According to the findings in the post above I patched the socket.c this
morning as follows:
--- bind9-9.3.4.orig/lib/isc/unix/socket.c 2006-05-19
04:53:36.000000000 +0200
+++ bind9-9.3.4/lib/isc/unix/socket.c 2007-09-10 07:45:01.000000000 +0200
@@ -2190,7 +2190,10 @@
if (sock->connecting)
dispatch_connect(sock);
else
- dispatch_send(sock);
+ /* Prevent reuse of a still
pending socket. */
+ if (!sock->pending_send) {
+ dispatch_send(sock);
+ }
}
FD_CLR(i, &manager->write_fds);
}
added an entry
bind9 (1:9.3.4-3) unstable; urgency=high
* Fix assertion failure in socket.c:1663
-- Thomas Antepoth <[EMAIL PROTECTED]> Mon, 10 Jan 2007 06:09:03 -0700
to the debian/changelog file and rolled my own local package of
bind9-9.3.4-3 and related libraries by issuing dpkg-buildpackage in the
bind9 source directory.
Right now the server runs since about 08:00 am in the morning without
any coredumps or other issues. No problem up to now. I know that this is
more a dirty hack^W^Wpragmatic workaround but it seems to do the trick -
at least for me and my configuration.
t++
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
