#53
pasocentral.org
origin = ns.pasocentral.org
mail addr = grog.lemis.com
serial = 2005022001
It took quite a while to work out that the path name for db.pasocentral.org was not relative to the directory in which both files were located. Stephen has a distributed DNS configuration, and each ozlabs user has a directory ~/etc/bind (and not, as you might expect, ~/etc/named), but the main configuration files are kept in /etc/bind, where I read:
=== root@ozlabs (/dev/pts/6) /var/log 19 -> cat /etc/bind/named.conf.options
options {
directory "/var/cache/bind";
And, of course, there was a db.pasocentral.org file there. Changed the pathname:
--- named.conf 2008/06/29 00:34:59 1.12
+++ named.conf 2008/06/29 05:05:39 1.13
@@ -1,3 +1,4 @@
...
zone "pasocentral.org" {
type master;
- file "db.pasocentral.org";
+ file "/home/grog/etc/bind/db.pasocentral.org";
And it still didn't work! This seems to be a bug in named: it didn't notice the different name until I removed the zone and added it again.
By the evening we still didn't have a reply from Ausweb, so did a bit of guessing what the zone might look like. Chris told me that she also had ftp access (apparently that's how they expect their customers to upload their web pages!), so I went and tried ftp.narrawin.com. It timed out. On further investigation, it proved that ftp.narrawin.com is really their primary name server:
=== grog@dereel (/dev/ttypi) ~ 101 -> nslookup -q=any ftp.narrawin.com ns1.ausweb.net.auServer: ns1.ausweb.net.au Address: 122.252.5.25#53 Name: ftp.narrawin.com Address: 202.155.174.209=== grog@dereel (/dev/ttypi) ~ 102 -> nslookup 202.155.174.209Server: 192.109.197.135 Address: 192.109.197.135#53 Non-authoritative answer: 209.174.155.202.in-addr.arpa name = ns1.ausweb.net.au.
As her signup letter told her, the real ftp server is at a different name. Given the insecure nature of the matter, I'm not going to reveal it in public. But what are Ausweb thinking by specifying their primary name server as ftp.narrawin.com, especially since (correctly, for once), they don't have the FTP port open?
Back home, tried one likely candidate for a CPanel replacement, webmin. The installation instructions are a little rudimentary:
To reconfigure webmin you should
run the following command as root:
${LOCALBASE}/lib/webmin/setup.sh
You won't have to perform this step
after every webmin upgrade.
Since 1.150_2, to run webmin from
startup, add webmin_enable="YES"
in your /etc/rc.conf.
It seems to be a bug that the make variable ${LOCALBASE} didn't get expanded (to /usr/local), but there are a couple of things that this message does not tell you:
After working that out, things worked fine. If you like web-based administration, it might even be good.
| Monday, 30 June 2008 | Dereel | |
| Top of page | ||
| previous day |
Mail from Ausweb this morning, answering one of my tickets. Here the statement and the reply:
We've been trying to add name servers to the zone narrawin.com, which you host. For some
reason, all attempts to add *.lemis.com fail with the message;
com.primus.tld.nameserver.NSChangeException: completed: NO {
com.primus.tld.registry.RTYException: completed: NO: 2201 Authorization
error }
I don't see anything at our end which is causing this. Please resolve ASAP.
Reply:
Your nameservers are set correctly to point to the server you are on however there is also a rogue entry pointing www.ozlabs.org. Can you please clarify why exactly you are wanting to change/add name servers to your existing domain.
So I replied and pointed to the other ticket, but it never got there because their web interface stripped off the information. Checked the ticket and found it had been closed, apparently because I looked at the empty “knowledgebase” article on Saturday. Left a comment in the ticket and received a call from Matt, who acknowledged that their knowledgebase was being updated. He also didn't understand why we wanted to add name servers; how would hosts know which one to access?
He then told me that we can't access the zone files, because this was a shared (web) server. I was unable to make it clear to him that web servers and DNS have nothing to do with each other He told me that it was too dangerous to let users update their DNS, because they could mess it up, and that I should get a DNS manager. On my asking, he explained that this is somebody else who can manage the DNS for us. I asked about ftp.narrawin.com and pointed out the obvious issues (you don't put ftp on a name server, but that's where it was pointing, and ftp service was available on a different server which isn't called ftp, but he insisted that this was correct.
When I continued asking, he put me on hold and then passed me to Peter, who didn't announce his name. This was clearly the same Peter who had also sent me a couple of messages:
Please do not double post for the same issue.
www.ozlabs.org is not a name server and as such is corrupting the zone file, please delete your incorrect input!
I don't know how he decided that these were both the same issues, since it's clear that he hadn't read the messages. And of course ozlabs.org isn't a name server—it's a host. But my subsequent discussion with Peter showed that he doesn't really understand DNS. When I told him that the name of the host was unimportant, he got up on his high horse and told me I shouldn't argue with people who had been doing this for years. When I pointed out that I had written a book chapter on this over 10 years ago, he said “OK, then read your book if you don't believe me” and hung up. Somehow it's typical that he didn't even understand what I said.
Just to be 100% sure, but also because it's the right thing to do, we changed the name of the server from www.ozlabs.org to ns.ozlabs.org (same machine), and of course we had the same problem. So, at the end of all of that: I still can't add a lemis.com name server, and we still can't transfer the zone from Ausweb. But since we don't have access to update the zone files anyway, the second issue doesn't seem to make much difference: we have to use our own zone files anyway.
About the only thing that I can get out of this is that there are even more stupid people out there than I thought. To make matters worse, Peter (the arrogant know-it-all) is the manager. As Stephen wrote yesterday:
It's amazing how many hosting services/ISPs are so ignorant about something so basic to the success of their business.
On the issue of security, ran nmap against one of Ausweb's web servers. The results were amazing: 1499 open ports! I've never seen that before. It took a while to realize that somebody else was to blame: BST, Boosted Satellite Transmission. It accepts all connections first and then checks whether it can really fulfil them. If not, it returns an ECONNRESET. Still, the issues I have are enough to warrant a products and services page about Ausweb.
It's just a pity that Chris has paid them for another year of disservice.
That didn't take all day, but after that I didn't really feel like doing much else. Played around with the web server, which really works quite easily. About my only gripe is that the root control panel is so difficult to read:
|
||||||||||||||||||||||
This kind of display works acceptably at resolutions up to 1024x768, but beyond that it's very difficult to read, comparable to the display of the “thumbnail”.
Do you have a comment about something I have written? This is a diary, not a “blog”, and there is deliberately no provision for directly adding comments. It's also not a vehicle for third-party content. But I welcome feedback and try to reply to all messages I receive. See the diary overview for more details. If you do send me a message relating to something I have written, please indicate whether you'd prefer me not to mention your name. Otherwise I'll assume that it's OK to do so.
| Top of page | Previous month | Greg's home page | Today's diary entry | Next month | Greg's photos | Copyright information |
