attack of the zombies
Alexander at Leidinger.net
Mon Nov 13 04:54:15 PST 2006
Quoting Bakul Shah <bakul at bitblocks.com> (from Sun, 12 Nov 2006
> About a week or so ago I updated -current and now linux
> binaries don't seem to collect all zombie processes.
> Eventually the maxproc limit is reached and further forks
> fail so that you can't even do ps (of course, dealing
> sensibly with such errors is another problem with most
> programs but that is a separate discussion). Work around is
> killing linux programs to allow init to kill the zombies.
You are the first one reporting this problem. I didn't noticed
something like this in my regression test runs with the linux test
project testcases and all other active developers in this area didn't
noticed something like this too (so far). Do you have some small
testcases (e.g. with programs in linux_base) or does this apply to a
specific workload (like the one below) only?
> This happens with skype, firefox and opera and may be more.
> I reinstalled linux_base-fc-4_9 and all ports depending on it
> -- all updated yesterday. The problem persists even with
To make sure there is no "garbage" somewhere:
- remove all linux ports
- remove /compat/linux/* (rm -rf)
- install what you need (only from ports)
> yesterday's -current. This problem showed up sometime
> between Oct 6 and Nov 6. One significant change I see during
> this time is the treatment of KSE. But presence or absence
> of nooption KSE does not seem to affect this problem. BTW,
> linux emulation is loaded as a module.
We are talking about i386, right? Please provide the output of "sysctl
compat.linux" (osversion should be set to 2.4.2).
> Also note that the old problem of linux-* programs gobbling
> up lots of memory is still present. For example, FreeBSD
> opera uses 96MB while Linux opera on FreeBSD needs 236MB + 48
> zombies to displaying exact same 24 pages (same session file
> and *just* after starting!).
> Is this a known problem? Am I doing something wrong?
This is not a known problem (at least not for 2.4.2 compatibility,
which is the default in -current). For known problems have a look at
"The only real way to look younger is not to be born so soon."
-- Charles Schulz, "Things I've Had to Learn Over and
Over and Over"
http://www.Leidinger.net Alexander @ Leidinger.net: PGP ID = B0063FE7
http://www.FreeBSD.org netchild @ FreeBSD.org : PGP ID = 72077137
More information about the freebsd-emulation