Greg
Greg's diary
September 2002
Translate this page
Select day in September 2002:
Su Mo Tu We Th Fr Sa
1 2 3 4 5 6 7
8 9 10 11 12 13 14
15 16 17 18 19 20 21
22 23 24 25 26 27 28
29 30
Select month:
2002 May Jun Jul Aug
2002 Sep Oct Nov Dec
2003 Jan Feb Mar Apr
Today's diary entry
Diary index
About this diary
Previous month
Next month
Greg's home page
Greg's photos
Network link stats
Greg's other links
Copyright information
    
Groogle

Sunday, 1 September 2002 Melbourne Images for 1 September 2002
Top of page
next day
last day

Up early this morning to help registrations, which went without any problems. The network worked surprisingly well, with the exception of DNS. Spent some time thinking about that and finally came to the conclusion that there was a firewall which was blocking UDP port 53, and probably all UDP, somewhere upstream. While thinking about that, noted that tcpdump insisted on incrementing port numbers on each hop, which made it difficult to use to investigate that kind of problem. Did a bit of work and came up with a patch to traceroute.c to stop that incrementation (-0 flag).

The others decided that the network required rebuilding, which took most of the day, during which we didn't have any network connectivity. Spent some time investigating the slides for my presentations on Wednesday. I was able to rebuild them, but for reasons I couldn't fathom, they ended up rotated through 90°. Took a lot of time trying to debug that, but by the end of the day was not much further.

In the evening off to Young and Jackson's on the corner of Flinders St and Swanston St to visit Chloe, then off to Little Bourke St for a cheap but surprisingly good Chinese dinner. Despite the impression, not too early to bed.


Monday, 2 September 2002 Melbourne Images for 2 September 2002
Top of page
previous day
next day
last day

In the course of the day, finally got the network up, but I still couldn't get my mail. After some telephoning with Yvonne, discovered that echunga.lemis.com had wedged, and at this distance didn't have much choice but the Big Red Button. Why do these things always happen when I'm away from home?

Still busy with my documentation, which took quite a bit of investigation. Finally reinstalled ghostscript version 6.52, which did the trick. I'm left with the question whether the difference represents a bug, a feature, or a bug fix. I suspect it's an attempt to optimize things.

In the evening out eating with people at a pizza place that I should remember in order to be able to avoid it in the future. Then out to a pub on Russell St where I thought that John Terpstra was liable to get himself thumped by a nearby drunk. Back with Adrian, whom we had decided should stay at the hotel until the end of the conference.


Tuesday, 3 September 2002 Melbourne Images for 3 September 2002
Top of page
previous day
next day
last day

Last day of the tutorials, and also the board meeting, where we barely managed to raise a quorum. Managed to get a fair amount of work done in time, including symbolically presenting Liz with a stockwhip.


This should be liz-greg-1.jpeg.  Is it missing?
Image title: liz greg 1          Dimensions:          1600 x 1200, 516 kB
Make a single page with this image Hide this image
Make this image a thumbnail Make thumbnails of all images on this page
Make this image small again Display small version of all images on this page
All images taken on Tuesday, 3 September 2002, thumbnails          All images taken on Tuesday, 3 September 2002, small
Diary entry for Tuesday, 3 September 2002 Complete exposure details

 

It didn't take Liz long to get the hang of things, with somewhat unexpected results:


This should be liz-greg-3.jpeg.  Is it missing?
Image title: liz greg 3          Dimensions:          1600 x 1200, 527 kB
Make a single page with this image Hide this image
Make this image a thumbnail Make thumbnails of all images on this page
Make this image small again Display small version of all images on this page
All images taken on Tuesday, 3 September 2002, thumbnails          All images taken on Tuesday, 3 September 2002, small
Diary entry for Tuesday, 3 September 2002 Complete exposure details

 

Talked about War Chalking to draw attention to the conference, but decided against it. Mark White bet me $20 that it would happen anyway before the end of the conference.

After the meeting, I had thought that I would be able to get a bit of time off, but somehow managed to spend the whole afternoon with a fair amount of stress.

In the evening to the Speaker's reception, after which Gordon Hubbard and I went to a very nice (but expensive) restaurant to meet Steve Alford of the NOIE, with whom we discussed an AUUG involvement in the use of Open Source in the Commonwealth Government. Gave us a lot of food for thought.


Wednesday, 4 September 2002 Melbourne Images for 4 September 2002
Top of page
previous day
next day
last day

Today the conference started. Things got off to a relatively good start, less stress than yesterday, and got my paper presented early.

As Mark had predicted, somebody war chalked us during the night. Mark swore it wasn't him, and considering he took my $20, I suppose I have to agree with him.

Annual General Meeting after the last session, myself as chair. It's rather interesting how much of a fool you can make of yourself.

After that to the networking reception, after which we had had enough to eat, so off for a drink or two at Young and Jackson's in order to introduce more people to Chloe. I talked about a story I had experienced on 25 September 2001: Peter Salus had talked about a stencil which had proved to be in the possession of Warren Toomey (then present). Peter had claimed that the stencil represented Rob Pike (who, as it happens, is here at this conference), but Greg Rose was there and filled in the missing link: this was round the time that AT&T adopted the death star emblem, and also the time that Peter Weinberger (not Rob Pike) had become a manager and apparently stopped thinking. They had used the stencil to paint a big death star like image of Peter on a nearby water tower and even charged the cost of the paint to his department. Some time later, Greg Rose received anonymously via mail in Australia. He has his suspicions who sent it, but he's not prepared to publish. He gave it to Warren Toomey. It's really nice to be able to fill in these stories at successive conferences.


Thursday, 5 September 2002 Melbourne Images for 5 September 2002
Top of page
previous day
next day
last day

Most of my duties were over yesterday, so I had time to relax a little today, though my usual barometer of such things, my list of unread mail messages, didn't reflect the fact.

Talked to Rob Pike today about the stencil story, since Peter Salus had implicated him. Without knowing what Greg Rose had said yesterday, he said that it was Peter Weinberger, not himself. That seems convincing enough.

Our dinner on Tuesday night wasn't without consequences: today we got a phone call from the office of Senator Richard Alston in Canberra, wanting to know more about us. It looks like Gordon will handle that one. It's certainly an indication that we have managed to raise our profile.

In addition to that, Nathan Cochrane of The Age came along to take a look at the conference. He certainly corrected my impression of the technical press: I took him to show the war chalking, and he knew exactly what I was talking about and even which programs to use.

In the evening had the conference dinner, which must have been something of a record. As usual, we had pre-dinner drinks in the foyer, and it became evident that there is a bit of a miscommunication: drinks started at 19:00, and the intention was to have people enter at 19:30, but nobody had told them that. Another one for the procedure guide we should be writing. When people came in, they seized on the balloons we had on the tables, and within five minutes we had some pretty impressive constructions, including one that later bridged two tables 10 metres apart. I put together a (non-BSD) team to judge the entries (Sarah Bolderoff, John Terpstra and Catherine Allen), and they chose one reading "BSD" as the winner and another one, also reading "BSD" as the runner up.

We must be getting old. We had been discussing the room party for weeks, including whether we should rent the presidential suite for the event (I decided against). Nobody wanted it in their room, so finally we found out Andrew McRae's room number and congregated there. Somehow it's the same old crowd, and I didn't stay very long.


Friday, 6 September 2002 Melbourne Images for 6 September 2002
Top of page
previous day
next day
last day

Up a little later this morning; I wasn't the only one. Had enough time to listen to a few papers, though I still had plenty of other stuff to keep me busy. Finally managed to find a two papers, both of which I wanted to hear, on at the same time.

People in Canberra keep calling. Today it was another bloke from the NOIE, wanting to know if AUUG is really the premier UNIX organization in Australia. Gordon didn't have much difficulty in convincing him; looks like he'll have another followup in Sydney.

The stockwhip seems to have been quite a success. Liz reports a feeling of power coupled with a look of respect in all whom she passes. Various people have been looking at it and trying it out. Rob Pike took a particular liking to it. Tried to get some photos of him, but this only highlighted one of the biggest weaknesses of digital cameras, especially when using flash: the delay from pressing the shutter until exposure is often over a second:


This should be pike-1.jpeg.  Is it missing?
Image title: pike 1          Dimensions:          2048 x 1536, 743 kB
Make a single page with this image Hide this image
Make this image a thumbnail Make thumbnails of all images on this page
Make this image small again Display small version of all images on this page
All images taken on Friday, 6 September 2002, thumbnails          All images taken on Friday, 6 September 2002, small
Diary entry for Friday, 6 September 2002 Complete exposure details

 
This should be pike-2.jpeg.  Is it missing?
Image title: pike 2          Dimensions:          2048 x 1536, 751 kB
Make a single page with this image Hide this image
Make this image a thumbnail Make thumbnails of all images on this page
Make this image small again Display small version of all images on this page
All images taken on Friday, 6 September 2002, thumbnails          All images taken on Friday, 6 September 2002, small
Diary entry for Friday, 6 September 2002 Complete exposure details

 

That's really something that needs work.

AUUG conferences die suddenly at 17:00 on Fridays. Managed to drag this one out a little longer by having a raffle of some books that Pearson Education had donated, but ten minutes later things were fading fast, and we even had our incredible mess of things which needed shipping sorted out by 18:00. Off to dinner in Little Bourke St with a few stragglers, most from the 08 conspiracy, and ended up at a Szechuan restaurant called the Post Deng, apparently in honour of Deng Xiaoping, where we had a reasonable dinner. Seems he has finally become an AUUG member, along with a colleague of his.


Saturday, 7 September 2002 Melbourne –> Bendigo –> Echunga
Top of page
previous day
next day
last day

Up and out of the hotel today without too much difficulty. The train back to Bendigo wasn't as pleasant an experience as on the way down: it was raining, there was no waiting room on the platform, and I had to wait 20 minutes on the platform in the drizzle. The train itself was in pretty terrible condition.

By the time I got to Bendigo, I still hadn't been able to contact Gloria Dunbar, an aunt whom I haven't seen for 30 years. Yana and I had intended to visit her this afternoon, but we agreed that we were probably too tired, and so didn't try again, just drove back to Echunga. It was pretty windy, and as a result a lot more tiring.

Back home, Yana reported an NFS hang on her laptop, which is connected via wireless on air-gw.lemis.com. Discovered that it was not responding, and that I couldn't talk to sat-gw either, so we were pretty much off the net, and had been for up to 3 hours. They were both alone on the same switch, so it looked like the connection was down; took quite a while to realise that both machines had crashed in their own manner, not made easier by the fact that both of them were headless. What is it about AUUG conferences that causes my hardware to crash? We had echunga on Monday, and last year the electrics and UPSs joined forces to take multiple machines down.


Sunday, 8 September 2002 Echunga Images for 8 September 2002
Top of page
previous day
next day
last day

The day after returning from the conference is always uneventful. Spent most of the day reducing my mail backlog from 4,800 to 250. I get the impression that people just wait for me to go away from home to increase the levels of mail requiring answers to unmanageable levels.


Monday, 9 September 2002 Echunga Images for 9 September 2002
Top of page
previous day
next day
last day

Another day spent in catch-up mode. I had time, but not the inclination, to do more work on my book. It's funny to have nothing pressing, though cleaning up my mail backlog is still a lot of work.


Tuesday, 10 September 2002 Echunga
Top of page
previous day
next day
last day

Things are gradually getting back to normal now. Was pleased to see that Nathan Cochrane did an article about last week's conference, including the War Chalking.

Started thinking about next month's “Trawling the Ports Collection”, and thought that it would be a good idea to install Wine, the Microsoft “Windows” emulator. Well, it would be a good idea to have it installed and working. In the middle of unpacking the Microsoft file system, had no less than three panics, which proved to be due to Yet Another Dying Disk. Worked around that and got Wine installed, but there was no documentation beyond the alibi man page. Searches confirmed that there's plenty of docco, but the port ignores it, and the tools necessary for building it are not present. Looked on the web site, but the tarballs they had there did not include the user documentation. Sure, it's on the web site as HTML, but there must be 100 pages to point and click your way through. What a pain.

Spent some time on the book. It's due next week, so it's certainly not too early.


Wednesday, 11 September 2002 Echunga
Top of page
previous day
next day
last day

I had intended to do more work on the book today, and managed a little, but there's this hint of spring in the air, and Guy came along today to continue tidying up the property. The shed was empty enough to start moving vehicles in, starting with the Citroën CX, which proved to be more difficult than we expected: even with jumpers from the tractor, we were unable to get the starter motor to move fast enough to start it. Started pushing it out and discovered that the hand brake was jammed on, so ended up pulling it out with the tractor, which didn't really care whether the wheels moved or not, though the brakes did unseize. After a bit of work, including a superficial wash, managed to get the shed to look relatively tidy.

Simon Hackett along in the evening, amongst other things to try to sell me an 802.11 wireless Internet connection to Mount Lofty, about 14 km away. He finally accepted that I was probably right that that wouldn't be a solution: the installation costs would be several thousand dollars. Talked about ADSL, and I discovered to my surprise that Telstra insist on a voice carrier on all DSL lines, though it substantially reduces the range that the cable will carry. From a technical standpoint, I could quite possibly get a DSL connection here if there weren't the requirement of the voice carrier. I'm surprised that there hasn't been more talk about this.


Thursday, 12 September 2002 Echunga Images for 12 September 2002
Top of page
previous day
next day
last day

Up late this morning after last night's late night, and almost immediately got roped in to a one-hour conference call with the USA, with particularly poor communication quality: I could only really understand one of the participants. I wish people wouldn't use conference calls to handle routine things.

After that into town to the ADUUG lunch, this time in the Austral Hotel off Rundle St. I think I'm getting fed up with pubs; self-service always takes for ever and holds up the conversation.

After that down to a camera shop looking for second-hand telephoto lenses, which I wanted for taking photos of the wildlife on the pond. Prices were strange. Most were in the $200 to $300 range, but I found an f/4 100-240 mm zoom lens for only $75; unfortunately, there was something wrong with it, and I couldn't focus to infinity at the short end of the focal length range. Ended up buying it for $10 (Australian) in the hope that I could repair it. If not, I have a fixed-focus 240 mm f/4 lens for that price, which can't be bad. Also bought a fixed focus 300 mm f/5.5 lens for $75.

At home, found nothing obviously wrong with the zoom lens. It also has a macro setting which uses the focal length ring to focus closely, so I suspect it is somehow maladjusted to be permanently in partial macro position. Googling proved pretty useless; it's a Sun 100-240 mm lens, f/4, and that doesn't give much to search on. I suppose I'm going to have to learn to repair lenses the hard way, just to save a few dollars. I must be crazy.


Friday, 13 September 2002 Echunga
Top of page
previous day
next day
last day

Not much work today. Googled around a bit looking for instructions on repairing lenses, but didn't find anything immediately relevant, though the issue seems to be pretty common, especially amongst lenses for video for some reason. Started dismantling the lens, but couldn't find how to get at the right part, so put it back together without the infinity stop, which enabled me to focus beyond “infinity”, giving me sharpness at infinity between 135mm and 240mm, which should be enough.

More mail cleanup, and I'm gradually getting things together. Also some serious work on the book, which makes me feel a bit happier.


Saturday, 14 September 2002 Echunga
Top of page
previous day
next day
last day

Another day with little to report: back to work on the book, and also tidying things up. Maybe I'm getting to a little more sensible work load. Looking back over the year so far, it looks as if the deadlines I had in January are the same that I have now. I need to get to a workload which enables me to get things done roughly on time.


Sunday, 15 September 2002 Echunga
Top of page
previous day
next day
last day

More work on the book. It has suddenly dawned on me that 5.0 is due any time now, probably before the book appears. I'll have to rework it all with that in mind.

NetBSD 1.6 has now been released. Spent some time trying to build it, and came to the conclusion that something is seriously wrong with the CVS server I'm updating from (anoncvs.NetBSD.org). I ended up with lots of messages like this:

cvs server: Updating bin/systrace
cvs update: move away bin/systrace/Makefile; it is in the way
C bin/systrace/Makefile
cvs update: move away bin/systrace/alias.c; it is in the way
C bin/systrace/alias.c

This occurred for many other directories as well. In this case, I renamed the directory and picked up a new copy, after which I compared the files. All the files were identical, but the timestamps in CVS/Entries were not:

--- systrace.old/CVS/Entries    Sun Sep  1 02:41:56 2002
+++ systrace/CVS/Entries        Sun Sep 15 16:06:45 2002
@@ -1,22 +1,22 @@
+/Makefile/1.6/Mon Aug 19 09:56:03 2002//
-/Makefile/1.6/Sat Aug 24 17:11:31 2002//
+/filter.c/1.5/Wed Aug 28 03:52:44 2002//
-/filter.c/1.5/Sat Aug 31 17:11:55 2002//

This diff is truncated and rearranged to make it clearer; a large number of files show different modification timestamps. Discussed this on IRC and discovered that there is an anonymous CVS server at cvsup.de.NetBSD.org, which on further examination proved to be a CNAME for baerenklau.de.FreeBSD.org. Started a new cvsup from there.


Monday, 16 September 2002 Echunga
Top of page
previous day
next day
last day

Came into the office this morning and found my cvsup from the previous evening still running, having transferred approximately a gigabyte of data in the process. That will cost me real money.

Finally the cvsup finished, and I checked things out and promptly had a build failure. I discovered I only had two directories in src/usr.sbin, those contained in $CVSROOT/syssrc/usr.sbin. I moved this directory aside, and in the parent directory I did another cvs co, this time getting the directories in $CVSROOT/basesrc/usr.sbin, but not the one in $CVSROOT/syssrc/usr.sbin. Merged it by hand and sent out a message to the NetBSD mailing list asking what was going on.

While doing that, got a call from Yana, who was in Adelaide. The power steering of the Magna had sprung a leak. Well, at least it didn't do it last week on the way to Bendigo or back. Into town with the Commodore, bought a litre of fluid and poured it in, only to watch it trickle out again. Off to the repair shop, where they diagnosed a loose clip, but left it there for repair anyway. Then to Leader, where I bought a replacement disk for the one that started dying a few days ago, and home to do Yet Another FreeBSD Install, which also didn't go as smoothly as it should. How can I describe this stuff in a book if I can't get it to work in practice?


Tuesday, 17 September 2002 Echunga
Top of page
previous day
next day
last day

More work on the upgrades today, and it became apparent how difficult it has become to upgrade anything on a FreeBSD system, let alone describe how to do it. At one point I got the machine in a state where it couldn't boot at all; I had managed to get a mismatch in /etc/fstab, and I couldn't remount the root file system, a typical chicken and egg problem. Found a way round that: mounted the compatibility slice on /mnt, so I had /dev/ad0s1a on the root file system, read-only, and /dev/ad0a on /mnt, read-write, and was able to modify /mnt/etc/fstab and reboot.

Spent the time after that with a very unsuccessful attempt to reinstall the ports I had. Ended up sending in a PR before I had to go into town and pick up the Magna. After that to the LinuxSA meeting, where Dan Shearer was talking about Hercules, an IBM System/390 emulator. I've always been puzzled about the attraction this has for a lot of people, Dan in particular, but it became apparent that it's at least partially due to lack of knowledge of the real system. Didn't participate in the discussion.


Wednesday, 18 September 2002 Echunga
Top of page
previous day
next day
last day

A frustrating day. Spent most of the morning trying to install various ports, and all of them seemed not to want to install. The last thing I wanted to do then. Ended up getting them working, including submitting a total of five PRs.

Didn't get much else done. My NetBSD build surprised me by still running after nearly two days.


Thursday, 19 September 2002 Echunga
Top of page
previous day
next day
last day

Finally my NetBSD build has finished, after 48 hours. Considering that this is a Pentium with about 133 or 166 MHz and 32 MB of memory, it doesn't compare very well to a build I did on diskless Sun 3/60s with 8 MB memory some time ago. That build took 100 hours, relatively fast in comparison. The problem was that I wasn't finished yet: I had to use a shell script to do the build, and it didn't install. Another thing to puzzle over.

More port installations, more bugs. People need to look at the end user perspective more. Installed StarOffice 5.2 because I couldn't find the OpenOffice CD I thought I had. It tried to download the binaries from the net. It seems that I had to explicitly specify -DWITH_CDROM. Where did it say that? In the Makefile, not as a comment but as the result of reading the rules. After installation, the build log finished with:

/src/FreeBSD/ports/editors/staroffice52/work/109939-03/share -> /usr/local/office52/share
/src/FreeBSD/ports/editors/staroffice52/work/109939-03/share/basic -> /usr/local/office52/share/basic
/src/FreeBSD/ports/editors/staroffice52/work/109939-03/share/basic/ssprint.sbl -> /usr/local/office52/share/basic/ssprint.sbl
===>   Generating temporary packing list
===>   Registering installation for staroffice-5.2_1
*** Error code 1 (ignored)

That's all. No indication of how to run the program, no obvious man page. With a bit of searching, I found I had to run /usr/local/office52/program/soffice. That's not easy to find: there are 118 executable files in that directory, including 102 libraries. This really needs to be done better.


Friday, 20 September 2002 Echunga
Top of page
previous day
next day
last day

More work on upgrading systems today. It's terrible work. I spent most of the day trying to find a clean way to upgrade /etc, and failed. mergemaster produced a lot of output, but didn't do very much else. Left me with about 80 files which had to be merged by hand, including such no-brainers as /etc/defaults/rc.conf. This system is just about unworkable.


Saturday, 21 September 2002 Echunga
Top of page
previous day
next day
last day

Came into the office this morning to find echunga had got itself into an NFS-related hang after the (el-cheapo Realtek) Ethernet card had wedged, requiring a reboot. As a result of that, I had missed my level 1 dump yesterday evening. Wrote a small script which would periodically take the interface down and then up again, and restarted the dump.

About two hours into the dump of my src file system, I noticed packets from echunga to wantadilla showing up on the tcpdump I keep running on sat-gw, the machine which runs my satellite Internet connection. After a bit of looking around, discovered that, although there was no connection problem between echunga and wantadilla (they're on the same switch), NFS traffic was going out of echunga's PPP interface, across the Pacific, up into space and back down via sat-gw. Took me a while to realise that this was due to the script I had written: part of the process of taking the interface down is to remove the route. It seems that if a request arrive while the interface was down, NFS took the logical choice and went via the default route. The only problem was that when the interface and route came back, it carried on using the old route. A little script rewriting (only cycle the interface if I can't reach either wantadilla or battunga) seems to have fixed that.

More work on updating configurations. This is still a royal pain.


Sunday, 22 September 2002 Echunga Images for 22 September 2002
Top of page
previous day
next day
last day

Still more work on configurations today, and finally managed to get some kind of method together. The trouble now is, should I document it in the book? It's becoming abundantly clear that the whole system is broken, and that we need to find a replacement. Still, it's given me the ability to move on, and it's high time for that.

Also finally got onto my two articles for Dæmon News, not too early. I wish I could find more inspiration.

Installed more ports on current, the new replacement for wantadilla. Documentation of some of these ports is terrible.


Monday, 23 September 2002 Echunga
Top of page
previous day
next day
last day

Up early this morning to pick my father up from the railway station from the Overlander from Melbourne. My aunt Freda had recommended checking with the railway first, so I called them up (note: “railway” in the phone book draws a blank, you need “Great Southern Railway”). The announcement was from the previous day, and the new one wasn't due until 7 am for a train which was scheduled to arrive at 7:25 am 40 minutes drive away. Finally got the announcement at 7:05 am: the train had been delayed by an hour, so it was worth calling anyway.

Got back home and found that I couldn't get much work done, all the more a concern because of the impending deadlines for my articles for Dæmon News. Managed to install mplayer, an MPEG player, and wondered idly how to get MPEGs off the TiVo. Fortunately, Tridge came online on IRC and pointed me to some software he had written for that. Spent a remarkably short period of time porting it to FreeBSD, and a remarkably long period of time getting it to work. It wasn't until I got further feedback from Tridge in the evening that I was able to pull the data off the machine and on to wantadilla, but then I had family obligations, so didn't get much else done.


Tuesday, 24 September 2002 Echunga
Top of page
previous day
next day
last day

More work on the TiVo copy stuff today, which kept me going all day. Nothing very complicated, just a lot of unclear relationships, like whether I was in fact getting the right formats. Spent a lot of time checking this, including building vplay under Linux (where it didn't compile cleanly at first) and trying to understand why it split video and audio, and why it invoked the video display program (mpeg2dec) with parameters which it didn't understand. Ended up being able to copy and display data, but not to display it directly. A lot more work is needed here.


Wednesday, 25 September 2002 Echunga
Top of page
previous day
next day
last day

Another frustrating day trying to get video stuff working. I sometimes wonder why I don't give up, but this is exactly the stuff I should be documenting both for my book and my porting articles. Somebody on IRC suggested that I should be running portupgrade, so spent some time investigating that, another can of worms. Then tried to rebuild mplayer, which had previously built. This time it didn't, but I didn't have the energy to investigate why. sigh.


Thursday, 26 September 2002 Echunga
Top of page
previous day
next day
last day

Still more work on the video today. Found that the lame port was what had died yesterday. The new C compiler found filth in there that had been ignored by previous compilers, including redefining array sizes in some source files only. Yet Another PR.

After that, to my surprise, it worked. There's still a lot of work to be done, possibly also on the TiVo side, but I was able to display and hear TV programs on the computer. A lot of jaggies on the display, unless I went to full screen, and the combination used up all the CPU on echunga (750 MHz Athlon). Couldn't run it on wantadilla (1.5 GHz Athlon) because I didn't have any sound hardware working. I suppose I'll have to work on that, but it'll have to wait. I've decided that there should be multiple authors for the Trawling The Ports Collection series, preferably written by the port maintainers, who should know what they're talking about.


Friday, 27 September 2002 Echunga
Top of page
previous day
next day
last day

Didn't get much work done today. Hugh Blemings and his wife Lucy came out to visit; they've been in Adelaide all week. Spent a lot of time looking at old junk hardware, but also went to the trouble to confirm whether Hugh could interoperate with my wireless network: last time anybody came here with a wireless Linux laptop was at the Hacker's barbecue last Christmas, when we weren't able to interoperate with Linux. It wasn't until the linux.conf.au in Brisbane in February 2002 that David Gibson and I worked out what went wrong, so this was the first test.

It didn't work. Well, not immediately, but that was a configuration problem: Hugh had the incorrect encryption key. After sorting that out, it worked perfectly.


Saturday, 28 September 2002 Echunga
Top of page
previous day
next day
last day

Quiet day. Finally decided on a (different) topic for my Trawling The Ports Collection article. Hopefully it will be the last one. I have spent a whole week on this, and I can't afford to do that all the time.


Sunday, 29 September 2002 Echunga
Top of page
previous day
next day
last day

More work on my article today, which is going to be about mutt. I've basically ripped it out of my book, which has the interesting effect that I have found all sorts of additions which will also go back into the book.

Spent some time investigating my bounced mail file, which I hadn't looked at for nearly three months, in which time I had accumulated only 30,000 bounces or so, a mere 350 per day. Some came from people who should have known better, so I decided to copy them back to the appropriate people. That worked out quite nicely: the bounce messages look like:

To: postmaster@lemis.com (Postmaster)
Subject: Postfix SMTP server: errors from unknown[164.109.58.147]

Transcript of session follows.

 Out: 220 wantadilla.lemis.com ESMTP Postfix
 In:  EHLO oracledb1.economist.com
 Out: 250-wantadilla.lemis.com
 Out: 250-PIPELINING
 Out: 250-SIZE 10240000
 Out: 250-ETRN
 Out: 250 8BITMIME
 In:  MAIL From:<subscriptionmaster@economist.com> SIZE=973
 Out: 250 Ok
 In:  RCPT To:<gregecon@lemis.com>
 Out: 450 Client host rejected: cannot find your hostname, [164.109.58.147]
 In:  RSET
 Out: 250 Ok
 In:  QUIT
 Out: 221 Bye

In other words, no reverse lookup. I added this little snippet to my .procmailrc:

:0
* ^Subject: Postfix SMTP server: errors.*164.109.58.147
| B u -

B is a script which dissects the message and sends a reply back telling people the errors of their ways. After all, if I get this many messages, why shouldn't they? Also wrote a web page explaining what it means; it's surprising how many people don't understand the message when it's sent: “The mail didn't go through, I got an error message”. “What did it say?”. “I don't know, I didn't read it”.


Monday, 30 September 2002 Echunga
Top of page
previous day
next day
last day

Spent some time looking at the results of yesterday's mail modifications and noted with some satisfaction that one of the culprits had accessed the web site and also fixed his reverse DNS, allowing the queued message to arrive:

202.58.62.105 - - [29/Sep/2002:20:03:13 -0400] "GET /reverse-lookup.html HTTP/1.1" 200 7910
202.58.62.36 - - [29/Sep/2002:20:15:36 -0400] "GET /reverse-lookup.html HTTP/1.1" 200 7910

Also discovered that mail to The Economist was not getting delivered. Further investigation showed that I had a problem with my local firewall: DNS replies to battunga were being dropped, but it didn't show because replies to echunga were being passed through. It might explain a lot of my modem traffic, though.

The Economist had a different problem: they have three, maybe four name servers, none of which show in their whois information. whois tells me:

   Domain servers in listed order:

   NS.ECONOMIST.CO.UK           194.205.68.226
   NS0.INSNET.NET               194.177.160.34
   DRDNS.DR.ECONOMIST.COM       213.212.70.100

nslookup tells me:

economist.com   nameserver = ns1.insnet.net
economist.com   nameserver = ns1.economist.com
economist.com   nameserver = ns2.economist.com
economist.com   nameserver = ns0.insnet.net

Authoritative answers can be found from:
ns1.insnet.net  internet address = 194.177.170.34
ns1.economist.com       internet address = 194.205.68.228
ns2.economist.com       internet address = 194.205.68.229

In other words, no overlap. It seems that one of the designated name servers (ns0.insnet.net) doesn't have an IP address, which is probably why things are so flaky. What a mess.

While looking around, I also discovered spam coming from a US .mil site:

To: postmaster@lemis.com (Postmaster)
Subject: Postfix SMTP server: errors from 130engbdeexch1.130thengineers.army.mil[140.156.104.11]

Transcript of session follows.

 Out: 220 wantadilla.lemis.com ESMTP Postfix
 In:  EHLO 130engbdeexch1.130thengineers.army.mil
 Out: 250-wantadilla.lemis.com
 Out: 250-PIPELINING
 Out: 250-SIZE 10240000
 Out: 250-ETRN
 Out: 250 8BITMIME
 In:  MAIL FROM:<bielerjeff@akrama.net> SIZE=2537
 Out: 250 Ok
 In:  RCPT TO:<rog@lemis.com>
 Out: 450 <bielerjeff@akrama.net>: Sender address rejected: Domain not found
 In:  QUIT
 Out: 221 Bye

I bounced the messages to abuse@army.mil, but that address doesn't exist either. I hope they know how to fight terrorism better than they know how to protect their computer systems.

Finally finished the article on mutt, but was too tired to do the Daemon's advocate. That will have to come in the very last minute.


Top of page Previous month Greg's home page Today's diary entry Next month Greg's photos Copyright information

Valid XHTML 1.0!

$Id: diary-sep2002.php,v 1.54 2015/12/30 01:20:41 grog Exp $