Bug#161483: libc6: syslog "init: Trying to re-exec init" causing later problems?
Package: libc6
Version: 2.2.5-14.3
Severity: important
I have had a problem for a while with pdnsd (see bug 101159), where
after a bit of uptime, it starts giving zombies whenever it is asked
to resolve a new name. Restarting the daemon killed the zombies, but
then they would reappear straight away, as soon as something new was
resolved. I found this was affecting other programs that used threads
as well, such as galeon/mozilla.
But I just noticed something new. I apt-get upgraded (tracking
unstable/testing) and a new libc6 was installed. I noticed a line in
my syslog that always happens when I upgrade libc6:
"Sep 19 13:26:32 scuzzie init: Trying to re-exec init"
And ever since about then, zombies haven't been getting reaped.
I didn't think init was meant to take part in reaping when the program
is still alive (I thought init only inherited the child once it's
parent died), but could this be the cause of my woes?
'ps axf' clearly shows the zombies parent is still the parent pdsnd
process, so I have no idea. I will try to keep this box alive, so you
can get me to experiment, as long as a developer gets back to me
reasonably quickly!
If it turns out that one shouldn't upgrade libc6 on a box that is to
remain up after the upgrade, perhaps having a big warning notice not
to upgrade libc6 until you are ready to power down? Even for
testing/unstable (I know I should know better than to run this on a
box that is meant to be stable)!
-- System Information:
Debian Release: testing/unstable
Architecture: i386
Kernel: Linux scuzzie 2.4.19+openmosix #1 Wed Sep 11 15:31:26 EST 2002
i686
Locale: LANG=C, LC_CTYPE=C
Versions of packages libc6 depends on:
ii libdb1-compat 2.1.3-4 The Berkeley database
routines [gl
-- no debconf information
--
TimC -- http://astronomy.swin.edu.au/staff/tconnors/
Confucius say: He who play in root, eventually kill tree.
Reply to: