Raspberry Pi Zero W, back from noboot

I had a brief power outage recently, and my Raspberry Pi Zero W did not survive - but the original Raspberry Pi Zero right next to it came through the power bounce without problems! (Power and cable modem internet have been bouncy this month.)

I tried to power it up on two known-good power sources. I tried another micro USB power cable. I didn't see any lights any time I tried, so I thought this Zero W was completely dead.

Then I ran across this suggestion to try a different SD card. Since that's so simple to try, I did, and now my Raspberry Pi Zero W works again! The old SD card spewed errors all over dmesg when I tried to mount it on my computer, so that was definitely the problem.

So now I'll add trying a new SD card to the list with power and cable to get a Raspberry Pi to boot.

Canon MX870 Print Head "Repair"

I'm usually cranky when the directions I want are only available on YouTube because I can read much faster than the video plays, and I find the extra time to be excruciating. Seriously, please just tell me in words!

A few months ago, my Canon PIXMA MX870 printer decided that it would mostly print yellow: no cyan printing, and only sometimes the faintest magenta printing. I suppose the other colors weren't cheerful enough? Some searching indicated that the problem was the print head. I tried all of the printer's built-in cleaning routines to no avail. I looked at replacement printers (I mostly print kid pictures), but I decided to see if I could fix it. It wasn't useful the way it was, replacing it seemed likely, so I had nothing to lose for trying. After a while, all remaining hints pointed YouTube. I decided that removing a print head might be best described in a video, so (prepare yourself for the shock!) I watched YouTube videos about printers.

The video that worked for me was "How to remove and clean a Canon printhead" and it delivered on the promise of its title! I thought I might need to watch "The secret on how to remove the Canon print head" but the first video with some gentle but persistent jiggling was sufficient.

What I had to do was to remove the ink carts, then remove the print head. I rinsed the print head in warm water from my kitchen faucet at a low pressure for several days (in between bouts of going about my regularly scheduled life), until I no longer saw any ink bleeding out when I did so. Yes, I spent days rinsing the print head, but it was essentially a free experiment. La, la, la, several days of rinsing and a day of drying later, I re-installed the print head and the ink carts. And my printer worked again! I couldn't believe something so cheap and easy actually resurrected my printer, but it did! I'm so glad I didn't replace it hastily!

Fixing Mail.app's Unread Message Count

After a few days of looking at the Dock badge on Mail.app reporting 172 unread messages (untrue), I searched for a solution. I found this one, Unread message count is wrong in Mail and Dock, with a suggested process.

• Quit Mail if it’s open.
• In ~/Library/Mail/V2/MailData, delete any file that begins with “Envelope Index,” such as Envelope Index or Envelope Index-shm.
• Your home Library folder is hidden by default. To display it, choose Finder > “Go to Folder” and then enter “~/Library.”
• Open Mail.
• Mail creates new Envelope Index files. This process may take a few minutes, depending on how many messages Mail is reindexing.

I gave it a whirl ... my laptop fan spun all the way up, noisy, long wait ... and it mostly worked! Mostly means that the Mail.app badge in the Dock showed the correct number of unread messages, but I was no longer seeing any messages in any of my Exchange (ewww) mailboxes. Oops! I deleted and re-created the account to get everything working again.
I think my Mac is jealous of my Xubuntu laptop, so it's trying to tie up more of my time. I think that's gonna backfire ...

Android Camera crashes

Unfortunately this struck while I was on vacation! The camera on my Android phone started saying "The application Camera (process com.android.camera) has stopped unexpectedly." with "Force Close" as my only choice more often than not. The suggestion of resetting all Camera application options didn't work for me until I also rebooted. I did not lose any pictures, just customizations like geotagging.

Easy Fix

Well, I think that was one of the easiest fixes ever! I had an external monitor hooked up to my home laptop for extra pixels. I used the mini DisplayPort to VGA adapter that came with my laptop. The monitor developed interesting colors and patterns, so I quit using. When my 4-year-old asked for his own computer (ORLY?), I borrowed a Mac mini from a co-worker, and decided to try the failing monitor. However, the Mac mini only had DVI so I dug up a DVI cable ... and the monitor has been working perfectly ever since! Something in my VGA chain (the adapter, the cable, or the monitor's VGA connection) had gone bad, but it didn't affect the display on the DVI chain.

I love it when monitors are that easy to fix!

Grandstream and Belkin

Our telephone service has been so awful (usually we have a lot of static on the line, and it goes out completely once or twice a year) and the price for basic service keeps going up, so I finally relented on my "different connectivity modality for redundancy" stance and picked a VoIP service. We're trying to get it to work right now. I hit a problem that's obvious to me as a network engineer, one that's probably tough to diagnose otherwise.

We have a Motorola Surfboard SB5100 Cable Modem that has its moments of showing its age. When it takes a nose-dive, it uses 192.168.2.100 as its WAN IP address; since that address is not routable (RFC1918), we lose Internet connectivity. However, it randomly comes back to life on its own, so Time-Warner doesn't want to replace my cable modem. The lesson here is that when I log in to the SOHO router and see 192.168.2.100 as the WAN IP address, I assume the cable modem is ill.

Following the directions, the next item we hooked up to the cable modem is the Grandstream HT502 VoIP box. On one side, it connects to the cable modem, so it gets the cable modem's WAN IP on its WAN interface. On the other side towards our home LAN, the default IP address is 192.168.2.1. In order to make almost all network configurations plug-and-play, it also serves DHCP out that interface for your home network. It uses the range 192.168.2.100 to 192.168.2.199 for DHCP by default.

To troubleshoot this, I plugged in my MacBook Pro to the LAN interface of the HT502. I checked my ARP table among many other things, and I saw the MAC address of 192.168.2.1 my default router change. That's when I knew the problem.

But the desired operation is to plug in our SOHO router next. We have a Belkin F5D7230-4 that has some pretty sweet advanced features with a simple interface. Yeah, it's also old and needs an upgrade since it has that Flash bug. Belkin hasn't released firmware to fix a router this old, so I have to reboot it when I can read email but not browse the web. I've had several Belkin routers (I love some of the unique features), and they all use the default IP address 192.168.2.1. Where have we seen that before? Yeah, you can't have two systems with the same IP address on the same network because IP addresses must be unique in order to address unique machines naturally.

Of course, the initial red herring was that I logged in to the Belkin router, and saw that its WAN IP address was 192.168.2.100 because the Grandstream HT500 series, like every DHCP server I've ever seen, starts handing out IP addresses from the bottom of its range. I was sure our cable modem had indigestion, but not so sure that I didn't follow proper troubleshooting data collection first.

Anyway, the fix was easy. Hook up cable modem to VoIP converter to laptop, then use my laptop to browse to http://192.168.2.1/ and go in to the settings to change the default IP address to another permissible RFC1918 private IP address; I picked 172.29.2.1 and changed the first two octets of the DHCP range as well (172.29.2.100 to 172.29.2.199) as well. Update, reboot, renew my laptop's DHCP lease (172.29.2.100 as expected). So then I put the Belkin router on the LAN interface of the Grandstream, and put my laptop on one of the LAN ports of the Belkin router, and network connectivity was back to normal!

We'll have to overlook, for now, that the Grandstream only once gave a dial tone, and I wasn't here to hear it! That will need to be fixed, but for now, I've climbed enough troubleshooting mountains for one evening.

T42 backlight

My T42 has gone from half-display (the left 1/3 of my screen would be black) to display wiggles (hard to read, hard to know exactly where to click; the jiggles were just the external monitor, though) to shorter times after cold boot with b(l)acklight. At first, it seemed like I only had the display problem if I didn't use this laptop continuously, so I thought I thought I had the display stays black after resume problem; you know, back in the days where I had some runtime before the backlight turned off. Right now, I get about one minute of backlight before it turns off and the screen is essentially black unless you squint in a bright room. Sounds like I'm not alone, so I have switched from massive abuse of my xorg.conf to asking the spare parts guy. I hope this new tactic works!

DNS: dispatch shutting down due to TCP receive error

I finally got annoyed at the "dispatch 0xhexcode: shutting down due to TCP receive errors: connection reset" in the logs on one server, so I asked Google.

The best answer seems to be from SANS ISC, but even the simple "the other end shut down prematurely" is enough to point out it's probably not a problem and it's not on the server's end.

Bad Cable

It's an obvious statement, but when you're troubleshooting, always verify that your cable is good.

I've been wondering why my Treo wouldn't sync recently. First it was a rare problem, and now it's rare when it does sync. I didn't associate this at first, but sometimes it hasn't charged when I plug it in (oddly enough, it does sync some of those times). However, this seems like the cable is going bad.

Whew! That's an easier problem to solve!

Buddi and Import QIF

I downloaded the Import QIF 3.0.0.1 plugin for Buddi 3.0.0.7, but then I had a terrible time importing the data I just exported from Quicken Deluxe 2002 for Macintosh (OS X).

1. Export each account one at a time. (Pain.) Do not check any export options, just a single account with no categories or budgets or anything else. I don't think I needed to convert line endings (tr '\r' '\n'), but I did anyway.
2. Delete all lines before the date of the first transaction (with the account information).
3. Now I could import that modified QIF into Buddi without ERROR: java.text.ParseException: Date could not be parsed:

In retrospect, that doesn't sound too bad, but I sure did spend a lot of time modifying date stamps before I tried this!

The good news is, I moved my data, and if I'm lucky, I won't go back to Quicken and Intuit's terrible customer service!

Keyboard Problems Again

My laptop keyboard's been out most of the past month. It was never right after it got doused in water, even though I replaced. There's a fuzz connector under the T, G, and Y keys that connects the keyboard to the motherboard. Ever since I replaced the keyboard, I've occasionally had to press down on those three keys very firmly in order to regain keyboard and trackpad. Well, one day last month that trick stopped working. (Luckily I have an external USB keyboard and mouse so I can keep working.)

The ambient light sensor for the backlit keyboard always works, so there's power to the keyboard. However, none of the keys work (not even the CAPS LOCK light).

This failure is different, though. I have keyboard right after the laptop boots, but after a while, it goes out (and no amount of pressing on TGY brings it back). For instance, one time I lost keyboard in the middle of typing 11 minutes after a reboot (I was on AC). The next time I was on battery, and it took two hours to go out. That would indicate a heat problem, and fuzz connectors aren't the most solid connection under the best circumstances. However, I think it's load-related (keyboard goes away when I see the CPU head up) more than power-related. So still a heat issue, but more about CPU than AC versus battery.

However, I collected a list of suggestions from the Internet on what to try for PowerBook G4 keyboard problems.

1. Disk Utility's Repair Permissions
2. Zap RPAM by pressing COMMAND-OPTION-P-R immediately after the startup chime. (This was tough, until I let the laptop cool off for a couple hours so that the internal keyboard worked because the external keyboard isn't loaded in time.) I also tried the Open Firmware zap by pressing COMMAND-OPTION-O-F immediately after the startup chime and then running these commands: reset-NVRAM, set-defaults, and reset-all which made it reboot normally.
3. Reset PMU by pressing and holding the power button for 5 seconds when the laptop has no AC and no battery.
4. Try safe boot by pressing the SHIFT key after the startup chime.
5. sudo rm ~/Library/Preferences/com.apple.BezelServices.plist /Library/Preferences/com.apple.BezelServices.plist
6. Reapply last combo update.

Yes, these are predominantly software fixes for what appears to be a hardware problem, but it doesn't cost anything to try. It does make the new MacBook Air very tempting, but I don't want onboard Intel graphics.

1. Update 1/19/2008: Repair Permissions didn't fix the keyboard.
2. Update 1/21/2008: Neither form of zapping the PRAM fixed it.
3. Update 1/27/2008: No luck resetting the PMU. Twice.
4. Update 1/29/2008: Safe boot didn't fix it.
5. Update 1/31/2008: Removing BezelServices didn't change anything keyboard.
6. Update 2/12/2008: Combo updater was not the ticket either.

So it looked like a hardware problem, and sure enough, I think this proves it. The fuzz connector between keyboard and motherboard can only be reset so few times, and I went over the limit. This laptop will just have to stay docked ...

Taboops

Ah, I finally figured out why I couldn't close windows (just tabs) in Safari! I still had the Taboo 0.3 bundle loaded. Remove it (Safari now warns you when you're quitting with multiple tabs or windows open), and I can close Safari windows again. Whew. It's pretty annoying when you can't close Safari windows!

Mac Sleep Issues

For years, I've seen a problem with sleep on desktop Macs running OS X (both mine and my mother's). What's odd is that I never saw this problem with laptops going to sleep. I decided that the problem was probably caused by having multiple hard drives in my Mac towers. Although I have a home network, my mother did not have one when her Mac crash-on-wake-from-sleep problem started. The work-around is to (remember to) select Sleep from the Apple menu instead of letting it go to sleep on its own.

Things I tried:

I tried unplugging all USB peripherals except mouse and keyboard (pain in the rear) before sleep.

Remove the SCSI card. (There's a G4 Firmware Update for that problem. On a side note, I got rid of a random kernel panic when I removed the CompUSA USB card, so don't discount removing cards.)

Use different time values for sleep, monitor sleep, and hard drive sleep.

On laptop, make sure that a/c and battery settings are the same in Energy Saver if having wake from sleep problems when changing a/c status.

Delete /var/db/SystemConfiguration/com.apple.PowerManagement.xml as often as needed.

Reset PMU. (How to do this depends on your hardware. Search Apple.)

Of course those didn't work.

Here's the twist, support coming from an unusual direction. My boss' boss also uses a Mac, and often asks me support questions. Yesterday, however, I learned a trick from him. John said he'd been having trouble with sleep until he changed the order of his hard drives. Apparently if OS X is installed on the second drive (and I'll have to think about what that means when I don't have a SCSI ID; maybe he means IDE slave instead of IDE master), he has sleep problems. He figured that out when Leopard wouldn't install on the blank second drive he uses for backups. (With cable select, you can just pull the master.) I think I need to open cases and take a peek to see if I can explain the problems we've had for years!

Fixing the Linux VNC oddity

This morning's annoyance was that old Linux VNC problem I had a while ago, and I finally solved it. I found a hint that TightVNC (a fork used by Apple for Apple Remote Desktop, known for good behavior on slow links) compression is incompatible with RealVNC (the main code base). I bet unknown message type 242 means I don't understand TightVNC encoding when you get down to it.

If I use Vine Server based on RealVNC, my Linux can control my Mac and my Solaris still can too! Don't forget to set a strong password and to open your VNC ports in the firewall.

I'm sure the reverse is also true: I could probably install TightVNC from DAG on my Linux, and then view Apple Remote Desktop (which is easier since it would open the firewall for me).

WebShell

So the cool tool of the day is WebShell (ssh for iPhone), but we couldn't get it working yesterday when we installed it.

The answer wasn't RTFM, but on the forum indirectly. Like most people, I run ssh on a non-standard port. You need to change webshell.py to look like this (replacing 22 with whatever you use instead of the default port 22):

[nethope@server WebShell-0.9.5]$ grep ssh webshell.py

cmd = 'ssh -p 22'

help = "set shell command (default: ssh localhost)")

The original line is just cmd = 'ssh', and that's where you make the change.

Whew!

Linux VNC oddity

I turned on Apple Remote Desktop on my MacBook Pro, but I couldn't VNC into it from my Linux box. However, I could VNC into it from my Solaris 9 box, and from my Mac at home (the one that mattered) with Chicken of the VNC 2.0b2. All Macs running 10.4.10. Linux gave me an unknown message type 242.

Here's what I saw on Linux (2.6.9.55 RHEL4AS):

vncviewer 1.2.3.39

VNC viewer for X version 4.0 - built Jan 10 2007 07:48:23

Copyright (C) 2002-2004 RealVNC Ltd.

See http://www.realvnc.com for information on VNC.

Fri Jul 6 17:30:58 2007

CConn: connected to host 1.2.3.39 port 5900

CConnection: Server supports RFB protocol version 3.889

CConnection: Using RFB protocol version 3.8

Fri Jul 6 17:31:02 2007

TXImage: Using default colormap and visual, TrueColor, depth 24.

CConn: Using pixel format depth 6 (8bpp) rgb222

CConn: Using ZRLE encoding

CConn: Throughput 20069 kbit/s - changing to hextile encoding

CConn: Throughput 20069 kbit/s - changing to full colour

CConn: Using pixel format depth 24 (32bpp) little-endian rgb888

CConn: Using hextile encoding

unknown message type 242

main: unknown message type

On the other hand, here's what I saw on Solaris 9:

vncviewer 1.2.3.39

VNC server supports protocol version 3.889 (viewer 3.3)

Password:

VNC authentication succeeded

Desktop name "a.b.c.d"

Connected to VNC server, using protocol version 3.3

VNC server default format:

32 bits per pixel.

Least significant byte first in each pixel.

True colour: max red 255 green 255 blue 255, shift red 16 green 8 blue 0

Using default colormap which is TrueColor. Pixel format:

32 bits per pixel.

Most significant byte first in each pixel.

True colour: max red 255 green 255 blue 255, shift red 0 green 8 blue 16

So if I can use something other than Linux, my VNC viewer to Apple Remote Desktop works. I wonder if Linux is just using a newer version (RFB remote frame buffer protocol version 3.8 instead of 3.3) ... but (intelligent!) backwards compatibility would be nice! I really didn't troubleshoot any further than this because it was Friday evening, and I could go home if I had reasonable confidence that I could VNC in to my MacBook Pro. Since Solaris worked, I didn't much care why Linux failed. Sure enough, my Mac at home also worked, so I went about finishing that long task, and enjoying my precious weekend.

Update: solved 10/02/2007!

Macintosh, Troubleshooting, Unix

ThinkPad T42 Oddity

I have a ThinkPad T42 at work. Usually it runs Linux (RHEL 4 AS), but once every 3 months I boot to Windows (XP SP2) to change my domain password (and to run Windows Update). If I knew a way to use SAMBA from Mac OS X or Linux to change my AD domain password, I might not boot to Windows as often as once a year! That would be nice, but unfortunately password changing isn't that easy.

I have an external monitor, a Dell 2405FPW, connected to it. Usually the monitor works fine. However, after I've booted to Windows, once I boot back to Linux, the monitor shakes. The scanlines don't line up anymore, and it looks for all the world like a hardware problem. However, if I unplug power to this monitor for two days (I haven't tested a shorter time period yet), it goes back to working. So Windows is doing something evil to my monitor that only shows up for Linux! Once the monitor has the shakes, it doesn't matter if I remove the BlackBox KVM or the Dock from the cabling.

I use the DVI connector to this monitor with my Mac OS X, and I use the VGA connector for the KVM and Dock (VGA KVMs are significantly cheaper!).

I'm still working on a theory to explain this, but at least I have a workaround, to leave it unplugged over the weekend.

UPDATE: a co-worker suggested that Windows detects the monitor, that also has card slot readers, and downloads new software (it would have to ask permission, and burn, new firmware, so I don't think it's technically firmware) to the monitor. Then when I go back to Linux, the new monitor software and Linux don't get along. It's a theory, but since I have a work-around (don't use Windows, or use a different external monitor with Windows, or leave it unplugged over the weekend), I'm not going to worry about it.

Brother Printer Settings

When we moved almost 4 years ago, our HL-1650N Brother laser printer lost its settings. That's fair, since it didn't have power for days. However, there's one subtle setting that needs to be changed from the default for printing to work! Last time, I printed the settings page and used highlighter so I didn't have to figure it out a third time, but I think I can find a blog entry more easily than a piece of paper (even one stored underneath the printer!).

From the printer's home page, go to Printer Settings and then to Printer Setup; change Emulation from Auto to BR-Script 3. Save.

That's it! One teensy change, but it's the difference between printing and frustrated!

Birdsnesting

Daddy bought an embroidery machine for me as a mom present while we were at the mall the day before my water broke the day before Karston was born. It is a Brother PE-180D, I knew it was well-liked, and it was an incredible price (1/4 the best price on Froogle, and 1/8 the best price I had ever seen in person) because it was the display unit. However, I've had to solve a number of open box problems. The first was that I didn't have the manual. Luckily the on-screen help covers almost everything. Then I found a link to the manual at SewForLess. (I couldn't find the manual with Google, and Brother's web find only wanted to sell me a printed copy. The kicker is that the link goes to Brother's web site!) The second problem was that I had the wrong hoop. Yes, the dangers of open box: not only are some components like the manual missing, others could be just plain wrong! After careful analysis of how the hoop went on the arm (don't forget, we're a two-engineer household!), we knew they didn't go together. There was no way to latch the hoop securely that also held the fabric flush with the bed so that it wasn't a hazard. (I broke two needles proving this before the careful analysis. The first needle break launched one shard just below my left eye, and now I always wear safety glasses when embroidering. I bet you didn't think safety glasses belong in a sewing room! but they do in mine.) I ordered the hoop from SewForLess since they supplied the manual link. But I still had a serious problem with a large snarl of thread on the underside of the fabric that eventually would break the upper thread. First I had to discover the name of this problem: birdsnesting. Once I knew that, I could troubleshoot with Google. At the sewing machine fault finder, I learned that birdsnesting is usually caused by improper upper threading. At first I scoffed, since I have a lot of machine threading experience. But I re-threaded it very carefully and slowly, following the on-screen directions, and the bird's nest disappeared. In this case I think my sewing and serging experience was part of the problem: I was sure I could just whip that upper thread through its path, no need to read the directions or to be careful! Now I know that I need to be very careful, and that following the on-screen directions helps. I still feel that I can thread the needle faster by hand than with the F.A.S.T. system, so I do let my experience over-ride that part of the directions. But slowing down vastly improved the results! Then my final problem, on my second test embroidery, the upper thread broke. When I grabbed the thread end, it was obvious: the large embroidery thread cone, 6 inches tall, wasn't unwinding freely. I was using my serger's stand and thread arm and reaching over to the embroidery machine, but it wasn't a good kluge. So Daddy installed the thread cone stand from my mom, the thread flowed easily, and the embroidery just worked. Ahh... since embroidery was part of our Christmas gift plans, the last two solutions were just in time!

I can now embroider when I want (OK, a year and a half later). The only remaining problem is that sometimes the bobbin thread pulls through to the front for a while. To keep it from showing up, I just wind a bobbin in the same color. Having equal weight upper and lower thread usually helps sewing anyway, and Singer Type 66 bobbins are easy to find. I might solve that problem in time, but for now I'm happy that I can now embroider what I want when I want.

dead keyboard

Well, one thing about mobile toddlers is more accidents. This time the accident was tipping my almost-empty water onto my laptop's keyboard. I shut down as fast as I could, let it dry a couple days ... but the keyboard didn't survive. I can't use seven keys on the right-hand side, including the letter 'i' and the '-' hyphen symbol (and shifted versions) as well as well as the ever-useful space bar and return key. I couldn't figure out how to log in at first, but then I remembered that I could (duh) use an external USB keyboard. No space and no return really slows down typing.

So this evening, I ordered a new keyboard from iFixIt since they have the right keyboard in stock, a reputation for good service, and detailed take-apart directions. Here's hoping!