Mac OS X: Will the glitches ever disappear?

Posted by Pierre Igot in: Macintosh
April 11th, 2004 • 10:44 pm

I realize I have a somewhat complicated setup. I use a number of USB and FireWire peripherals. I run a number of pieces of software, including a few that involve kernel extensions or “application enhancers”. I also use a dial-up connection that is shared through an AirPort Base Station, which is probably stretching it as far as networking is concerned.

In other words, my computing environment is not exactly a “plain vanilla” setup, and is probably rather difficult to duplicate for the purposes of isolating and fixing bugs.

But still. We’ve been through several generations of Mac OS X now, and today, in April 2004, I am still struggling with basic issues that constitute a major disruption and waste of time and energy.

I don’t know exactly when it started, but it’s fairly recent. I didn’t have the problem in Jaguar, and I don’t think I had it in the first couple of months after installing Panther. These days, however, if I leave my computer running during the night, with the displays automatically put to sleep after 20 minutes, but the CPU and the hard drives staying awake, and various applications running in the background — and if the AirPort Base Station happens to lose its dial-up connection to the Internet during that time (my ISP disconnects me automatically after 50 hours, and sometimes the connection is lost before that too), when I come back to the machine in the morning and try to wake it from sleep — nothing happens.

No matter how hard I press any of the keys on the keyboard, no matter what USB device I try to activate, no matter what FireWire device I abruptly disconnect in a desperate attempt to cause an I/O error that would somehow force the computer to wake from sleep — nothing happens. The computer is frozen in its sleep.

I can go to my wife’s PowerBook and re-establish the dial-up connection in the Base Station — but it is to no avail. My G4 will not wake from its frozen sleep. I have no choice but to do a hard reset and lose all kinds of settings, unsaved things, etc.


(To make matters worse, Safari seems to be unable to remember visited pages that are older than 48 hours. I have no idea why. It seems to me that Safari used to remember things for up to a week or even more. These days, when I restart my G4 after such a freeze, I only have the pages visited in the last 48 hours in my Safari history.)

Of course, I have my suspicions. First I suspected one of my external FireWire hard drives, a Kanguru QuickSilver, which had been acting up. (It’s being sent back to Kanguru, for the second time…) But last night the Kanguru drive was powered off. And it still happened.

Today, I suspect Panther’s Mail application. It has had freezing problems in the past. It has been updated as part of the latest Mac OS X updates. And it’s one of only a handful of applications that I leave running during the night and will be trying to establish Internet connections (to my various POP servers) at regular intervals during the night. I wouldn’t be surprised if the loss of the dial-up connection would somehow cause Mail to freeze (even though it normally just takes accounts offline in such a situation).

So I guess my next troubleshooting step will be to quit Mail every night before going to sleep, and see if that changes anything when the dial-up connection is lost some time during the night.

There is absolutely no point in my sending a bug report to Apple at this point. The behaviour is too vague. There is too little available information about what might be happening during the night. I have managed to isolate the probable cause (the loss of the dial-up connection) — but I do not know for sure whether it is Mail, or another application running in the background during the night, or one of my USB/FireWire devices (somewhat unlikely, since they don’t rely on the Internet connection to operate as far as I know). To make matters worse, it’s not exactly something that I can troubleshoot in the space of a few hours. I have to try different things and leave my computer running through the night and see what happens if I get disconnected during the night. It might take me weeks before I can more exactly pinpoint the source of the problem!

Still, the fact that I am still struggling with such fundamental glitches now that Mac OS X is supposed to be a mature, ultra-stable computing environment is rather discouraging. I know some people might say that I need to get rid of all the gadgetry (USB devices, software enhancements, etc.). But my gut feeling is that it wouldn’t change anything. After all, until recently, I was using all these “gadgets” with no significant problem with waking from sleep. I was able to get weeks of continuous uptime sometimes.

What has changed recently? Well, I have upgraded to Mac OS X 10.3.3 and Mail has been updated to version 1.3.4…

Yeah, I am afraid that discouraging is the word here. <big sigh>.

8 Responses to “Mac OS X: Will the glitches ever disappear?”

  1. Pierre Igot says:

    (a) I think the main issue with my network is that it’s not necessary part of the most commonly tested configurations. So if something in a system update breaks things for my particular config, it won’t necessarily be noticed…

    (a2) Thanks, but no thanks :).

    (b) It’s another problem with my config: I have 20 different e-mail accounts in Mail. Needless to say, Mail sometimes has trouble cramming all its connecting needs through my pokey 28.8 kbps connection. And I have seen Mail freeze on me in the past, sometimes temporarily (for 30 seconds), sometimes completely. I wouldn’t rule it out.

    (c) I’ve tried pinging or ssh-ing from the PowerBook, but never got any response. The system.log for last night ends with the following lines:

    Apr 11 17:27:23 DualG4 kernel: USBF: 200722. 21 AppleUSBOHCI[0x229d800] Watchdog detected dead controller (hcca #: 32212, hc #: 32667)
    Apr 11 17:29:01 DualG4 Toast This: pluginFactory
    Apr 11 17:29:01 DualG4 Toast This: queryInterface
    Apr 11 17:29:01 DualG4 Toast This: examineContext: null
    Apr 11 17:35:40 DualG4 Toast This: examineContext: null
    Apr 11 23:24:04 DualG4 Toast This: examineContext: null
    Apr 12 03:15:00 DualG4 CRON[1268]: (root) CMD (periodic daily)

    (23:24 is about when I checked the computer for the last time before going to bed.)

    Doesn’t look abnormal to me…

  2. ssp says:

    (a) Your network setup doesn’t sound extraordinary or complicated at all. You seem to be using features that are all ‘advertised’ for the different hard/software you bought.

    (a2) If you’re up for a challenge, solve the following network problem: I have a Cisco VPN connection coming in over a second wireless PC card and want to share it with a Windows computer via the built-in Airport card. The Windows computer will see the network I offer and be assigned an IP address by my DHCP server. Still, neither of the computers can ping the other one. Why is this the case and how can it be fixed?

    (b) Mail may trash your e-mails but I doubt it is powerful enough to freeze your machine.

    (c) To which extent did the machine freeze? Could you still ping it? ssh into it? What do the console/system.log files say?

  3. ed says:

    get off 28.8k, your problems will more than likely disapper. With 20 different email accounts in mail all trying to get out of a 28.8…I wouldn’t let you get anything either.

    Apple has done a superb job of bringing the rock solid FreeBSD to everyday users like you. Have you ever tried to do all these things in slackware or anything other than a gui, then you know how difficult it is to get something to work. My dual G5 and dual G4 (quicksilver) have been up since I got em with the exception of updates with no hiccups. Now my powerbook 1.25 is a different story with their flaky batteries…I have to switch them every once in a while.

    So to answer your question, glitches will disapper when users (and developers) stop creating them.

  4. Pierre Igot says:

    I wish high speed was an option… but it isn’t — unless you have a couple thousand dollars to spare.

    I find your attitude (blaming other users for allegedly creating problems) rather counterproductive, I must add. It’s all about enhancing the user experience for everyone, not just the happy few. And no one is forcing you to read about other people’s problems.

  5. Pierre Igot says:

    For what it’s worth, since disabling Déjà Vu, I haven’t experienced the freeze… 4 days of uptime and counting.

  6. Pierre Igot says:

    If Mail featured the ability to specify a different time interval for e-mail checking for each account, believe me, I would gladly spread the e-mail checking over time. (Some of these accounts only need to be checked a couple of times a day.) But in spite of my repeated attempts to get the message across to Apple, Mail still only has one setting, forcing you to check all accounts at the same time.

    As for the freezes… yes, fingers crossed. 5 days of uptime now :).

  7. ssp says:

    Pierre, I suppose I am with Apple here… this is an option that very few people will use and that only makes the already complex e-mail setup more complex for everyone else.

    My favourite strategy (as I described on my blog a while ago) remains to simply forward all my e-mail accounts to a single one that I actually check with Mail. While I mainly do this do reduce the connectivity issues caused by firewalls that may ‘protect’ me from accessing my data, it has the added benefit of having the least overhead. If you are dealing with large amounts of mail, having two separate (important, few spammed and public, highly spammed) accounts that you check may give you the best balance of easy access to your important e-mails and little overhead.

    And if you really want different checking intervals, why don’t you try an AppleScript that gets executed periodically and checks mails in the desired accounts?

  8. ssp says:

    While I’d agree with ‘ed’ that checking 20 mail accounts simultaneously on a slow link may not be wise (forwarding them all to one account and checking that one may give you less overhead an thus much better performance) – I also totally agree with Pierre that this should never make a system freeze or even crash. There may be bad performance but no crashes.

    And it looks like there might have been another culprit… fingers crossed.

Leave a Reply

Comments are closed.