|skip past calendars
(12/24) Getting Away From
After I wrote my lengthy "End of Two Weeks of SPAM
Purgatory!?", I almost discarded (did not publish) it
because I thought it was too long and not that interesting.
In retrospect, it is obvious that my perspective was distorted,
since there has been significant subsequent positive impact:
What I do really is simplistic.
I am surprised (delighted) that it works as
well as it does because I know so many ways the spammers could defeat it.
Like many good programmers, I am basically lazy in the sense that
I try to get the best results with the least amount of effort.
Of course, that attitude is not limited to programming.
For example, the late, great
Kamakawiwo`ole, in his video "IZ: The Man and His
Music" talking about making music, says "basically ... what
I do it's minimum effort but maximum pleasure, and that's part
of being Hawaiian".
ex-colleagues from the 80s, both of whom I still think of as
friends, are trying to lead the fight against spam, but
didn't even know of the other's efforts.
I've been privileged to get them engaged in an intense dialogue
and read what they have to say to each other.
If that was all my efforts produced, that would be
now and then, Chris Pirillo finds one of my
"tidbits" worth republishing to his audience, which is
orders of magnitude larger than the routine audience for my
postings. While I feared that what I had written was too long and
boring, Chris obviously thought otherwise, since he
reproduced it in his Lockergnome Windows Fanatics feed day before
I've been so busy that I haven't even been keeping up with
my usual RSS feeds, so I started getting queries in response to
Chris' republishing before I knew of the republishing!
ex-colleagues, who are much more spam-fighting experts that I
am, seem to have concluded that my simplistic approach is more
effective and reasonable than they would have thought without
This "tidbit" is even more technically presumptive than the
I'm hoping it will be helpful to a broad audience, yet
definitive enough that I can get back to some of the other topics I keep
saying I'm going to bring to completion, e.g.,
My intention here is to explain my practices in enough detail that anyone
who runs their own mail server can adopt (with or without modifications)
my practices. By far, the biggest assumption is that the mail server
is a Unix oriented machine.
(And to make it easier for me to get this
written, I make some very weak assuptions that the server is
running something similar to recent Red Hat releases or Fedora.)
I've not even thought about doing similar things on a Windows-based
Though everything I've done has only been run on recent
Red Hat Linux or Fedora,
I assume that my approaches would work with any of the
BSD flavors and any of the vendor proprietary Unix flavors, but I
don't even have easy access to most of those.
(As those who have read my past tidbits know, I am very proud of what my
team did in creating Dell Unix V.4 Version 2.2 and I still have a machine
that can run Dell Unix. But that is irrelevant in a production environment.
I also have a machine that can run Solaris 9 X86 or FreeBSD 5.1, but I
haven't found the time to work with either of those. When that machine
is powered on, it is most likely running Fedora or some flavor of
Finally, in terms of clients, what I have has mostly been exercised
with Outlook 2000 for POP and Outlook Express 6 for IMAP.
environment is radically different.
and others are attempting far more ambitious approaches for
Exchange servers. (Somebody must have a good reason for trying to run a
Microsoft-based mail server that doesn't use Exchange, but I've
never heard one. From my perspective, you either use Exchange or a
Before there is any backlash from Mac advocates, (a) Macs have
yet to be established as significant in the server competition, and
(b) I think of OS X as another flavor of Unix.)
I hope that is enough background. I am assuming that anyone who reads
further has already gleaned the basic strategy from the prior posting and is ready for more detail.
One of my challenges in describing things is that my personal usage has
been strictly IMAP oriented, but I expect that most people are more
interested in POP.
Assume an e-mail gets in far enough that this discussion is
I'm assuming that the default Red Hat/Fedora mechanisms are already
in effect, plus all spam-oriented options in sendmail.mc are
enabled, for example, sendmail.mc has
I am pretty sure, based on my server's log files, that such
settings are pretty important.
On the other hand, I don't have any evidence one way or the other
whether spamassassin as supplied/configured by Red Hat does any
good. In my experience, Red Hat has good judgement on such things,
so I accept their judgement when I don't make the effort to make my
All of the above could/should be seen as disclaimers.
The substance of what I do is best visible at
and the referenced files visible as links in
I hope the above is enough to help people use these tools for themselves.
- When I
started this two years ago, I had no procmail experience.
I looked through many examples of procmail-based spam fighting. I should
be giving credit to the examples that influenced me most,
but it was so long ago I don't remember who/what deserves the
credit and thanks.
above links show a very generic POP setup. But what I use
in production for IMAP for myself and my wife isn't all that
different from what you see in those links.
depend on renattach
to neutralize potentially hazardous attachments.
is really not the reference version -- when I see an
address or id that seems suspicious, I run virc.spam, which
changes rc.suspect (the version I use for IMAP) and the perl
expression embedded in virc.spam derives rc.suspect4pop from
I am trusting my "white lists" rc.fromaddressbook and
rc.exempt, I am brutal in rc.devnull and more brutal in rc.suspect.
All the spammers reading this should immediately realize that my biggest
vulnerability is forged "from" addresses.
I see something suspicious, I run virc.spam and change
rc.suspect (and thus change rc.suspect4pop). On rare occassions I
find something so obviously spammish that I change
is a simplistic way to create rc.fromaddressbook
from Outlook "Contacts" exported as comma separated
- rc.suspect4pop is adding an "X-Suspect: [Suspect]" header to
the message. The client must be looking for this header to put the
mail wherever suspect mail should go. For example, with Outlook,
the "Rules Wizard" can be used to put mail with this header in
a folder named "Suspect".
- I used to have some domains in rc.suspect that I would really like to
have left in there. For example, except for my monthly bill,
anything I get from att.com is almost certainly forged. But some of
the most important users of my mail server get lots of genuine mail
from att.com. So I had att.com in rc.suspect, but took it out
to make things right for the majority of the users of my mail
(12/21) End of Two Weeks of SPAM
This will be longish, definitely not a "tidbit".
I hope you will find it worth reading. It concerns
spam, spam filtering solutions, and ISP customer
If those topics do not interest you, you need not read further.
Some of this will seem very technical to some of the e-mail recipients,
but I will try to explain the technical aspects as I write.
Spam is frustrating to all of us. Some say that more than half of e-mail is
now spam. It seemed like spam started escalating dramatically after the 9/11
My wife and I seemed to be victims of the early escalation of undesired
e-mail two years ago, presumably because we had made our e-mail
addresses very visible publicly, especially on our web sites.
Starting in early 2002 I have been crafting a custom solution that has been
satisfactory for the two of us.
Technical Issues: There are two primary Internet e-mail protocols
for picking up mail: POP and IMAP.
Most people use POP (Post Office Protocol). POP stores the mail on the
client, so (unless you tell it otherwise) it deletes the mail from the
server when your client gets it.
If you only use one computer, that's fine. But if you use more
than one computer, POP can be frustrating.
My wife and I use IMAP (Internet
Message Access Protocol) because it stores the mail on the server in such a
way that it is the same regardless of what client computer you use.
Originally, my spam solutions only worked reasonably with IMAP.
(On the other hand, IMAP is inefficient and can be frustratingly
A good friend, very astute technically, called a few weeks ago and
asked about using what I had done. Then the answer was wishy-washy,
since he wanted to continue to use POP.
Week before last, one of my client's people and my client
complained to me about spam. They all use POP, with Outlook 2000.
I told them I would make him a guinea pig for a modified
version of what my wife and I use. I spent midnight to 4:30 a.m. that night
reworking what I had done to make that possible, making a coordinated
IMAP and POP version, got some more sleep, then
spent much of the afternoon tweaking/testing what I had done earlier.
I applied it to my client's account and he seems happy with the changes.
I think what I have done is immediately applicable to anyone who uses my
First, what is spam? Some think it is any unsolicited e-mail.
My wife likes to get e-mail telling her how to enlarge her penis!?
My male friends don't!?
More seriously, if you've ordered stuff from Amazon and they suggest
you buy something similar, is that spam? Some say yes, some say no.
If an outfit you've never heard of tries to sell you Vicodin, we
probably all would call that "spam", even if Hormel wishes
Second, in some sense the spammers are winning.
They're tricking a lot of people.
If you get spam and it gives you a "take me off this list"
link, the last thing you want to do is click on that link.
Spammers are looking for viable e-mail addresses.
Most of the stuff they send goes to invalid addresses.
If you click on a "take me off this list" link, they've
suddenly discovered a valid address and will add your address to their
list of viable addresses, exactly the opposite of what they said and
Third, the e-mail protocols were designed without thinking about this
Unless/until those protocols change, which is not easy, there is
no 100% solution.
It is very easy to forge e-mail addresses.
Spammers have lots of other tools at their disposal.
The most we can hope for is to make spam no more annoying than the junk
paper mail we receive and recycle.
Server vs. Client solutions: Ideally, this would all be dealt
with at the e-mail server.
That way, your dial-up connection wouldn't waste the time of
downloading a virus you didn't want in the first place.
(I'm not going to try to distinguish between spam and viruses.
They're different, but I don't want either of them, and I
use coordinated mechanisms to keep them at bay.)
However, many of the commercial solutions, and there are some very
good ones, deal with things at the e-mail client (i) because there can
be more control at the client and (ii) maybe they can make more money
selling solutions per client than solutions per server.
Open Source vs. Commercial Solutions: There are many good efforts
both from the free software advocates and those trying to make money.
(1) I didn't
want to spend money or time sorting through all of the options and
(2) I wanted to understand as best I could how to deal with the
It turns out that everything I use is either free software or stuff I've
crafted myself. However, my client's request forced me to look at
how to make what I did work with commercial software, specifically
Microsoft Outlook, and I think I have done so.
Perfection: If you're looking for a perfect solution, stop reading.
I don't have one.
What I have is good enough for me, good enough for my wife, and,
I hope, good enough for everyone who uses my mail server.
Because of all the problems listed above, any attempted solution is
going to fail to some extent, either by throwing away mail you want to
see, or making you look at mail you don't want to see.
My bias is to try to never throw away good mail, even if bad mail gets
through. (I have a strategy for neutralizing viruses in bad mail, so
even if bad mail gets through, it is unlikely to harm the computer.)
My Basic Strategy
First, I use an automatically generated "white-list" -
anyone that I (or other user of my mail server) says they want to receive
mail from gets to send me (or the other user) mail.
If George W. Bush (probably forged, since he said he stopped using
e-mail entirely when he took office) wants to tell me how to enlarge my
penis, and G-dub is in my white-list, the mail gets to me.
Part of what I have done is to make it easier to make this
"white-list" be based on addresses the user has put in their
Second, anyone not in my white list who has VIAGRA or Vicodin or
similar words or common mis-spelling of those words in their subject line
gets their mail thrown away.
They can be clever with mis-spellings and get the mail through.
Every day, I (and other users of my spam filters) get a list of who had
their mail thrown away, so if
someone I really wanted to hear from wrote me, I can write them back
and say "so sorry, my spam filter threw your mail away".
Third, I have a growing list of "suspect" domains and
Anything from those lists gets re-routed to a "Suspect" folder, in the
IMAP case, or gets an X-Suspect header in the POP case.
Either way, the "suspect" mail is in a
separate folder and can be quickly scanned, when/if it seems worthwhile.
95%+ of what goes in my Suspect folder is immediately deleted.
Finally, anything that doesn't pass/fail the above tests ends up
in my inbox.
Most of the above is excerpted and abstracted from an e-mail I sent to
clients, friends, and family December 14. The youngest recipient
was my niece just turned 15, so I didn't think I would offend her
or my sister with the word "penis" and so forth.
About 15% of the recipients had addresses at one of the largest
ISPs, which I will refer to as XYZ hereafter. I think everything
I am saying is factual, and there are only two reasonable
interpretations of "XYZ" but I am trying to avoid offending either
one of them. My telephone conversations with XYZ have intended to be
polite and constructive, in spite of XYZ severely trying my patience and
forgiveness. Anyway, the December 14 mail got through to all of the
recipients, even with the potentially offensive content.
December 16 I sent a family-letter, to the same addressess, and the
ISP (XYZ) rejected all of the copies going to their clients. The rejection
message was very unclear and truncated. For my personal account with
XYZ, the rejection said:
----- The following addresses had permanent fatal errors -----
(reason: 554 TRANSACTION FAILED: (HVU:B1) The URL contained in your
email to XYZ members has generated a high volume of complaints.??
Per our Unsolic)
This is literally what it said, except that I have substituted XYZ
for the ISP's domain name.
(I assume they intended to say "Unsolicited" and continue
further, but the many rejection mails I got all stopped at that same
This made absolutely no sense. If my spam descriptive e-mail got
through, including potentially offensive words, why was this
rejected? (A slightly excerpted version of the e-mail is visible at
What URL could be the problem? Certainly not the one for the Methodist
And seemingly unlikely any of the http://technologists.com/ URLs.
I sent e-mail to the postmaster at the ISP and got no response. Surprise.
So I started calling their customer support numbers. I probably spoke to
20 people, most of whom were seemingly not competent for the discussion
at hand. They would give me a ticket number and say they were transferring
me to someone who could help. Half of those transfers were disconnects!
Finally, I got a toll-free number for the postmaster's office.
I called that number, waited on hold for an hour and 20 minutes,
then finally spoke to someone who seemed to have a resaonable idea of how
to diagnose the problem. The first thing he did was have me forward the
rejected e-mail to an address at Yahoo.com! (XYZ is not Yahoo!) When he
read the message, he couldn't see any reason why it was rejected.
He gave me a new ticket number, admitted they were swamped with
technical problems, and said that someone would resolve. He couln't
say how long that would take.
Since I knew that most messages I sent to my XYZ correspondents were getting
through, I realized there was an obvious workaround: put the e-mail on
my web-site, password protect it, and tell the XYZ recipients where
to find it and give them that id/password. That worked. So besides
clumsiness/frustration, all of the problems were solved.
Last night I received an e-mail from my pastor, who is very computer
savvy and aware of what was going on, saying "Thank God I don't
use XYZ. A friend just upgraded to their latest software and now his system
This afternoon, when I was intending to write this, I suddenly
realized there were 3 URLs, not two that might be offending XYZ.
I was pretty sure that http://nwhillsumc.org/ was not the
problem, and I doubted that anything related to
http://technologists.com/ was an
issue, but there was a third domain name in the e-mail.
In the postscript of the e-mail, I had said
P.S. This is not the end of my project, just a milestone. I still want more family e-mail addresses to add to the lists. I still want more photos. I'm also beginning to make MP3s of my out-of-print LPs, etc. One of my accomplishments last week was to help the Red Clay Ramblers make CDs of out-of-print albums they recorded! I'm astonished that I could help them in this regard to help them recover lost recordings of their own music. The MP3s are in a separate password protected directory to avoid copyright violations.
In doing so, I had given the URL for the Red Clay Ramblers web site,
http://members.tripod.com/~RedClayRamblers/. Note that I am not making this
a hyperlink, because that is the URL XYZ is rejecting.
There is no sense in this at all that I can recognize.
http://members.tripod.com/ was one of the first, after XYZ, to
inundate their users with pop-up/pop-under windows.
With that exception, I know of no reason why XYZ should be blocking
references to http://members.tripod.com/~RedClayRamblers/.
When I realized all of this, and made tests that proved to me
conclusively, that I had diagnosed the issue, I called the
XYZ postmaster toll-free number, expecting to be put on hold for
an hour. To my delight, a human answered immediately, seemed to
understand what I was saying, said she was adding the info to my
trouble ticket and that even though they are horribly back-logged,
they should fix this problem in about a week.
(12/17) Quarter Decade Photo Project;
MP3s; Simplistic Spam
Quarter Decade Photo Project
Somehow it always seems like I spend my time on things other than my plans.
Sometimes this is logical, sometimes it is serendipity.
For 2 1/2 years I've been working on archiving family and friends'
photographs, for a number of reasons:
(The URL for the photos is http://technologists.com/photos/.)
- If the physical photsgraphs get lost/destroyed, the digital versions
are much better than ashes or whatever. In my case, this strikes
close to home, since my father's birthplace had a
bad fire over a decade ago, and many family treasures were lost.
On the other hand, I have photos of my mother's
mother's mother's family.
- Many of these photos I had never seen before. If I had never seen
them, then it is likely that other/younger family members had never
- The digital versions allow for editing/enhancement that is impractical
for those without a convential darkroom. (I used to have a darkroom in
the 60s and had access to one in the 70s. Now I depend on what I can
do with scanners and software.)
- I've reached a major milestone, having scanned almost 1700
photos/slides/negatives in the last 2 1/2 years. Not only are these
available on the web to most family members, I've made paper
copies for those who are too old to want to learn to use a browser.
Rise and Fall of MP3.com
One of my real thrills of 1998-99 was the emeregence of MP3.com, and the
ability to make Caroline's 70s/80s/90s recording available to a broad
We were both thrilled in 1999 when some of her songs hit the top of the
charts on MP3.com, not only in folk/country genres, but even her
tribute to Bob Marley Tuff Gong
and some of her other songs e.g.,
being promoted by mp3.com.
Unfortunately, it looks like only the domain name "mp3.com"
will survive, and all of the 250,000 artists' music will
disappear unless/until something is done to make it available elsewhrere.
Fortunately, all of Caroline's MP3s are vislble at http://kaybuena.com/songs/.
Simplistic SPAM Filtering
The other thing I've been doing is making my simplistic spam filtering
solutions useful to all who use my mail server. If they endose what
I've done, I'll tell more, but the biggest limitation
is that my solutions only work for those who receive mail on my server.
(12/1) Making Peace With Windows
back, I admitted that I needed to make peace with Windows XP.
I had three main issues with XP:
(Others have other issues with XP, e.g., the "activation"
requirement. Those issues do not particularly bother me.)
- The new "Start Menu" seemed to slow me down, especially
on notebooks and other machines with limited pixel layouts.
- I had been unable to get my WiFi card to work with XP.
- Some administrative tasks seemed unnecessarily harder than with 2K.
I've newly started attending services at a neighborhood church.
The senior pastor called me and suggested a 1-1 meeting.
At the end of that meeting, I volunteered to help with any computer
problems at the church, other churches and/or non-profit organizaitons.
Bill, the pastor, immediately said he was having problems making his
WiFi connection as secure and functional as he would like.
It turns out that Bill has been working with computers about as long as
I have, and has been working with PCs longer than I have!
Though he's quite adept with managing his own and the church's
computers, sometimes he gets stuck, as we all do.
When I arrived the next day to follow-up, I found out that
(a) Bill's notebook was running XP and (b) the church had
802.11g equipment, with capabilities beyond my obsolescent 802.11b
At first I was stumped, and didn't get things working much better
that they already were.
I installed XP on my notebook, yet again, with several significant
differences from before:
Though clumsier than my experience with Windows 2000, I did get my
802.11b stuff working with XP, including enabling WEP.
- I installed XP SP1 before trying anything else.
- I read the knowledge base articles on the WiFi manufacturer's web
- I set XP for the "Classic Start Menu" and made the other
user interface tweaks that I routinely make when I setup a Windows
machine for myself.
Then I went ahead and fetched the church's 802.11g equipment,
got everything working the way I thought it should, including enabling
I've taken 802.11g stuff back to the church and have it working well
So now my attitude toward XP is similar to my attitude toward Outlook --
in general I'm not a fan of Outlook, but for some situations it is
the tool of choice.
All things being equal, given a choice, I'd use Windows 2000
before using Windows XP.
However, there is at least one thing I can do easily with XP,
enable WPA, that I can't do easily with Win 2K.
So I think I've reconciled with XP at least as well as I have with
The church's primary server is running NT4 -- yet another motivation for
me to get back to
(11/21) Disks STILL Fail (Sometimes
Those of us who remember computing before the last decade probably
remember the great improvement in disk drive reliability that occurred in
the early 90s. Before then, disk drives seemed to be the most
failure-prone component of computers.
"Everyone" was concious of "head crashes" (when a
recording head hits the spinning magnetic platter, usually destroying
both of them).
Backups, mirroring, "Redundant Arrays of Inexpensive
Disks" (RAID) and other strategies were emphasized to cope with the
Seemingly overnight, disk manufacturers dramatically improved
At a time when disk drives seemed to last a couple of years,
manufacturers started quoting "Mean Time Between Failures"
(MTBF) of close to 30 years!
It is critical to realize that this is predicted average
behavior, and that any given disk can fail at any time.
Still, it is very easy to lull oneself into thinking that disk drives
last forever. They don't!
Between my own premises, other commercial premises, and residential
premises, I probably control forty to fifty disk drives. They do fail.
I think I'm well prepared for failure of the most important drives.
(I'm usually obsessive about backups and redundancy.) However, I
got caught this week.
In my experience in the last decade, when a disk drive fails it is
almost always gradual, not catastrophic.
Presumably, the magnetic material fails in spots, and sectors of the
drive become unusable.
Depending on the circumstances, this may go un-noticed, but more
often than not, even the in-experienced user will notice that
something is wrong and at least ask for help.
However, this Tuesday I saw the first catastrophic disk failure I can
remember in over 10 years.
Unfortunately, it happened to the disk drive that is most important
to me, the primary drive on my Linux production server.
My NT4 production server was designed to be a rack-mount server, has
a built-in RAID system and good monitoring software. As long as I keep
an eye on the monitoring software, any significant problem is very
unlikely. (One drive failure would probably only be noticed by me and the
warranty service person.)
However, my Linux production server was really designed to be a desktop
machine and has had minimal disk redundancy.
I had been planning to institute much more formal mirroring when I upgraded
that machine from RH 9.0 to Fedora, probably Thanksgiving weekend.
I still intend to institute the mirroring, but right now I am humbled
and embarassed that that machine failed Tuesday, with a small loss of
data and an outage of several hours.
It could have been worse. I was on premises and noticed the problem within
an hour. My existing redundancy strategies worked as expected so that the
loss of data was minimized. I decided to go ahead with Fedora on
Tuesday, since I needed to do a complete OS install in any case.
That went well. I had been out of town three of the previous four days
and would have had much more of a challenge fixing things remotely. (I
believe I could have done so reasonably, with one of my hot spare
machines and backups. I don't think there would have been any worse loss
of data, but the problem would have not been recognized so quickly and
the recovery would have taken longer.)
For those of you in the U.S., Happy Thanksgiving!
(11/12) Fedora's Fine; nt4eol;
So far, I have no complaints about Fedora. It feels like a good
successor to Red Hat 9.0. The only obvious omission is tripwire.
I created my own ad hoc, simplistic analog of tripwire
in 1998, before I knew of tripwire, and have continued to maintain
it. So the omission of tripwire
probably is a concern to others, but doesn't directly affect me.
I have Fedora installed on all of my Linux machines except for the
production machine that is still running RH 9.0 (and the museum machine
that runs Red Hat 5.2).
Assuming things go as I expect, Fedora will replace 9.0 on the
production machine in a couple of weeks.
Of course, the big questions revolve around updates, business
practices and other potential changes as Red Hat proceeds with Fedora.
For now, I'll hope that those questions are resolved positively.
Tangentially, I have learned a lot more about multi-booting many of
the operating systems in my
In other words, I've spent many frustrating hours installing and
reinstalling many of those operating systems.
The big problem seems to be that they make different, incompatible,
assumptions about disk geometry. I won't rant about that the way
I might want to, but I will say that NT4's "Disk
Administrator" tool was my best friend in resolving the problems.
NT4 Server End of Life
All the above and other activities have impeded my nominal plans.
I'm filling in my experiments and experiences in
but have much more to do.
Because of the above, no new news about
However, I plan to use Fedora to test/fix/enhance mod_auth++ before
I put Fedora on my production Linux server.
(11/06) Brave New World: NT4 2004
NT4 Server End of Life
Huxley probably wouldn't notice, but 2004 is when we'll have
to deal with the real demise of NT4 Server.
I've started nt4eol to
describe my experiments and experiences.
Right now there are four placeholders for additional pages I plan to add.
"and all those
things" (mod_auth++, Fedora)
Except for the citations in the October 30 and November 5 editions of the
Lockergnome IT channel
(thanks Chris!), I don't have much to add about
I continue to use it, test it, and recognize bugs, but I
need to allocate time to fixes/enhancements.
In the Linux world, there's lots of news, especially the Core 1
release of Fedora and Novell's
acquisition of SUSE. I got the Fedora ISOs quickly, thanks to BitTorrent, and am beginning
to assess Fedora as a replacement for Red Hat 9.0. Obviously, there are
going to be many assessing/wondering this, e.g., Red Hat's
Fedora released - the upgrade path for the rest of us?.
My assessment so far is definitely "thumbs up". I think Red Hat
has done the right thing.
Technically, Fedora feels to me like an incremental Red Hat Linux
release. I probably grumbled more about the changes between RHL 7.0 and
RHL 7.1 than I will grumble about what has changed between RHL 9.0 and
Fedora. My evaluation of a new Red Hat release goes through three stages:
Fedora feels more like an incremental Red Hat Linux release than something
new. If I were a product manager at Red Hat, I would be grumbling about
all of the places Fedora still seems like Red Hat Linux 9.x from a business
perspective. For example, on one of the early pages, it says
"Welcome to Fedora Core 1 ... If you have purchased Official Fedora
be sure to register your purchase through our web site,
http://www.redhat.com/." Since you can't purchase Fedora, this
is nonsense. But the similar message that existed with shrink-wrap Red Hat
Linux was apropos. Anyway, so far I am very pleased with Fedora both
from a technical and a business perspective.
- Install "everything" on a machine that doesn't matter and
look for obvious problems. I've done that with Fedora. No obvious
- Install on my "hot spare" server. That server is intended to
be able to take over if either my Linux or my NT4 server fails.
I've just started installing Fedora on my hot spare server.
- Install on my production Linux server.
(10/31) mod_auth++ Beta; "it's the end
of NT4 as we know it"
(I was going to post this 10/30, but how could I not wait for more burnt orange on Halloween?)
1. mod_auth++ Beta
I've solved the biggest problems I had with mod_auth++. Let's call the current version
"Beta". I'm expanding my production use of mod_auth++. If
you're curious, (and willing to assume any risk involved) please
give it a try. The usual disclaimers apply -- I take
no responsibility if something goes wrong.
A special thanks to
Matthew Gregg at the mod_auth_any
project for telling me of their approach to avoiding the "browser
2. "it's the end of NT4 as we
Microsoft is bringing Windows NT4 Server to "end-of-life".
Windows NT Server 4.0: Changes in Product Availability and Support.
My interpretation is that there will be no new fixes, except for
security issues, after this year. Security fixes will stop a year
later, after January 1, 2005.
This seems perfectly reasonable. NT4 is ancient. Microsoft has released two
successor products, Windows 2000 Server and, now, Windows 2003
(Of course, there are sub-versions of both 2000 and 2003 Server.)
However, there are lots of production NT4
servers going strong. My two production servers run NT4 and Linux,
The biggest problems in the upgrading are the directory issues.
Because of the radical changes between NT4 and the successors, there is
no easy answer. Here are some possibilities:
I used to be proficient in dealing with NT4 directory issues, but had
gotten out of practice. I've given myself a refresher course. Soon
I plan to add a new section to this site devoted to all of the above,
plus, LDAP, which is even more important than I realized before.
- (Ostrich mode) Pretend there is no problem. This might actually be
viable for my production NT4 server, since it has minimal
directory issues and is behind a firewall.
However, I'm assuming that by January 1, 2005 it will be
running some flavor of Windows 2003 Server.
- Samba 3+ on Linux (or some other Unix-like environment). This is
plausible. I've experimented with the latest Samba build
(Samba 3.0.1pre1) and see much promise. However, I'm not as
optimistic as Samba 3.0 Does Windows Even Better.
- Windows 2000 Server
- Windows 2003 Server
(10/20) mod_auth++ "and all those
"mod_auth++" started with my frustration with existing
authorization mechanisms that are available with standard browsers (IE,
et al) and web servers (Apache, IIS, et al).
I wanted to be able to control access to web cams, photographs,
and other static content on my web servers in ways that seemed impractical
with the commonly used mechanisms.
After investigating and thinking, I believed I knew how to do so.
I think I have successfully prototyped what I envisioned, at least
with IE and Apache, and believe
what I call "mod_auth++" will also work with other browsers and
servers. There is a first draft document at mod_auth++ which describes what I've done, how mod_auth++
might be used, and the limitations and problems I've recognized.
"and all those
A friend who read that I was making
and hardware work
asked if I was a "masochist". I said "no, I am a
historian". I took his comment as a challenge and brought my Dell 320N+
386SX 20MHz back to life running Windows for Workgroups 3.11, including
an alpha version of Mosaic 2.0.
A different friend said I had created a museum and should make it visible
on the Internet. I wish I could. Unfortunately, 12-year-old software
(and lots of more recent software) would be very vulnerable in the
currently dangerous state of the Internet.
I did install Windows 2003 Server on a couple of machines, but have not
done much more than that. Perhaps for good reasons, Windows 2003 Server
is much less friendly to multi-booting other operating systems (Microsoft
and non-Microsoft) than previous Windows Server versions. In particular,
on one machine that already had Windows 98 and Red Hat 9.0 installed on
it, the Windows 2003 Server install disabled the Windows 98 and
corrupted the Linux install. However, on a different machine that had
Windows NT 4.0 Server, Red Hat 9.0 and Windows 2000 Professsional,
installing Windows 2003 Server did no harm to any of the existing systems.
So I have to assume that the Linux corruption on the first machine was not
intentional. The Windows 98 disabling clearly was intentional.
(10/6) "If Tomorrow Wasn't Such A Long
When I said "Diving
In", I thought I would stop updating this page for a month or
so. I had no idea it would be 8 months! But everything always takes longer
than you think it will. The words of one of my main muses, Bob
Dylan, have resonated with me as I've tried to overcome
bloggers' block and get back to writing.
(I don't really think of this as a daily blog, but I have meant to
write something every few days, not
allowing lapses of months and months.)
What have I been doing?
- Spending my time with personal and family challenges
and blessings. In particular, June 21st I was
father of the bride. Not only was my daughter's wedding a
blessing, I tremendously enjoyed my role and helping/seeing it
happen. (I was
not at all like Spencer Tracy or Steve Martin in the movie
- Pursing the "unifying access control approach that will be both
secure and usable".
So far, this has worked out fairly well.
This was the nominal reason for
taking the writing hiatus and "diving in". I have
alpha+/beta- code working with Apache.
I've used some of the new capabilities for my own production
purposes for a couple of months.
One of my next steps is to finish
and document what is visible at http://technologists.com/mod_auth++/ --
what is visible there now is mostly incomprehensible unless you
look at what I've done to mod_auth.c.
- Becoming a self-taught expert regarding Microsoft Active Directory,
as implemented in Windows 2000 Server. This ties back to my interest
in making LDAP usable for non-experts, since Active Directory is
based on LDAP. However, Active Directory is at least as intimidating
as plain LDAP. Next I plan to go back to plain LDAP and also explore
the reported improvements in Active Directory in Windows Server
- Expanding my already eclectic interests in alternate operating
evironments. What if SCO really puts a damper on Linux? I've been
looking at Solaris and FreeBSD much more closely, understanding
how they work on their own and how they fit with Windows and Linux.
What if Samba 3 really is a satisfactory replacement for a Windows NT4
- Both because of this expansion of interests and my desire to preserve
my access to ancient environments, I've been setting up some
multi-boot machines that allow me to run any of the following,
though not all at the same time. (These are listed in approximate
order of the age of the OS, oldest first. These are in addition
to my usual operating/testing environments: Windows 2000
Professional, Red Hat Linux 9.0, and Windows 2000 Server.)
1 through 4 are on a 12-year old Dell 450 DE/2 DGX!
Part of what started this all was seeing if I could get the DGX running
again, and to see if I could get Linux running
on that machine. It turned out that 5.2 is the most recent Red Hat
release that I could get to work with a machine that old.
- Dell Unix V.4 Version 2.2, which, 11 years ago, was the
best x86 implementation of Unix. It was based on the latest
AT&T SVR4 and included many extras, notably the Roell
X-server (pre-cursor to XFree86)
and lots of useful public source packages.
- Windows 95 (OSR2) with IE 5.5.
(I'm tempted to bring up a Windows 3.1
environment that works with TCP/IP -- I've got a 20MHz 386sx
notebook that only knows NETBEUI and IPX/SPX right now.
- Windows NT4 Workstation with IE 5.5.
- Red Hat Linux 5.2
- Windows 98 with all the latest Microsoft updates.
- Windows NT4 Server with all the latest Microsoft updates.
- Solaris 9 X86
- FreeBSD 5.1
- (Soon to come) Windows 2003 Server.
There are at least two serious omissions from this list:
- Windows XP. I've tried it numerous times on different machines
and just don't like it. One of these days I'll have to
make peace with XP, just as I had to make peace with
Outlook, but that took several years.
- Macs. I have two ancient, non-functional Macs that might be
cobbled together into one functional system. What I really need
to do is buy a modern Mac. But I haven't bought a modern PC
for myself in quite a while, so I'll probably get a Centrino
notebook before I get a new Mac. (I do have NextSTEP 86 and
compatible hardware, since the X86 port was developed on
prototypes of the Dell 450 DGX, but getting that working again
seems much less important that a modern Mac.)
(2/6) Valuable Distractions and Discoveries:
I've not written one word of my intended
Rather, I've been pursuing a
"unifying access control approach that will be both secure and
- I discovered a grant solicitation that seemed closely related to what
I'm working on, so I submitted a funding proposal.
This may have seemed a distraction, but the thinking and learning
were very valuable even if my proposal is not funded.
- I think I have come up with a secure scheme for new authentication
and access control mechanisms that will integrate nicely with existing
web browsers and servers.
It seems to fit nicely with the Apache web server.
There seems to be a natural way to do equivalent things with
This is a meaningful discovery if, as it seems, there is a way
to provide improved authentication and access control mechanisms that
fit well with existing code.
It will be a victory for software architecture if this works without a
huge coding effort.
- Now it is time to (i) dive in to the details of the existing
Apache authentication modules and (ii) build new modules with new
Having never even built Apache from source
code before now, there is probably much to learn.
already found what looks like a minor bug in one of the existing
authentication modules, and think I have a fix for the bug, so
the next step is to build the repaired module and test.
[koko] (welcome to …) eight Jurassic O.S. on 1992 Dell 486D/50September 26, 2019
koko: reviving timbl's WorldWideWeb browserJuly 1, 2019
koko: exploring NEXTSTEP 486July 1, 2019
koko: prolonging JAWSJuly 1, 2019
koko: sustaining Dell UNIXJuly 1, 2019
1992 JAWS demo for Stewart CheifetMay 17, 2019
a plug for TUHS (just sayin...)November 27, 2018
Let's start at the very beginning... 801, ROMP, RT/PC, AIX versionsMarch 8, 2017
NeXT, give Steve a little credit for the WebOctober 8, 2011
U.S. Patent 6,212,547 — It's not over 'til it's over, but ...Febuary 2, 2011
Definitely better than being there — IMTC annual meeting over the InternetNovember 3, 2010
Mainstream Videoconferencing available againFebruary 14, 2008
A brief history of Dell UNIXJanuary 10, 2008