How to set up WEP encryption:

• SMC Barricade Basestation
• Airport Extreme Wireless Card
• Mac OS X 10.2.6

Note that if things don’t go 100% right, you may lock yourself out of your wireless network. Be sure you know how to get back in before you follow any of these instructions.

1. Set up non-encrypted transmission first. Make sure you can connect to your basestation and surf on the internet.
2. Visit your basestation’s URL - default is http://198.168.123.254
3. Log in (You have changed the password so it’s not ‘admin’, right?)
4. Click on Wireless.
5. Create a Network ID (SSID). I’ve read that leaving it at ‘default’ sometimes causes problems, so change it.
6. Check the ‘Enable IEEE 128 bit Shared Key security’ radio button
7. Now we need to enter a key. This should be a 26 digit Hex code. Simply go to the magic WEP Strong Key Generator. Click on ‘generate 128 bit key’ and scroll down to see your new key. Pick the one labeled HEX. Enter this value as ‘WEP Key 1’. Keep this key visible somewhere, since you’ll need to feed it to your Mac so you can login.
8. Click ‘Save’ and then ‘Reboot’ your wireless router.
9. OK, now you’re locked out of your basestation. Let’s get back in.
10. Click on the Airport Icon in the menubar. Choose ‘Turn Airport Off’. Wait a few seconds and then click ‘Turn Airport On’.
11. Wait a few seconds for your Mac to recognize your network. Then, click on the Airport icon again and choose your Network-ID. A password box should pop up.
12. Click the drop-down menu and choose ‘128 bit Key’ and then re-enter your 26-digit HEX key. And no, copy and paste won’t work - you need to enter it digit by digit. Check the ‘Keychain’ box so that you don’t have to re-enter that hideous key each time.

Enjoy your new encrypted network!

Ugh… I woke up this morning with a nagging pain in my right knee. It kinda improved through the day, so I wasn’t too worried about it, but when I tried to run on it this evening, it really hurt. So I’m gonna keep to the bike for awhile. I remember having this problem (which I think is patellar tendonitis) during the last marathon and I remember that it got better pretty quick, so I’m praying it’s the same this time.

Ha! I found another reason not to get married. Apparently, marriage quelches genius. On the other hand, if I got married, I could blame marriage for the fact that I’m not a genius…

I spent most of this afternoon struggling with a mail issue and I didn’t find it documented well anywhere else, so I’m going to write about it here in the hopes that it might help someone else.

What I was trying to do: Setup qmail + vpopmail + courier imap for imap and pop3d

Problem I was having: For some users, checking POP3 mail would work for a while (about 5 minutes), but then would fail:

      Jul  5 00:30:34 vkurup pop3d: LOGIN FAILED, ip=[::ffff:68.173.27.35]
      Jul  5 00:30:34 vkurup pop3d: Disconnected, ip=[::ffff:68.173.27.35]

The failure would only happen for users who had vpopmail accounts and shell accounts. Users who have only vpopmail accounts were fine.

Eventually, after much searching, I found this message which advised to configure courier-imap with --without-authdaemon. This solved my problem.

According to the author of courier-imap, this problem is solved in newer versions of vpopmail, but he mentions version 5.3.38 and I could only find version 5.3.20, so I didn’t try that fix.

Not only has MarkD produced a book, but he’s made SlashDot as well. Guess it’s about time to retire now…

I’m going to get it eventually, but I’m still working my way through Cocoa Programming for Mac OS X

I’m going to sign up for the National Do-Not-Call Registry, but I’m not sure that it’s going to reduce the number of calls I get. Look at the exemption list:

Exempt businesses include:

• long-distance phone companies
• airlines
• banks and credit unions; and
• the business of insurance, to the extent that it is regulated by state law.

However, many telemarketing calls are placed by professional telemarketing companies, and even if the company whose goods or services are being sold is exempt, the telemarketing company may be covered.

You may still receive calls from political organizations, charities, telephone surveyors or companies with which you have an existing business relationship.

The list of telemarketers who don’t fit into the exemption list seems kinda small to me…

I have a love/hate relationship with daemontools. It’s powerful way to manage services (like web servers and mail servers), but it sometimes doesn’t act the way I expect it to.

Today, I tried to set up qmail-pop3d. I set up a directory called /var/qmail/supervise/qmail-pop3d and put a file named ‘run’ inside it. This contained:

#!/bin/sh
PATH=/var/qmail/bin:/usr/local/sbin:/usr/local/bin:/usr/bin:/bin
export PATH

tcpserver -H -R 0 pop-3 
/var/qmail/bin/qmail-popup vkurup.acornhosting.net 
/home/vpopmail/bin/vchkpw /var/qmail/bin/qmail-pop3d Maildir &

Then I connected the run script to my /service directory

# ln -s /var/qmail/supervise/qmail-pop3d /service

Then I checked to see if it was running:

# svstat /service/qmail-pop3d
/service/qmail-pop3d: up (pid 13524) 0 seconds

Hmmm… up 0 seconds. That’s not right - it should be at least 5 or 6 seconds (I can’t type that fast). That usually means that the service is repeatedly failing and restarting itself.

ps shows that qmail-pop3d is running, but trying to connect to port 110 doesn’t work:

$ telnet localhost 110
Trying 127.0.0.1...
telnet: Unable to connect to remote host: Connection timed out

And now I’m stuck. Nothing is getting logged to the qmail-pop3d logs. After some headbanging, I do another ps -ax and this time notice a process called readproctitle:

11919 ?        S      0:05 readproctitle service errors: ...lready used?tcpserve
r: fatal: unable to bind: address already used?tcpserver: fatal: unable to bind:
 address already used?tcpserver: fatal: unable to bind: address already used?tcp
server: fatal: unable to bind: address already used?tcpserver: fatal: unable to 
bind: address already used?tcpserver: fatal: unable to bind: address already use
d?tcpserver: fatal: unable to bind: address already used?

What’s that? I look it up and find that readproctitle is a kind of scrolling-log for daemontools which shows up when you run ps. So, it looks like qmail-pop3d failing because it’s trying to bind port 110 even though it’s already been bound. So this makes me thing that qmail-pop3d isn’t failing, but that it’s running over and over again. And then I look back at my run script and look at that pesky little ‘&’ at the end. Doh! That tells the process to detach once it starts. It starts normally, then detaches. Once it detaches, daemontools thinks it’s down, so it tries to start it again. (At least that’s the way I understand it). Getting rid of the ‘&’ and adding ‘exec’ to the beginning fixes it and everything works now.

Me: 1 Daemontools: 2123

I’m setting up vpopmail 5.2.1 - a POP3 email manager for qmail, but it won’t compile out of the box.

gcc -I. -Icdb -g -O2 -Wall -c vconvert.c 
In file included from vconvert.c:35: vmysql.h:53:22: missing terminating " character 
vmysql.h:60:35: missing terminating " character 
make[2]: *** [vconvert.o] Error 1 
make[2]: Leaving directory `/usr/local/src/vpopmail-5.2.1' 
make[1]: *** [all-recursive] Error 1 
make[1]: Leaving directory `/usr/local/src/vpopmail-5.2.1' 
make: *** [all-recursive-am] Error 2 

Whenever make gives you an error, look for the first error you can find. Any errors after that often fix themselves once you fix the first error. The relevant code is here, (vmysql.h, starting at line 42):

42: #ifdef CLEAR_PASS 
43: #define TABLE_LAYOUT "pw_name char(32) not null, \ 
44: pw_domain char(64) not NULL, \ 
45: pw_passwd char(40), \ 
46: pw_uid int, pw_gid int, \ 
47: pw_gecos char(48), \ 
48: pw_dir char(160), \ 
49: pw_shell char(20), \ 
50: pw_clear_passwd char(16), \ 
51: primary key (pw_name, pw_domain ) " 
52: #else 
53: #define TABLE_LAYOUT "pw_name char(32) not null, \ 
54: pw_domain char(64) not null, 
55: pw_passwd char(40), \ 
56: pw_uid int, pw_gid int, \ 
57: pw_gecos char(48), \ 
58: pw_dir char(160), 
59: pw_shell char(20), \ 
60: primary key (pw_name, pw_domain ) " 
61: #endif

Looking around line 53, I see that line 54 and line 58 are missing backslashes - adding ‘em back in fixes the error. The interesting thing to me is that version 5.2.1 has been out for over a year and, according to an email on the vpopmail list, this error still persists in more recent versions. How come this bug hasn’t bitten anyone else? I suppose it’s because the install docs recommend setting CLEAR_PASS (see line 42), which would cause the compiler to miss the bug. Or maybe I’m just lucky

Not all of OpenACS is Postgres 7.3 compliant yet. OpenACS 5 is compliant and the core packages in OpenACS 4.6 are compliant, but some non-core packages haven’t been fixed yet. Noncompliant packages will complain:

[24/Jun/2003:15:21:36][26623.196621][-conn:kurup::4] Error:  Ns_PgExec: result status: 7 message: ERROR:  Function  pa_collection__new("unknown", "unknown", "unknown", timestamp with time zone,  "unknown", "unknown", "unknown") does not exist
Unable to identify a function that satisfies the given argument types
You may need to add explicit typecasts

It is looking for a function with timestamp with time zone as one of its parameters, but most OpenACS pl/pgsql functions were defined as accepting timestamp. In PG 7.3, timestamp means ‘timestamp without timezone’. timestamptz means ‘timestamp with timezone’. So, the quick fix, is to change timestamp to timestamptz.

Read more at Bart’s posts in this thread.

psql doesn’t deal well with tabs. If you feed it a function that has tabs in it, and then call that function, you’ll often get this: ERROR: parser: parse error at or near "for" at character 15. Thus, if you ever see this in your error log, it’s time to M-x untabify.