Motor needs push? Replace Start Capacitor

"Blog helps people find solutions. Problems: Google Search based on statistical pattern-matching. Machines don't understand human sentences. Solution: Multiple simple descriptions of problem. Headline writing style freed from articles and helper verbs. Facilitates computer's pattern-matching."

Ack. I can only mangle my natural language for a paragraph at a time. But here's the actual post topic:

"AC Induction Motor runs well. But Motor needs push start. What's wrong? Start Capacitor normally pushes motor. Capacitors degrade often. Capacitors cheap. AC Motor fixed for under $10. Blog post saves Motors from landfill. World saved."

Quite sadly, it takes real research and experiment to discover this basic fact about motors. A running motor, but one that needs a push-start to actually spin, is almost certainly missing its designated pusher -- the "start capacitor". These things are cheap, and easy to replace. There's no reason to throw out a perfectly good motor or, worse, the entire product it's attached to. Split-phase AC Induction Motors (start phase, run phase) are used in everything -- from pumps to furnaces. The fact that they're tossed on the scrapheap, often with the appliance, is nearly criminal.

The major problem is that symptoms are described differently by different people. Sentences are infinitely variable, though highly constrained (actually, any infinite set is highly constrained). So the meaning of "motor runs if I push it" (and the more vague symptoms that precede this discovery) can be expressed in an infinite number of ways. If Google Search was designed with the rather specific internal circuitry of the human brain, it could 'get' what we 'mean' and search the web on our behalf. Unfortunately, that circuitry is unknown to the natural sciences, although a tiny bit of progress is made every year. This is too slow for Google: they've abandoned the idea of understanding how the brain actually works, and instead hang out with people who wrongly think they can divine internal structure from statistical analysis of surface behavior.

Until someone competes with Google on this front (they seem committed to the wrong direction), or until the science of cognition matures, here's the best we can do, to empower people, and to save the planet: encourage everyone to discover the most commonly used words and phrases that could express their problem.

Don't give up! Save the machine!

Google Sites URL mapping

I manage a number of sites with Google Sites. It's clearly not a tool that Google feels heavily invested in. There are occasional fixes, typically years in coming, but there are still significant problems with the HTML generated by the WYSIWYG interface, and many UI fragilities with the editor and manager. I use it anyway.

I found one site unexpectedly redirecting/forwarding to a sub-directory page, in this case a "/home". I wondered what was causing this problem, which I certainly didn't want.

I eliminated the possibility of a problem at the registrar -- the thought was that maybe I'd accidentally added this subdirectory to a field in a zone forwarding directive, or maybe I'd put a redirect in an html page, or in the header of a server response on another host server … although this seemed unlikely.

But, no, those possibilities were not realities.

So, I was back into the mysteries of Google Sites. What was prompting this strange behavior?

The only difference I could find between this and other Google Sites, was that in the "web mapping" section I wasn't using the lower-case name of the sites subdirectory.

But instead of generating an error, Google Sites mistakenly interprets this as a redirect directly to the default directory, visible to the user as a "something.com/home" directory, which was not desired.

It would be nice if they would fix this. And, in general, the mapping console could provide precise information about the nature of the mapping.

So, I deleted the old entry, added a new "web mapping" with a lower-case name, and all was fine.

Demagnitizing Camcorders

Although the reasoning in this original article is interesting, it turns out to be incorrect. It's possible to clear a bit of magnetic dust using blank tapes and cleaning tapes, but serious problems that damage subsequent tapes are caused by dirty rollers, capstans, and video heads. After following the method below, the problem came back a few tapes later, and it was not fixable through the same method. Cotton swabs for the non-head components, and paper or chamois swabs for the video heads, with isopropyl alcohol, are the only pieces of equipment that resolve this problem. The tiny specs of magnetic dirt that interfere with a video head, and damage tape image frames permanently, are very easy to remove with a manual cleaning.

My SONY DSR-PD150 is over a decade old, but I still use it, because it's good in low light and very reliable.

Unfortunately I recently threw it into the back of a car full of big self-powered speakers, and it became magnetized. Perhaps the cause was direct contact with the speakers' magnets, or perhaps moving through their magnetic fields, or perhaps from some kind of reverse "Faraday Cage" effect (magnets inside a Faraday Cage, which usually protects from electric effects outside the cage, but, who knows, might amplify the effects when magnets are inside the cage).

The previously reliable image and sound recordings became spotted with dropouts, digital noise, etc. Definitely magnetized heads.

So, how to demagnetize a camcorder? Is it possible?

Yes, it is possible.

I'll say what worked, and then reconstruct the discovery process.

The solution is:

1) get new tapes
2) with each tape:
3) record once for the length of the tape
4) play back once for the length of the tape
5) throw away the tape
6) repeat 3-5 with new tapes until the magnetizing goes away


This sounds unlikely, I know. Like many things related to magnetism, our intuition fails us. Mine did.

At first I tried two different miniDV cleaning tapes. They were no help. These tapes might be helpful in other situations, but apparently not with magnetized heads or magnetic particles, which I imagine were magnetically attached to the heads. I also tried just recording and playing back on the same miniDv video tape, over and over again: this also did not help.

I found one intriguing post, here, and decided to use it as the basis of a little theory, and think through the likely repercussions. If it's true that the use of a camcorder tends to neutralize, rather than magnetize, the recording and playback heads, then we would need to optimize the conditions under which the camcorder could eliminate the magnetic particles depositing on its heads. The only thing that could do this is the tape, since, being magnetic tape, it is both the source of the particles and a potential mop of them. So, after having tried so many things to no avail, I simply broke out a new tape, recorded the full 40 minutes (DVCAM mode, but it wouldn't matter what the tape format is) and played it back. Still dropouts. But it seems a little better. I threw away the tape, broke out a new one, and did the same thing. On the playback, it seemed like there were markedly fewer dropouts. I threw it out. One more time, and this time, the playback was flawless.

Apparently I'd used the new tapes as "clean magnetic mops" to pull tape particles, and their attending magnetic charge, off of the camcorder head. It takes a certain leap of faith to throw away brand new camcorder tapes, but that seemed to be the only answer. I know from previous experience that demagnetizers are tricky and can make matters much worse. Throwing away a few new tapes is well worth it, since it can refurbish a camcorder.

This previously highly-magnetized piece of equipment had demagnetized itself, tending towards neutral, given neutral tapes to help it along with the clean-up. I don't guarantee it, of course, but, for me, the effectiveness of this technique spoke for itself.

Sleepless MacBooks

My MacBook, like many people's, is an insomniac. If you put it to sleep, from the menu or by closing the shell, it wakes up again, in a few seconds. This default behavior will empty your batteries and overheat your machine in its case.

Now, many things can cause this: mostly hardware interrupt handlers containing instructions to wake up. I first Googled this problem with the phrase: macbook does not stay asleep, but the Apple discussion board is completely disorganized. At the top of the page, it announces "1 solution" but doesn't give you a hint as to where it may be, in pages of discussion! I found it eventually, but I thought it would be helpful just to say:

... turn off Bluetooth.

It's in Systems Preferences under Hardware -> Bluetooth. Your computer will then get the rest it needs, without pills.

Addendum:

If the Bluetooth hardware is failing, the Bluetooth symbol has a little squiggle through it.


This makes the system write a hardware error to the system log (/var/log/system.log) every second. This wakes your Macbook consistently.

In this case, without repairing the hardware, you'll need to put your machine into hibernation, instead of sleep. There's a useful widget for this, called Deep Sleep.

Your Bluetooth hardware may start working again. Anytime you see the grayed out Bluetooth symbol in the upper right, without a squiggle, you'll be able to sleep normally ... if you turn Bluetooth off.

Open Source OCR on a Mac

Tesseract-OCR is an effective open source project from the mid-1990's. The actual character recognition is terrific, so it's not hard to get good results, and it's quite a time-saver.

But, it's not exactly turn-key. You need to:

1. download the source
2. compile it for a Mac
3. download a language file
4. copy it to the appropriate directory
5. run it on TIFF files that need to be renamed to a .tif extension.

Tesseract won't run unless you copy the language file to /usr/local/share/tessdata. Which is strange, because it uses it very irregularly. Most of the miss-read results are simple English words: you get "iist" instead of "list", "lf" instead of "if". It makes you wonder how exactly it is applying this language file.

If you use a Mac utility like Textedit, or Word, or Open Office, the spell-checker can find and help you fix these in a matter of moments. But, still, it's irritating, when you have a long document. This software needs to be 'productized'.

So, the actual sequence:

1. go here, and download tesseract-2.03.tar.gz.

2. In a Terminal window (Applications->Utilities), find your download directory, cd there, and:

: gunzip  tesseract-2.03.tar.gz
: tar xvf tesseract-2.03.tar

3. cd to the tesseract-2.03 directory, then:

:./configure
:sudo make
:sudo make install

4. Go back here, and download tesseract-2.00.eng.tar.gz, then, find your download directory, and:

: gunzip tesseract-2.00.eng.tar.gz 
: tar xvf tesseract-2.00.eng.tar
: cd tessdata
: sudo bash
: cp * /usr/local/share/tessdata/

Then hit control-d to exit the sudo bash shell.

Make a TIFF file, be sure it has a .tif extension, and then issue a command like this:

tesseract document-image.tif document-results

... and then you'll have text in document-results.txt

Works great. It should come standard with a Mac. With a graphic user interface. And some corrections to the language file use.

Server heartbeat script

Simple bash script on Gnu/Linux or Unix system, to monitor the uptime of your server, and tell you, call you (well, text your cell phone) when something's wrong. I just call this one heartbeat:

#! /bin/sh -v
while true
do
  if wget --no-dns-cache --no-proxy --delete-after google.com
  then
      if wget --no-dns-cache --no-proxy -t 1 -T 60 --delete-after \
your.domain's.IP.address/beat.php
      then
        echo "success"
      else
        echo "Heartbeat stopped" | mail -s "MACHINE DOWN" -c \
your@emailaddress.com \
yourphonenumber@txt.att.net; sleep 20m
      fi
  fi
  sleep 10m
done

We need a server-side script ... the beat.php above:

<?php
header("Cache-Control: no-cache, must-revalidate"); // HTTP/1.1
header("Expires: Mon, 26 Jul 1997 01:00:00 GMT"); // Date in the past
?>

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" 
"http://www.w3.org/TR/html4/loose.dtd">
<html>
<head>
<title>Beat</title>
</head>
<body>

<? print(Date("l F d, Y ... g:i A")); ?>

<br>
</body>
</html>

A few explanations:

* We hit Google first just to make sure the web is working. No sense queueing up an unecessary e-mail to yourself.

* wget's delete-after cleans up the result of the wget, which is built to fetch pages from a URL.

* wget doesn't error when your local resolv cannot find a domain name. This is why we use the IP address. If you need to test to see if your domain's zone name server is running, create a script for that purpose. Don't rely on this one.

* most cell phone have a way to receive e-mail / text messages. Try sending a message from your phone to your e-mail address, and reply to it, to discover, and test, your cell phone's e-mail address.

* Of course, all the "cache expire" stuff, and the fresh-content dynamic server side script, is necessary, otherwise when your server goes down, some helpful cache server along the way will provide you with the beat.php results anyway, from your numerous previous wgets. It's best to test this, despite my precautions ... there are lots of cache mechanisms being inserted on the web these days.

Ubuntu command line mail

Ubuntu doesn't come pre-packaged with command line mail. That's strange, because this is one of the most simple, important commands in Linux / Unix admin network scripting. It lets you interact with the real world.

To install, unfortunately, the older package mailutils is deprecated for Ubuntu, even though it's listed in all Google results for [ "command line mail" ubuntu ].

So, installing good old unix command line mail (e-mail) on Ubuntu, as of late May 2008:

apt-get update
apt-get install postfix
apt-get install mailx

Then use mail freely.

Safari form spacing problem

Safari will render a form with extra space at the bottom (typically, right under your submit button), whereas more compliant browsers like Firefox do not. This makes login boxes and other colored form divs and tables render improperly.

The fix is simple ... add this to your opening form tag:

form style="margin: 0px; padding: 0px;"

Safari crash on mismatched tags

If Apple's mac web browser, Safari, crashes on your web page, look for mismatched beginning and end tags. With a fairly complex page, using lots of CSS, we found that Safari was crashed by a simple mismatch of h1 (opening) and h2 (closing) tags ... note that we first debug-interpolated the error to the CSS file, but there was no error to fix ... the typo was in HTML that used the CSS file ...

(keywords: crash, crashes, crashed, safari browser)

Firefox img border problem

Of course, since Firefox is probably the most compliant browser, this is more an artifact than a problem. But I'd like to help people out with this, so I'll keyword it: problem, bug, issue ...

In Firefox an image (img) with (a href) link tags around it, often shows up with a visually unwanted blue border. The fix is simple. Set " border=0 " in the img tag:

< img src="file.jpg" border=0 > 

Good habits: lengthen your history

It's important to know what you've done, as a sysadmin. If you're one of a group of admins, history becomes even more important.

The first thing I do on a legacy Linux system, is edit /etc/profile, and change HISTSIZE to 100000 or more. Note, this does not erase previous history (although etting it to zero will).

Of course, "history" records only one side of the conversation. The responses of the machine are just as important. Unfortunately, there's no straightforward mechanism like history, which would automatically log everything you see. Your terminal client can often do this for you, but it would be nice to do it on the machine itself.

To bug the bash people about this, for their next release, go here.

The "script" command is a temporary fix ... it records your session in a transcript file. You type "exit" or ^d to stop recording. It writes into a file called "typescript", unless you specify something different.

Trying to automate this:

... you cannot just put "script" in your .bashrc file ... it will go into an infinite loop, because "script" itself starts a new shell (when you stop this, amusingly, you'll find yourself at the bottom of a few hundred nested shells ...)

You can't put it in .login ... because script starts a shell. When .login exits, so will the "script" shell. But that's not what you need.

What you need to do, is set an environmental variable in .bash_login, unset it in .basrc, and then start "script" just once.

MSN Messenger very broken on Macs

Microsoft's closed-protocol Windows Live, Microsoft Messenger IM environment will work for a mac under some conditions. If you first authenticate on the machine you first sign in on, you're ok, and the client works passably well (although there are many irritations, like a nearly useless search feature) ... but if you need to take on previously authenticated identities, you're hosed. It simply doesn't work on a mac. The re-authentication e-mail never gets sent. Also, the web ui, "webmessenger.msn.com" is simply broken for any browser that runs on a Mac (Firefox, Safari, and the old IE) ... it loses connections and messages quite easily.

However, MSN Messenger even works well on 8-year-old machines running Windows 2000 Pro ... set one up so you don't need to buy a new machine and give money towards Microsoft's monopoly-driven attempts to create working software.

cp -r hangs. use cp -R.

My primary purpose for this blog, is to make small little nuggets that a search engine will find, when you have a problem, and you haven't slept for three days, and you type in the first thing that comes to you.

I could do this with keywords: cp, cp -r, cp -R, hangs, takes forever, special devices, licq_fifo, linux, unix, large copy, massive copy, etc.

But that's not the way you think when you haven't slept. If you're a human, you ask questions "why is my cp hanging?" or "my copy is hanging, but why?" or just statements "my cp hangs" or "my copy hangs." Very commonly, you think in terms of time-based sequences of statements, as if you were talking to a doctor ... something like "I'm on Linux. I had to copy a lot of data from one disk to another. One disk is remote, using NFS. I used cp -r. It took forever, so I did df, and found that the new disk size wasn't increasing. Then I killed it with control-c. And ran it cp -r -v, to see what was going on. It got hung up on a file licq_fifo. I googled licq_fifo, and found out it was a special device. cp -r doesn't follow symbolic links, why would it not recognize special devices? Oh, wait, I remember this, it's a normal problem. I typed "man cp", and then the more recent "info cp". And, right, cp -R will ignore special device files. I should go to sleep."

This is of course the simplest problem, quite quick to resolve. But look at the volume of problem-solving information in the above paragraph. What can we do to make search engines more like expert systems, so that we can preserve more of the world's problem-solving-sequence information? Not just with IT, but with everything involving such traversals of ordered fuzzy graphs?

Discovering a legacy purpose for LDAP / slapd

If you're on a legacy Linux/Unix system, and you notice slapd running a directory service, you might wonder: what is this directory service for? What is slapd doing? Why is ldap here? Why do I have slpad running? What data is slapd serving? [I'm playing with search engines here -- I typed these questions into google and found nothing quickly.]

A search of the data will answer most of your questions.

ldapsearch

.. is a common LDAP client command on Linux/Unix. It will tell you if no ldap is available ("Can't contact LDAP server").

But, even with a slapd daemon running, you won't find any data by typing this command with no arguments. You need to tell it where to start looking: the base of the ordered data on the local server.

So, find the config file -- usually somewhere like /etc/ldap/slapd.conf. If it isn't there, use "locate slapd" or "locate ldap" to find it.

In the file, you'll see something vaguely like this:

# The base of your directory
suffix "dc=onething,dc=something,dc=else"

Use the string in quotes to search your directory:

ldapsearch -b "dc=onething,dc=something,dc=else"

You'll see the data, and then the purpose of LDAP on your server should be quite clear.

BIOS disappears

I'm putting an EIDE drive in a 7-year-old server, which has SCSI disks from that time -- still working, but they're very small.

There's only one startup setting I need to change:

I hit "delete" to enter the "CMOS Setup Utility".

I go to "BIOS Features Setup".

The setting I need to change is "HDD Sequence SCSI/IDE First".

I need to set it to SCSI, which is still the boot drive. The default is IDE ... the shop that made the server didn't need to change this, because there were no IDE drives ... they should have, but oh well.

But, inexplicably, sometimes, the system doesn't boot up properly. What's up?

Oh. 7-year-old server. The little CMOS backup battery is dead. I've been turning the machine off at the power strip, and it doesn't get the 3 volts it needs to keep my one change to the BIOS ...

In a deadirectory

After some apt-get update and apt-get install, I get this funny error message when I try to restart apache:

"shell-init: could not get current directory: getcwd: cannot access parent directories: No such file or directory"

But the command runs fine.

Something felt familiar ... oh right. This is a UNIX error message for "you are in a directory that no longer exists".

Mysql broken increment

Changing table names, I changed the name of the primary key as well, and forgot to set all the attributes.

If you don't, things start to break, because the attributes aren't "saved" for you in any way. For example, the auto-increment stops working, the table is no longer functional, and you get this error:

Duplicate entry '0' for key 1

To fix this, add the attributes to the primary key field. This won't hurt the values (at least, it didn't for me):

alter table X change column key_name key_name int(12) unsigned not null auto_increment

Upgrading to PHP 5 -- fix 3

And, of course, there's the famous "backslash" or "magic_quote" problem. The only way to sensibly develop a large PHP app is to turn magic_quote Off in the php.ini file. That way, you can get reliable behavior from addslashes & stripslashes. When you port to a new php implementation, however, you'll find that, by default, magic_quotes is On. Edit php.ini, turn magic quotes Off, issue a /etc/init.d/apache2 restart, and your text inserts & fetches shouls be fine.

Upgrading to php5 -- fix 2

I get:

"Warning: mysql_connect() [function.mysql-connect]: Can't connect to local MySQL server through socket '/var/run/mysqld/mysqld.sock' " ....

I notice that my old command line entrance into mysql isn't working, and I get:

"Can't connect to local MySQL server through socket '/var/run/mysqld/mysqld.sock' "

Ah ... mysqld isn't running. So:

mysqld_safe &

... which finds the old config files fine. Errors disappear. Everything's up.

Upgrading to php5

After a major Linux distro upgrade

This message (from php):

Fatal error: Call to undefined function: mysql_connect() ...

Mysql wasn't bundled in. So I typed:

apt-get install php5-mysql

... which fixed the problem

Odd characters in PHP server output

Switched from php4 to php5.

First problem: strange control characters that the browser renders as black diamonds with a question mark inside. Like this: ��

So I saved the html file.

Looked at it with vim -b.

The confused black diamonds were "a0" in hex.

I typed "a0" into Google.

Others had this problem, with perl. It was a matter of content-type character sets (charset): is0-8859-1 vs. UTF-8 ...

Looked at the HTML output of another server I had (with php4) for phpinfo(). Compared side-by-side with the php5 output. Very useful ... a bunch of sections with settings in tables. It was in a section entitled "HTTP Headers Information", subheading "HTTP Response Headers". The good server had the name/value pair:

Content-Type text/html; charset=iso-8859-1

... the bad, problem server had:

Content-Type text/html; charset=UTF-8

So, I looked at the /etc/php5/apache2/php.ini file.

there's a line that read:

;default_charset = "iso-8859-1"

I tried removing the leading semi-colon (uncommenting). Typed:

/etc/init.d/apache2/restart

And that fixed it. The diamonds disappeared.