Re: lots of Zombie Processes on Vaio PCG-Z600LEK
In linux.debian.laptop, you wrote:
>Hello,
>
> I have a problem with zombie processes on my laptop(debian woody).
>Applications that seem to spawn child processes don't seem to get rid
>off them when finnished, for example XMMS creates a process for every
>track, but these never get killed, and just pile up until the system
>process limit is hit. When the offending application is finnished the
>zombie child processes are killed as well (by init?), but not before
>then.
It's been a while. Have you solved the problem yet?
> The problem only starts to occur a few hours after boot. Other friends
>said it might be an issue with my kernel(custom 2.4.18), but I rebuilt
>it from scratch (new kernel soruce) and that didn't help.
>
> I vaguely remember (I realize that's unhelpfull) this starting to occur
>after I fiddled around with pcmcia-cs and wavelan to try and get my
>wavelan card to work, I used versions of those packages that I compiled
>my self, and recompiled the kernel as well to get it to work. Could this
>be related, if so how might I go about trying to tell?
I first encountered this problem about 1.5 years ago, on both my
laptop and desktop. I downgraded both to testing from unstable -
mainly to get libc back to the older version. I also removed a few
packages, and my problems mysteriously dissapeared. The problem
initially persisted across a large range of kernels - about 2.4.10 to
18 or so, IIRC.
>I would be greatfull if anyone has any suggestsions/ ideas of where to
>start looking?
For the past few hours, I have been hacking on noflushd, trying to get
it to talk to ext3 partitions (with little success so far - it
*should* be working, in theory....)
Then I gave up for the night, and noticed that I have all these zombies everywhere again. What have I not been using for the past 6 months, since the problem disappeared? noflushd.
> 283 ? 00:00:11 xfs
> 285 ? 00:00:10 noflushd
> 294 ? 00:00:00 atd
And then I notice you're running it too. Suspcicious.
I wonder if the killing of a kernel thread (kupdated) - a process I
always thought of as being rather clumsy, could be upsetting threaded
applications. Just a random stab in the dark, but you may want to
concider turning it off, rebooting, and seeing whether the problem is
gone.
--
TimC -- http://astronomy.swin.edu.au/staff/tconnors/
Calm down, it's *only* ones and zeroes.
Reply to: