Thu Mar 21, 2013, 3:15 AM  --  What does it mean if you don't post any updates to a page in four and a half years? In my case, it means that I have better things to do. Does anyone still maintain vanity domains anymore? Once upon a time every self-respecting nerd had a vanity domain, and if you didn't it meant that something was wrong with you. Now everything is all GMail and Facebook, and nobody has a domain. Well, nobody except me anyway. I've had this darn thing since about 1998, and I'm keeping it. So there.

What was my point? Oh yeah, I migrated the content into an Amazon S3 cloud site. Haven't you heard? Everything's all cloud now. This one is cumulonimbus, or cirrostratus, or nimbulo-strata-fluffy or something. Whatever that means. Who made up the term Cloud Computing anyway? It had to be someone in marketing. If the engineers named it it would be calleed Dynamically Allocated Virtual Datacenter Emulation or something like that, and nobody would buy it. It's probably a good thing that they had some people in marketing.

Anyway welcome to the cloud.

Fri Aug 8, 2008, 11:15 AM  --  The server was down again last night for about an hour and a half. The UPS batteries failed.

Wed Jan 9, 2008, 12:45 AM  --  The server was down for about eight hours last Friday when a storm knocked out the power to my whole block. I thought about hooking the server up to a ten-speed bicycle and a generator and peddling like mad for hours on end so that y'all would be able to read your important emails, but I decided to go out for beer and pizza instead.

Since you're reading this, the power is obviously back on.

Mon Oct 8, 2007, 1:45 AM  -- is in San Jose! It will take a while for DNS changes to propagate, but I set up a fake hostname ( as an alternate MX host. Hopefully a few servers have cached that information and will successfully deliver mail until Network Solutions updates their info. The mail log seems to be filling up normally.

Sun Mar 25, 2007, 2:20 PM  --  I noticed today that the mail traffic was looking awfully low. A bit of investigation revealed that the sendmail daemon, which hadn't been restarted since I fixed the time, was still running on PST. Mail was being sent and received just fine, but the log files all had time stamps an hour in the past. The scripts that collect mail statistics run in hourly blocks, so they thought that there was no mail traffic recorded for the current hour.

The sendmail daemon has been restarted and things are back to normal.

Tue Mar 20, 2006, 11:30 AM  --  Yesterday I fixed the server time zone. Now it's on the screwey new Bush Administration time. Is there any real documentation that proves the argument that Daylight Savings Time really saves any energy in a modern-day society? I don't know about you, but I work in an office building where the lights are on all day no matter what the clock says. They only turn off late at night.

What does it say about our society that people have to have their clocks changed by the legislature because they can't just look out the window at the sunshine?

I propose a new time zone, which I call Sunrise Normalized Time or SNT. If we are going to mess around with the clocks at all, why not do it continuously instead of all at once? A clock on SNT would always start out the day at sunrise. This could be called zero or twelve or whatever; it doesn't matter what we call it. Let's make it six so that people won't have to get used to getting out of bed at one in the morning (eveen though it would be more like what we call seven in the morning today). By the same token, we may as well keep the same 12-hour system so that people don't have to adjust. The point is that sunrise would always be at the same time every day. The end of the day on the other hand would move around quite a bit as the year passed. On the very shortest day of the year, the sunset would be at 3:35 PM. On the very longest, it would be at 8:44 PM. This is what Daylight Savings Time tries to do in a primitive way; I am only proposing that we make it continual.

The first obvious argument against this plan is that it would be complicated. My response is that any clock or watch which has a battery is computer-based, and that it wouldn't be hard to have them calculate time relative to the sunrise. Certain time-critical computer servers and space sattelites would want to keep a different standard, but most of them already do. To prevent trouble with time zones, most real computers are really on Greenwich time. The time zone you set only changes the displayed time, not the computer's clock. As for the people, millions of years of evolution (or seven days and nights if you prefer that worldview) have adapted people for a variable-length day. I believe that humans would actually do better on SNT than under the current antiquated system which hasn't changed since the most advanced timekeping technology amounted to keeping lint out of the gears.

The astute observer will note that sunrise is of course relative to your position on the planet. SNT would still have hour-wide time zones to allow people to agree on what time they get to work or what time the train leaves. In America for example, we would replace PST, MST, CST and EST with PNT, MNT, CNT and ENT. We would replace PDT, MDT, CDT and EDT with nothing.

I'll put together some sort of Java applet for the home page to display the current time in PNT.

Mon Feb 5, 2007, 11:00 AM  --  There were some problems this weekend with spam handling and email. I upgraded the system Perl version, and it broke Spamassassin and Openwebmail. This is just another reaason to dislike interpreted languages; If I break my C compiler then I can't compile anything, but all of the programs I already have will keep working. If I break the Perl interpreter, it's curtains for anything written in Perl.
At any rate, it's fixed now.

Sun Dec 10, 2006, 1:55 AM  --  The system software has been updated again, and the system rebooted. Yay!

Wed Oct 11, 2006, 2:15 PM  --  The system software has been updated with some security enhancements and bug fixes. The system has been rebooted.

Wed Sep 20, 2006, 11:15 PM  --  The power supply is fixed. I had the system down for about an hour to complete this ten minute project. Why is it that it always seems to work out that way? It would have been smooth sailing if the fan connector hadn't been glued on to the power supply board so that I sort of had to shave the goo off of it to install the new one. Still, PC Power and Cooling gets props for building a power supply that can run continuously for a decade, and then when the fan finally quits it just keeps going anyway.

For the record, is a lot more like Serenety or the Millennium Falcon than it is like the starship Enterprise; It's dusty and broken, but I keep it running anyway.

Wed Sep 6, 2006, 11:00 PM  --  The server crashed this afternoon. The power supply fan failed, and it got really hot. I haven't fixed the fan yet, but the machine seems to be running okay with the side panel off for ventilation. I'll fix the fan as soon as I can. In case I forget, it is 3 1/8 inches between screws, and 4 1/2 on the diagonal.

Wed Jun 28, 2006, 3:14 AM  --  One of the hard drives on the server died. It has been replaced. The power supply fan is also sounding like it's not long for this world, but I don't have an extra one in my pocket.

Sleeping now. Bye.

Sat Dec 10, 2005, 3:45 PM  --  Last night I replaced the UPS with a power strip. The batteries are completely dead, and every time the UPS does a self test the system goes down. I am not sure why exactly the batteries have reached this sorry state seeing as they were fine about two months ago, but I will look into it.

Wed Dec 7, 2005, 7:00 PM  --  The system went down at 5:41 this morning because the UPS was freaking out. The system log shows 52 power failures of two or three seconds duration over a twelve minute period. It also reports that the UPS batteries need to be replaced. It seems unlikely that the power actually went out fifty-two times, but I suppose stranger things have happened. I suspect that it may be due to weirdness with the UPS. We of course have our best team of network engineers working on the problem, but until they solve it everybody can feel free to give me a call if the system seems to have gone down again.

Sun Aug 28, 2005, 2:45 AM  --  For those of you who may be looking, the gap in the email traffic graphs for most of Friday is due to a problem with the logfile analyzer, and not the email delivery system. Email was working just fine, but it wasn't getting counted.

Fri Jul 22, 2005, 12:05 AM  --  The server has been up and down for about the last hour while I worked on the DSL 'modem' problem. Some of you (does anybody actually read this stuff?) may recall that back when the server was in Palo Alto and had an SBC dsl line, the 'modem' kept freakin' out and needed to be regularly power cycled. You may also recall that I built a little black box that allowed the server to do just that as needed. Well, I reinstalled that very same box so that the server may again power cycle the 'modem' all by itself. I had eliminated the box to free up the serial port for an actual modem, allowing for remote administration if the network went all haywire. That worked great until the new DSL 'modem' began having similar problems last week. Anyway, after some IRQ shuffling and playing with undocumented DIP switches I have managed to get the modem, the UPS and the LBB (little black box) all up and running on their very own serial ports. Yay!

Tue Jul 13, 2005, 9:45 AM  --  The network has been down since about 2:30 AM last night. The DSL 'modem' kind of went to lunch, and had to be manually power cycled. Hopefully it won't be happening again any time soom.

Sat Jul 9, 2005, 10:15 PM  --  SpamAssassin has been upgraded to the latest version. Yay.

Tue Jun 14, 2005, 10:45 PM  --  The server is now doing antipipelining for incoming mail messages. If this doesn't mean anything to you, you can just smile and nod.

The short version is that there is some polite handshaking which happens when a regular email message is received by the server whereby the sending system says hello, and the server says hello back, and so on and so forth. A lot of spam senders try to slam all of this stuff into the server at one time, without waiting for the polite responses. This is called pipelining. Antipipelining does just what it sounds like -- it rejects this sort of connection.

This is in addition to the private blacklist that has been in place for about a month and a half now.

Sun May 29, 2005, 1:30 AM  --  At about 2:30 AM yesterday morning, the average nearest-hop ping time for the server dropped from close to 40 milliseconds to about 30. The 'floor' (see graph) dropped from around 33 to around 25 milliseconds. Speakeasy may have installed a new router or something, but they haven't said anything. The improvement is not spetacular, but I thought it was worth noting.

Mon May 2, 2005, 12:30 AM  --  There hasn't been a lot of news lately. As they say, no news is good news.

I set up an in-house proprietary spam solution on Wednesday, and so far it is working really well. I am not going to go into just how well it works, or how it works, because the spammers will only read this and figure out a workaround. I will however take this opportunity to point out to any unsolicited bulk email senders that there is a special place in Hell reserved just for you, where you will experience the eternal sensation of having your eyeballs gouged out with icepicks while being hung from the ceiling by your genitalia. I'm serious, you fuckers can Burn.

Tue Dec 28, 2004, 12:30 PM  --  Happy Birthday Heather!
The server software has been upgraded again.

Fri Dec 24, 2004, 9:30 PM  --  The server is now running on 196 megs of RAM, which is a little better than the 64 megs it had half an hour ago. It ought to see 320, but for some strange reason the 256 meg DIMM that I put in only shows up as 128 megs. It is an old motherboard, it might not be able to see any more than that.

Fri Nov 26, 2004, 2:30 PM  --  Since I am up here in Placerville with the server for Thanksgiving (doesn't everyone visit their server for Thanksgiving?), I rebuilt the system software last night with the latest updates and installed it today. The system is now...updated.

Tue Oct 5, 2004, 11:15 AM  --  Several services have been intermittent since about three this morning. The system logfile drive got filled up and confused the system to some degree. It should be better now.

Tue Sep 28, 2004, 3:10 AM  --  I got a message at about 3:30 yesterday afternoon that the server was experiencing a loss of utility power and shutting down. When the power was restored a couple of hours later, the server did not start up properly. There were two reasons for this: The first problem was that the new system software takes a little bit longer to shut down than the older software did, and the UPS was actually turning off before the system was done shutting down. This caused the backup disk not to be unmounted properly. The second problem was that the backup disk was not configured to be automatically checked at boot, and the system refused to continue the boot process without it.

Both problems have been fixed.

Fri Jul 23, 2004, 1:40 PM  --  The server ran out of logfile space today while I was working on some things, and I rebooted it. All is better. is moving towards supporting SPF email sender authentication. This is an anti-spam scheme endorsed by a number of large email providers. has been forged as the sender for a pretty good amount of spam and viruses lately (this is not unusual these days), so SPF may be able to help out with that a little bit.

If anyone is still using a third-party mailserver for outgoing mail, I would recommend changing to as soon as it is convenient. It is not yet required that be used for outgoing mail, but it may be in the future.

Fri Jun 4, 2004, 1:25 AM  --  After a rather long build, the system software has been upgraded to release 4.10.

The UPS is currently reporting a temperature of 129.2 degrees F. The last time I was there my highly calibrated hand did not seem to detect a temperature that high, so I am assuming the sensor is a little out of whack. It isn't exactly lab grade. Just in case you were wondering, the UPS data cable did come unplugged in March, but that would be unlikely to calse erroneous temperature readings. The UPS talks to the server over a serial connection, so it pretty much either works fine or it doesn't work at all.

Sun Mar 21, 2004, 1:10 AM  --  Serial communications with the UPS seems to be down. The cable has probably fallen off or something. The UPS will work fine without communicating to the server, so it isn't really a cause for concern. The only potential problem is that in the event of a power failure of greater than about an hour in duration, the system would not shut down properly. I don't anticipate this problem happening very soon.

Sun Mar 14, 2004, 3:30 PM  --  The long awaited software upgrade has been performed! The system is now running FreeBSD 4.9.

Sat Mar 13, 2004, 5:30 PM  -- is now happily located in Placerville. Our DSL is now being provided by Speakeasy.

Sat Mar 13, 2004, 11:15 AM  --  After 283 days, 22 hours and 36 minutes without a reboot in Jon and Rosie's laundry room, is saying goodbye to the Bay Area and venturing Eastward. Bye!

Wed Mar 10, 2004, 2:00 PM  --  Not much exciting stuff going on lately, so here is a little status update. The server will be moving to Placerville this weekend, either on Saturday or Sunday. Downtime is expected to be about six hours. Also, I have deleted the accounts 'bob' and 'twilight1138' because they have not been used in over a year and collectively were taking up 207 megs on the mail spool drive. Mike Palumbo and Bob Van Zant, if you want your accounts back you had better promise to delete your email at least twice a year.

Fri Feb 6, 2004, 11:20 AM  --  There hasn't been a lot of news lately, things have been running pretty smoothly. At least until now. Over the last week or so there have been a couple of developments worth noting. First, the MyDoom Microsoft email virus has been hitting the server with about 2,500 messages per hour. I thought it was interesting enough that I set up email traffic graphs on the system status page. The other interesting development is that the DSL at Jon's house seems to have been pretty spotty for the last couple of days. Not that there is anything I can really do about that, I just thought I should mention it in case anyone has noticed the lack of availability.

Now back to your regularly scheduled websurfing.

Sun Oct 26, 2003, 1:00 AM  --  The server went on battery power for nine seconds yesterday morning. That isn't really very exciting news, but the UPS doesn't get used very often so I thought it was kind of neat.

Let's do the numbers:
1:07AM up 144 days, 12:27, 1 user, load averages: 0.23, 0.38, 0.31

The astute reader might notice that it has not been 144 days since I replaced the CPU fan. I did not bother to take the system down just for that, it was easy enough to remove and replace it with the machine running. Some of the newer chips have such a small heatsink contact area that they might very well overheat in the half a minute it took to replace. I did not consider that to be a problem in this case. The K6-2 that runs on is underclocked and has a pretty large contact patch, so I figured it could dissipate an adequate amount of heat by itself. Frequent poking revealed that indeed the chip temperature did not exceed acceptable levels.

In other news, the system software uprade which was supposed to take place a couple of months ago has been (obviously) postponed. Hopefully I will find the time to do it as soon as I finish some of my other projects.

Mon Sep 29, 2003, 3:30 PM  -- went through another CPU fan today. I replaced it and added one of those boxes that plays an annoying little tune if it detects fan failure. This one happens to play Yankee Doodle Dandy. I offer my sincere appologies in advance to Jon and Rosie.

Tue Jun 10, 2003, 6:30 PM  --  The server has temporarily moved to Jon and Rosie's house in Mountain View. It will eventually be moved to Redwood City and hopefully stay there for a while, but it needeed a place to be while Brendan gets his act together about the whole DSL line thing.

Sat Apr 12, 2003, 1:45 AM  --  The new motherboard is installed and working! is now running on the absolute latest nineteen-nineties technology: A 500 MHz AMD K6-2! There is still only 64 megs of RAM on the board, but it should be quite a bit faster. Not that most users are likely to notice, but it should extend the lifespan of the current machine for a few years.

Mon Mar 24, 2003, 4:30 AM  --  The new root drive has been installed and seems to be working fine. It took me much longer than expected due to some problems getting the new drive to boot correctly. It is all properly configured now. The new drive is larger than the old one, so I have grown the user home directories and some system partitions.

While the system was down I took some time to play with the UPS configuration. I have set it up to shut the system down after 30 minutes running on batteries. My hope is that this will protect against the UPS overestimating the remaining battery run time. Not that this has ever been a problem, but better safe than sorry.

I also noticed that the CPU fan had siezed up. A drop of oil seems to have done the trick. I should try to look at it every six months or so.

Fri Mar 21, 2003, 2:45 AM  -- is backing up files again. For those of you who remember, this hasn't been done in about two years. The tape drive has not been replaced, but I have found somewhere to put the backup files. Lee contributed a 4 gig IDE drive to run the porno spooler back when we were living in Sunnyvale, and even though the spooler is no longer running the drive is still in the machine. So, I have re-tasked it to store backups. The bad news is that the bad spot on the drive which needs to be replaced is in the user home directory partition. There are some bad files in my personal web page stuff, and in Brendan's Trash mail folder.

Wed Mar 19, 2003, 4:30 AM  --  The network routing tables just got hosed. I couldn't get the kernel to do anything to them, everything just timed out. So I rebooted the system, and it seems to be working all right now. Too bad, the uptime was at 239 days.

Wed Mar 19, 2003, 3:00 AM  --  The named process died unexpectedly on Tuesday at about 1 PM. There was a problem with the script which is responsible for detecting and correcting this situation, and it is now fixed. What is more interesting is what caused the name server process to die. The root disk shows several read errors over the past couple of days, and these appear to be the cause. This is is not a good thing. It doesn't seem to be due to cabling problems, which means that the drive is likely going bad. I will try to locate a new one.

Fri Jan 31, 2003, 7:00 PM  --  The web server has broken the gigabyte mark this month for data served. Not that this has any particular relevance to anyone, it is just an interesting bit of information. It is largely due to the nearly 400 megs that the server spit out on the 23rd. There doesn't seem to be any one site responsible for all of the hits that day, they seem to have come from all over the place. The top two URLS this month have been the Terrorist's Handbook and the Lockpicking Guide, which account for about 2,000 and 3,000 visits respectively. The next biggest URL is the web page itself, with a whopping 300 visits.

In other news, the DSL connection was down for an hour on the 28th along with SBC's entire Bay Area DSL network. The box worked perfectly.

Tue Dec 17, 2002, 9:20 PM  --  The script I wrote last night to reset the DSL 'modem' had a bug in it, and it wasn't actually getting reset. And of course, murphy's law being what it is, it went offline about an hour after I left. It is now fixed and working properly.

Tue Dec 17, 2002, 4:30 AM  --  I have built and installed the 'magic box' that I mentioned below. The server is now capable of healing itself when the DSL 'modem' goes offline.

For the really curious, here is how it works: I have a solid state relay wired up to a serial port on the server. The relay control pins are connected to the data terminal ready and signal ground lines of the port (pins 4 and 5 respectively on a DB9) through a current limiting resistor. The load pins of the relay are wired into the hot line of an extension cord, and the ground and neutral pass straight through. The whole thing is enclosed in a little black box for safety and appearence.

This means, among other things, that the "Last 10 network resets" section of the system status page actually reflects the network being reset.

Sun Dec 8, 2002, 3:15 AM  --  The system was unavailable on Tuesday the third from about 7 AM to just after 11 PM. The DSL 'modem' went offline and required a reset. The last time that happened was in June, so it seems not to be too frequent of an occurrence. If I feel especially productive some time soon I may build a box and write some scripts to allow the server to power-cycle the 'modem' all by itself. Then the 'modem' can be automatically reset whenever the network goes down and the server starts feeling lonely.

Thu Oct 3, 2002, 3:40 AM  --  Reverse DNS works properly for the first time since the move back to Palo Alto! I have finally gotten around to having PacBell Internet point reverse lookups for to It is a really complicated configuration with a sub-class-C address space allocation (I have three bits), which can safely be ignored by non-system-administrator types. I am excited though, because it means that the SpamCop engine works properly again. It now sees reverse IP lookups correctly and traces spam back to the source. Previously it was picking as the origin because the reverse DNS lookup didn't match the hostname given by sendmail.

Mon Jul 22, 2002, 1:30 AM  --  I have finally replaced the UPS batteries, and the UPS is up and running. is now thoroughly protected from the almost unheard-of event of a power failure in Palo Alto.

Sat Jul 20, 2002, 3:30 AM  --  Well, the bad news is that the new motherboard I got for the server appears for all intents and purposes to be DOA. I will work on it some more to see if I can get it going in my workstation, but I think it is no good. The good news is that it doesn't really matter all that much because the server doesn't do anything which takes a great deal of CPU power. I just want to get an upgrade in while I can still find an AT form factor board. For the overly curious, my upgrade configuration of choice is an FIC VA-503+ motherboard with an AMD K6-2 350. I already have the chip, I just need a board that works.

The replacement batteries for the UPS have been ordered and should be arriving soon.

Thu Jun 6, 2002, 3:15 PM  --  The system has been unavailable since some time yesterday afternoon due to a problem with the DSL 'modem'. The 'modem' has now been reset, and everything should be fine.

UPDATE 11:30 PM  --  It happened again. I reset it. Again. This is clearly a recurring problem. Either PacBell is doing something to their network that requires a reset, or the unit is going bad. We'll see if it happens a third time.

Tue May 21, 2002, 2:35 PM  -- extends a hearty welcome to, who tried to come by for a visit last night at about a quarter to one in the morning with some kind of portscanner. The connections from (, AKA were refused.

Fri May 18, 2002, 2:00 AM  --  The UPS batteries have failed. I took the system down and replaced the UPS with a power strip. I will take the unit back to my secret lair for repairs. Here are the numbers:

 1:51AM  up 88 days,  6:04, 1 user, load averages: 1.32, 1.27, 1.25

Thu Feb 14, 2002, 12:15 PM  --  I replaced the CPU fan on the system last night, and upped the clock speed to 180 MHz. I am contemplating a hardware upgrade to an FIC VA-503+ motherboard with an AMD K6-2 or K6-III chip on it. It looks like that is about the last AT form factor board that I will be able to buy, and to go to ATX I have to replace the case and power supply and yadda yadda yadda...

Sun Feb 3, 2002, 10:15 PM  --  After three days of downtime, some system setup, and a few nameserver gymnastics, has returned to Palo Alto. Connectivity is now being provided via DSL by SBC/PacBell. Hopefully this will result in better upstream data rates than the wireless Sprint Broadband contraption, and it has to be better than the crappy cable modem we were using the last time we were in this city.

Wed Nov 28, 2001, 2:25 PM  --  The system went down a little before 3:30 this morning due to a hardware problem. I found the server locked up and the SCSI bus hung. Power cycling fixed it. It is probably time for a major hardware upgrade, but I am a little strapped for cash at the moment.

Tue Oct 16, 2001, 12:10 AM  --  Bingo! The system upgrade required an update to the configuration file for the password authentication mechanism. Everything should now be back to good.

Mon Oct 15, 2001, 10:45 PM  --  I still have not fixed the login problem, but I am working on it. I am also working on a webmail gateway, which caused the web server to be down for several hours. The installer for the web mail software automatically upgraded the web server software, which was nice of it, but the shiny new web server didn't understand the old configuration files. I had to patch the new stuff into them by hand. At any rate, the web server is back but logins are still broken.

Just in case anyone is curious, the email system itself is working flawlessly. Now that I have said that it will probably break, but at least up until now it has performed perfectly.

Sun Oct 14, 2001, 11:00 PM  --  The security issues are now resolved, but I broke the login handler in the process. Until I get it fixed, users can read their mail using the POP or IMAP mail reader of their choice. That most likely means Netscape.

Sun Oct 14, 2001, 6:00 PM  --  Someone got in to the system last night or this morning through a telnetd exploit. They were running a port scanner and some sort of brute force remote login script. I believe that I have successfully removed the software that the cracker brought in, and I have temporarily disabled telnet logins until I get the daemon patched.

Sun Oct 7, 2001, 11:45 PM  --  The UPS seems to be going on battery an awful lot. Usually at exactly 8 AM for about 4 seconds. I am guessing that some company around here is turning on a large piece of equipment, and that it causes a sag in the power line. If it keeps happening I will reduce the sensitivity on the UPS.

In internet worm news, my bogus default.ida is being exported nicely. The transfers seem to be getting stopped after 16K. Either the worm is smart enough to stop them or it is running out of buffer space and crashing. I am fine with it either way. I have put the following additional bogus files in place to try and trip up the other worm that I am getting hit by.


Sun Sept 23, 2001, 3:15 PM  --  I have installed a UPS. Unfortunately, it turns out that I can't use a regular serial cable with APC's Smart-UPS brand equipment. I will have to build a custom one so that the system and the UPS can talk to each other. The cable looks like this.

In other news, the web server took a huge number of hits on the 18th and 19th of this month, and a little over twice the normal number the last three days as well. They look to be coming mostly from hosts on Sprint Broadband's network, but I haven't really looked into it yet. The number of bytes served hasn't gone up much, which leads me to believe that they are mostly requests for pages that don't exist. I will have to take a look at the log files and see if I can learn anything.

UPDATE: It looks like I am getting hit by some sort of distributed attack targeting NT systems. There are a bunch of GET requests for cmd.exe and root.exe with some funny stuff in the URL. It doesn't pose a threat to but I will explore it some more anyway. There are also some hits from the Code Red worm, but those go back further. I have set up the web server to spit out the entire Windows 98 CDROM image in place of the default.ida file that Code Red looks for. That should give the worm something to chew on. Hopefully it will crash the machine that the requests are coming from.

Mon Sept 17, 2001, 8:00 PM  --  I have changed the time servers that synchs to. Nobody should notice any difference.

Old servers:

New servers:

Mon Sept 3, 2001, 7:50 PM  --  The system went down last night sometime after 2 AM. Something went awry with one of the swap partitions. I am not quite sure what happened, perhaps the tape drive hosed the SCSI bus or something. At any rate, I rebooted it and everything is fine now. Although there have been several previous instances when was unavailable, most notably the DSL fiasco, this is the first time that the system has gone down completely out of the blue. I suppose that there could be a legitimate software bug somewhere, since no software is completely bug-free, but I feel more inclined to blame the hardware. Code as fundamental as virtual memory drivers is usually very well debugged simply by virtue of the fact that it is in constant use.

Sat Jun 30, 2001, 8:15 PM  --  The system was up and down a few times over the past hour while I was installing a second parallel port and a third hard drive.  Everything is back up now.

Thu Jun 28, 2001, 10:30 AM  --  After fighting with failed kernel dependencies all day yesterday, I just deleted the kernel sources and had the system fetch them fresh last night.  I rebuilt the kernel this morning and installed it and rebooted and everything is running perfectly. is now running FreeBSD 4.3.  I still need to do some cleanup in the ports tree, apparently there are some layout changes since 4.0, but everything is back to running.

It is distinctly possible that some mail was rejected yesterday -- the name server was hosed for a while.

Tue Jun 26, 2001, 3:00 PM  --  I broke the system by running out of disk space in the middle of a "make installworld".  I am working on getting everything synched to the proper code tree.  Presently all important services are up and accessable, but that might change.

Fri May 18, 2001, 5:00 PM  -- is back up!  On April 1, 2001 (I'm not kidding) at 11:42 PM PDT, my DSL connection was severed by Telocity.  For no apparent reason.  I could go on for hours about how I called their tech support department every single day for three weeks, but let me summarize:  They did not know why my connection wasn't working, they were unable to connect me with someone who could tell me why it wasn't working, and they were unable to connect me to someone who could fix it.  On May 1, I moved.  No, I didn't move because of Telocity's ineptitude, although that is certainly a compelling reason.  :-)  At any rate, I am now in a new location and have a different provider.  I actually got the server back up yesterday, but the name server changes didn't get around until today.

For the insatiably curious, here are the httpd log entries surrounding the major block of downtime.  I have crossed out the source IPs to protect the innocent.

XXX.XXX.XXX.XXX - - [01/Apr/2001:23:30:20 -0700] "GET /~akira/home/lockpick/ HTTP/1.0" 200 4872 "" "Mozilla/4.76 [en]C-CCK-MCD {TLC;RETAIL}  (Win98; U)"
XXX.XXX.XXX.XXX - - [18/May/2001:02:48:36 -0700] "GET /~akira/home/lockpick/ HTTP/1.1" 200 4872 "" "Mozilla/4.0 (compatible; MSIE 5.0; Windows 98; DigExt)"

Fri Mar 9, 2001, 2:00 PM  --  There hasn't been much news for quite a few months now. That's okay, no news is good news. But there is some now. First off, the tape backup drive isn't working at all anymore. I am loking into possible replacements. I don't consider it a crisis situation, the chance of major data loss is very small, but is is at least a newsworthy event. The other interesting development is that's IP address will be changing some time in the next few weeks as a result of the buyout of PhoenixDSL by Megapath. Megapath is farming the residential service out to Telocity, so whenever I change over to Telocity I will get a new IP address. There will probably be a day or two of downtime when that happens.

Thu Nov 2, 2000, 7:00 PM  --  Okay, I figured out where the mystery web hits came from. They were legitimate. On Monday, I updated my secondary nameserver ( to get its' data from my new IP address, and I forced it to do a synch. So most of the universe was in fact able to get to by querying the secondary server. The difficulty was this: I was using my CSU Chico account for testing, and Chico for some reason wouldn't go to my secondary nameserver to do lookups, whereas most of the internet at large had no problem with that. On closer inspection, the email that all arrived in one pile was all backed up at Chico's mail server, being forwarded since I don't really use that account much any more. The two messages which weren't being routed through that mail server were in fact dated today. The upshot of all of this is that the chance of mail having been returned is a lot less than I thought it was this morning. Although there are probably other places like Chico which were late to get the update, most people will have been all right since Monday evening.

Network Solutions' whois database has still not been updated. They are on crack. I was instructed by their tech support dudes to call if it wasn't fixed by tomorrow, so that is what I will do.

Thu Nov 2, 2000, 11:00 AM  --  Finally, after four days, Network Solutions has processed my name server change for I got a small pile of backed up mail in my inbox this morning. There is a pretty good chance that some time in the past four days someone somewhere tried to send mail to a user and it was returned as undeliverable. I appologize for the inconvenience, and I blame it all on Network Solutions. The professional way to have done the move would have been to set up a proxy host on the old IP address which would have routed everything to the real host until DNS reflected the change. Since I don't happen to have an extra server just lying around, I went with the downtime option instead.

I have noticed an odd peculiarity about my web server stats log tool. It shows way less usage than usual on Monday, which is what I would have expected since nobody knew where was that day. However the log shows pretty typical usage on Tuesday and Wednesday. I know that the root name servers had not yet been updated, so I am a little puzzled as to where the hits came from. Well okay, I know exactly where they came from because it is in the log file. What I don't understand is how they got here. A few of them look like robots, and they could just be looking at my IP address, but most of them look like they are legitimate. Wierd...

Sun Oct 29, 2000, 10:00 PM  -- is now in Sunnyvale! Everything is working perfectly, except for DNS from the outside world. The root nameservers will be updated tomorrow at 5 in the morning, and then everything should be just peachy. I have ordered a new motherboard for the server, and it will probably arrive some time this week. Not that the 166 MHz Pentium overclocked to 200 isn't working, it is doing fine. However the case that the system is in, which I like a lot, is an AT box and if I don't grab a motherboard soon I might never get a Pentium II/III board in an AT form factor. So there will be another hour or two of downtime later this week, but I highly doubt that my huge user base will even notice.

Sat Oct 28, 2000, 8:30 PM  --   Four hours, two 50-foot lengths of Cat-5 cable, one extension cord, two boxes of cable staples, and one pepperoni pizza later, my closet is now ready for the server. Methinks I shall go collect it tomorrow. Is it just me, or is there something fundamentally right about networks and pizza together in the same place? Maybe I am just wierd that way, it wouldn't be the first time...

Sat Oct 28, 2000, 4:30 PM  --   My DSL line is up and running. I'm getting 79.9 K/sec downstream and 42.4 K/sec upstream. Not too shabby. I am going to Fry's to grab some ethernet cable to run into the closet, and tomorrow or maybe even this evening I will move the server. A few days of downtime will probably ensue while the name server changes propagate through the DNS system.

Wed Oct 25, 2000, 8:00 PM  --  The PhoenixDSL people called this morning, they are coming Saturday afternoon to install my DSL line. I'm not selling the beach house yet though, not until I can actually send packets out and have them come back.

I'm kidding, I don't have a beach house. I would rather have a mountain cabin anyway.

As a follow-up to the last entry, I never did locate any customer-service people from AT&T Broadband, but they seem to have resolved the problem by themselves. Soon I won't have to care about those loosers anymore.

Nothing to report about the backups, everything is going swimmingly. I will need to buy a UPS for the Sunnyvale location. The server is going into my closet, and the nearest source of power is on a wall switch. I am considering hard-wiring it to always be on, but a UPS would be a good idea anyway.

Wed Oct 11, 2000, 11:20 AM  --   I guess I spoke too soon, the ping times are back to crappy. It is definately AT&T's problem, though. I will attempt to locate some sort of customer service person.

Tue Oct 10, 2000, 3:30 PM  --   For the last several days, ping times have been absolutely terrable. I sent an email off to AT&T Broadband and proceeded to try messing around with the gateway address, which didn't help at all. A few hours later things magically got better, so I guess they fixed the problem.

Now you, my dear reader, are probably wondering "Hey! What about that crappy ISPChannel place you keep whining about? What happened to them?" Well, since you asked I'll tell you. Since the last time I had any interesting news to report, Cable Co-Op in Palo Alto was sold to AT&T, and the cable modem service went with it. I am still getting bills on Cable Co-Op stationary, but the remittance address is now AT&T Broadband. My opinion of ISPChannel is still the same, they are crappy. The jury is still out on whether or not AT&T's cable modem service is crappy, and I don't expect to be with them long enough to form an opinion because of...

The DSL Saga:

Early in August I moved to Sunnyvale from Palo Alto. On August 11th, I placed an order with Phoenix DSL for a 768/408 DSL line. That is somewhat faster downstream and quite a bit faster upstream than the cable modem, so all three users should notice a pretty good speed improvement whenever the DSL line actually comes in. I received a confirmation email and a welcome packet from Phoenix, in which I learned that installation takes an average of two months. Not great, but I have heard that installation takes forever with pretty much all DSL providers so I don't hold it against them. Well, a month later on September 11th I decided to call them up and see how things were going. After much difficulty, I manage to learn from their customer support team that they seem to have lost my order. Great. They refer me to sales, and it takes about a week and a half to get a new order placed. I managed to get my new order confirmed on October 5th, almost two months after I initially placed the original order.

So why am I still trying to get DSL from these total bozos? Because I want a static IP address. There are other DSL providers who are a better deal if you just want to surf the web really fast, but Phoenix is one of two providers who will give you good bandwidth and a static IP at a reasonable price. So I am not giving up on them just yet. is pretty happy where it is, so there is no desperate crunch to get the system moved just yet. I will post more of the DSL saga as it develops.

In other news, the tape drive seems to be pretty happy with an occasional cleaning. Backups are backing up just like they should.

Fri Jul 28, 2000, 1:30 PM  --  I rebuilt the system this morning in preparation for an upgrade to FreeBSD 4.1.  If anything is broken, it is probably my fault. In other news I cleaned the tape drive and it seems to be working so far, and I feel like crap today. I think I am getting a cold. That is all.

Mon Jul 24, 2000, 11:30 AM  --  The backup failed last night. The system reports 'device not configured' on attempts to use the tape drive. I am hoping that it just needs cleaning and that I don't need to replace it. The Friday backup did get run after I rebooted on Saturday, so the most recent backup isn't too old.

Sun Jul 23, 2000, 1:40 AM  --  My tape drive has gone off line, and it is blinking error codes. The Friday backup was only partially successful. I am going to reboot the machine and re-run it. Hopefully power cycling will clear the error.

Thu Jul 20, 2000, 9:30 PM  --  ISPChannel deleted my statc IP address again.  I called them up last night and spent 45 minutes on hold, then 30 minutes talking to a tech support guy whose final answer was "I don't know, call customer service."  That really inspired confidence in their tech support staff for me.  This morning I called customer service and told them I wanted my static IP back.  They said that I couln't have a static IP with the 'bronze' plan, and I would have to go to the 'silver' plan.  I explained to them in no uncertain terms that since the silver plan costs $87 a month, it would be a much better deal for me to go with DSL.  They said they would call me back, which they did at about 3:00 in the afternoon.  They said that I had my stat IP and everything should be working.  So far, all appears well and good.  I am a little upset that they didn't make any effort to warn me that this was happening, they just did it.  I am still thinking about DSL, but since things are working once again I will just go with inertia unless they get worse.

Sun Jul 9, 2000, 3:00 PM  --  I changed the system backup schedule to be more frequent. I am now doing backups on Sunday night and Thursday night. Well, technically I am actually doing them on Monday and Friday mornings at 4:10 AM but I think of it as Sunday and Thursday nights. Although the official purpouse of the system backups is to protect against hardware failures and the like, if anyone accidentally deletes some files or something that you need, I can restore them from the bacckup tapes if they were around for one of the two most recent backup runs. The directories that get backed up are /mnt/home (user home directories), /mnt/capweb ( web stuff), and /var (user mail inbxes and some misc. system stuff).

Thu Jul 6, 2000, 1:00 AM  --  I was in Chico this weekend for a festive fourth-of-july party, at which much alcohol was consumed and much fun was had by all. I left Chico this morning and drove straight to work, where I found out when I tried to check my email that my network connection was down again. After work I went over to a friend's place for dinner, and when I eventually returned home I tracked down the problem: ISPChannel had decided that I no longer needed a static IP address and therefore had relegated me to DHCP-land. I don't like DHCP-land, so using my DHCP-assigned IP address I hunted down their 24-hour tech support number and called them up. So that I don't forget it in case I need it again, the number is 888 450-3390. For the record the tech support guy was really nice and very helpful. He got my name and IP address, adjusted the routing tables, and then I was back in business, so to speak. ISPChannel has redeemed themselves somewhat.

Tue Jun 27, 2000, 11:20 AM  --  Yesterday at 2 PM, the server died. I still don't know quite why, we'll see if it happens today. On a better note, I got the kernel all compiled with the proper drivers in it and everything seems to be working great. As it turned out, I forgot to include the needed file in the config, so it was all my fault. That's okay, it usually is.   :-)   At any rate, is now officially running FreeBSD 4.0. The network seems to be a little slow as I type this, it may be a configuration problem or it may just be the fact that Cable Co-Op's cable modem service is a piece of $#!+ and I really ought to go DSL, which was my original plan but when I called Orconet up they told me that their DSLAM in my area was full and they couldn't hook me up. But I've been over this before. At any rate, everything is better now.

Mon Jun 26, 2000, 3:40 AM  --  After some ado and a few hours of down time, I have successfully upgraded the system to 4.0.  The upgrade seems to have left out a header file that I need to compile a kernel to work with my fast ethernet card, but everything else appears to be working. There was a little bit of wierdness with one of my hard drives, I may make the backup schedule more frequent. I don't know for a fact that all of the services which are supposed to be up really are, because I can't talk on my LAN. I can ping the outside world though, so I will assume that all is well until I venture forth into said world later today and can actually test it.

Thu Jun 15, 2000, 10:50 AM  --  The 108-degree heat yesterday seems to have caused my tape drive to go to lunch, I can't send it any commands and I can't eject the tape that is in it. I am going to reboot the system and see if it helps. Adios to 81 days of uptime. :-(

Note after reboot: Yep, that fixed it. I wonder if it will be okay this evening, it is supposed to be another hot one today.

Tue March 28, 2000, 4:30 PM  --  I just noticed that the system rebooted at around 8:40 PM on Saturday, after about 65 days of uptime. Okay, it was 62 the last time I bothered to check, which was some time last week, so it was probably 65 or so on Saturday. Anyway, I'll have to look into what happened. I suspect that there was a power failure.

I don't think that this was related to the maintenance I did last week, although that is a possibility. I was doing a lot of mucking about in /var.

Thu March 23, 2000, 6:30 PM  --  Last night I sent Chris an MP3, and it didn't go through because the mail spool partition was full. So today, I moved some stuff around and made the mail spool partition 450 megs instead of the pitifully small 20 megs that it used to be. There is a caveat, however: I did all of this with the system up and without rebooting. This is a little bit like replacing a radiator with the engine running. Consequently, there may be something broken. I restarted the three services which I had to stop in order to move files around and I have tested everything which comes to mind, but if something seems really out of whack then it probably is. Drop me a line if you notice anything unusual.

Sun Feb 20, 2000, 2:00 AM  --  My cable modem is on the fritz again. It is telling me that it can't establish a connection. Well, the RF indicator light is blinking amber, which according to the documentation means Initialization and Acquisition. Basically, ET can't find the mother ship. I have mixed feelings about this whole cable modem thing. Plusses: 30 percent faster downstream than the cheap DSL, no term committment, and in theory the coax infrastructure should be better than twisted pair copper. Minuses: 20 percent slower upstream than cheap DSL, can only get one static IP, and ISPChannel isn't doing a very good job in the reliability department. It is possible that my particular cable modem is just screwed up, or my particular cable or something. The thing has to reset several times a day even when it works, but that usually takes less than 30 seconds. That one time the gateway's MAC address changed though -- that was definately not me. If I haven't printed out their tech support phone number after the link comes back up and people can read this (not that they would want to or anything), someone shoot me. My only real reservation about switching to DSL is this: Then I am locked into a term committment. What if their service is crap as well, but I am stuck with it, and what if the rates go through the roof while I am locked into a contract?

For the record, if I go to DSL my service provider of choice is Orconet.

Tue Feb 15, 2000, 2:30 PM  --  A miracle has occurred -- I am actually backing up files! All three users can now sleep with the knowledge that their junk mail is safely sequestered away on magnetic tape in case one of my hard drives crashes or something. I expect to make backups about every weekend, so if someone accidentally deletes something important I can get it back for you, at least if it was around for the previous backup. If my house burns down then you are out of luck, the tapes go with it.

Mon Jan 24, 2000, 8:00 PM  --  Yesterday evening starting at about quarter to seven my cable modem was unable to establish a connection.  It stayed that way until about 8:30 this morning.  I think that ISPChannel is really lame, and I will most likely get a DSL line soon.

Tue Jan 18, 2000, 11:00 PM  --  My connection to the rest of the internet died today at about ten to four for reasons which I was unable to determine while I was at work.  I only had a little time between arriving home and leaving again to go see TheatreWorks' latest show, so I rebooted the server.  It didn't help.  When I got home, the system log showed that at about half past eight the ethernet address of my upstream gateway had changed.  I interpret this to mean that they had a hardware failure.  Everything seems to be working properly now.

Thu Jan 13, 2000, 1:30 AM  --  I have decided not to keep the mail server configured as an open relay.  This will prevent exploitation by any would-be spammers trying to send through my mail server, but it prevents anybody else from sending outgoing mail through from any machine which is not on the local network.  If any of the constituents of my enormous user base have a problem with this, there is a workaround.

Retrieving mail from the outside world is still dog slow.  Since it works great on the local subnet, I have a feeling that this might be due to network latency introduced by my cable modem provider.  The latency I see when telnetting in from outside is mediocre, but not nearly as bad as the mail retrieval.  It can take more than ten seconds just to connect to the IMAP server.  I wonder if the slow ethernet card should be replaced, my little script reset it twice in the 24 hours since I upgraded everything.

Wed Jan 12, 2000, 1:45 AM -- The system upgrade is complete. There are some local file sharing problems which aren't working right, but everything else is done and should be back to normal. Hopefully that problem with the slow ethernet card will go away now.

It is now possible to export your own web stuff to the world at large (Becky, once again this means you).  Just create a directory called public_html under your home directory and the URL wil be

Tue Jan 11, 2000, 9:30 PM -- I am presently doing some system maintenance right now, so email service might be a little erratic. No incoming mail will be lost, but you might not be able to read it for a bit.


This page is maintained by