What the *hell* is wrong with my Leopard system? What is “callback_client” and why does it die?

OK, I’m really mystified. I cannot, absolutely cannot, get this Leopard (10.5.2, clean install and then reimport of all the old stuff) installation on my Powerbook 1.67GHz with 1GB RAM (see, this means nobody can complain I didn’t tell them what my system was) to stay up for any length of time.

I mean, the longest I’ve managed is not quite seven days. That’s really not very long, compared to what Tiger used to manage. That’s rubbish. That’s awful. And no, I’m not running any weird extensions or APE or that lot. It’s Leopard and some applications.

And it almost always happens on waking from sleep – and then on waking automatically from sleep. Particularly in the mornings, when it is set to wake from sleep at 0655. Then I come down, hoping to find it’s been sucking up email and web feeds and opening pages and so on, and what do I find? Frozen machine – no response from the GUI; only way back is to power off, power on.

Any clues in the Console, you ask? Yes, I ask that too. Here’s what we find. The machine wakes up at 06:55:00 (we assume); the first kernel note on the console is at 06:55:07 saying “System Wake”. Jolly good. One second later:

20/02/2008 06:55:08 Mail[10442] *** Error while selecting [Google Mail]/All Mail: (null)

Then a few things about Appleshare shutting down and starting up, and then it gets the Airport up (on interface en1). Then two seconds later:
20/02/2008 06:55:32 mDNSResponder[25] Note: Frequent transitions for interface en1 (192.168.0.4); network traffic reduction measures in effect
20/02/2008 06:55:33 mDNSResponder[25] Note: Frequent transitions for interface en1 (FE80:0000:0000:0000:0211:24FF:FE28:18E9); network traffic reduction measures in effect

Ur, OK, so maybe there’s something about the Airport? But here’s the next thing, which is the killer, 20 seconds later.

20/02/2008 06:55:53 fseventsd[30] callback_client: ERROR: d2f_callback_rpc() => (ipc/send) timed out (268435460) for pid 235
20/02/2008 06:56:03 fseventsd[30] callback_client: ERROR: d2f_callback_rpc() => (ipc/send) timed out (268435460) for pid 235
20/02/2008 06:56:03 fseventsd[30] client: 0x80f800 : USER DROPPED EVENTS!

Which just goes on and on and on until I come along and kill the thing. And the really strange thing? I can’t find a process id 235. On restarting the machine, there’s a 234 (which is Apple’s Activity Monitor, the second process to start on my login) and then a 237 (pmtool, the command-line program that feeds Activity Monitor).

I can’t find anything from Googling that explains what callback_client might be, nor anything that looks even slightly like this. (There’s a discussion about iDVD on Apple’s support boards, but I don’t use it.) It’s totally perplexing. Come on, can anyone enlighten me even a little?

8 Comments

  1. If possible, you might try a ethernet connection overnight and see what it does in the morning. It sounds like you are getting major interference over your WiFi network that is causing the Airport to drop out repeatedly. Maybe that is causing the system to get sick. I would report this to Apple if that is the case.

    The fseventsd is part of Time Machine. It is the daemon that keeps Time Machine updated on what files have changed since your last backup. If you want a technical explanation there is a good one on Ars:
    http://arstechnica.com/reviews/os/mac-os-x-10-5.ars/7

    Hope this helps.

    James

  2. @artMonster: it’s not – that’s a graphics driver problem, and there’s no graphics driver output in my console.

    @James: yes, that’s feasible. However, I don’t use Time Machine – never have. It’s not turned on and hasn’t been. Which makes it very odd. I’ll see if any other apps are using fseventsd since that might be something to do with it.

  3. My guess, and this is only speculation, is that some process is registering to receive fs events then dying (which is why you can’t find a pid 235) without un-registering, then being resurrected by some process (launchd?) then repeating the same issue. Then when an fs event does happen (which happens whenever any file is changed on the system) it repeatedly tries to contact the dead process. It then gets hung up trying over and over again to start this errant process and notify it of file system events.

    If you can figure out what went in that pid 235 slot and prevent it from running in the first place, you’ll probably find your culprit. Since it registers for fs events, high on my list would be something to do with Spotlight, Time Machine, or a background task related to the Finder itself.

  4. You’re not using growl by any chance? I had to kill some apps associated with growl (growl mail in particular) that produced the symptoms you see.

  5. @Paul – weirdly, though I didn’t install it explicitly (and had it set aside in /Documents, not /Library/PreferencePanes, so it wouldn’t load), it has been running.

    I’ve uninstalled it. Let’s see what happens…

  6. Did you ever solve this? I am having the same problem.

  7. Charles

    Tuesday 5 August 2008 at 10:10 pm

    @John: well, eventually I got the machine fixed – it had a fault motherboard. So if you’re using a machine which has had reported problems, that’s a possible cause: hardware, not software. Unfortunately.

Comments are closed.