bind lockups

Frazier Scott scottf at oit.edu
Thu Apr 26 18:31:02 UTC 2001


Thanks a million.  Knowing where the h_errno symbol resides enabled me to 
hunt down a faulty library and fix the issue.  Bind 9.1.1 compiled and 
installed and so far not a single lockup.  I'm sorry to say that it looks 
like it might have been a PEBKAC error.


At 06:20 PM 4/25/2001 +0100, Jim Reid wrote:
> >>>>> "Frazier" == Frazier Scott <scottf at oit.edu> writes:
>
>     Frazier> Can anyone help me troubleshoot my dns server?  I'm
>     Frazier> running Bind 9.1.0 and can't upgrade to 9.1.1 because of
>     Frazier> compile-time errors.
>
>Well report those compilation errors to bind9-bugs. Nobody should be
>running 9.1.0. It has a number of threading bugs that cause it to lock
>up. These are fixed in 9.1.1, though some platforms with broken
>threads implementations will need to run a single-threaded name
>server. This is done with a --disable-threads argument to configure
>before compiling.
>
>     Frazier> cc -pthread -std -g -o dig dig.o dighost.o 
> ../../lib/dns/libdns.a ../../lib/isc/libisc.a -lpthread
>     Frazier> ld: Unresolved: __h_errno make[2]: *** [dig] Error 1
>
>__h_errno is a standard symbol in a Unix resolver library. It's used
>for holding an error code whenever a resolver routine returns an
>error, just like errno and a Unix system call. Try linking the
>programs with libresolv.a (ie a -lresolv argument to cc/ld) or
>whatever it's called on your system. It might be an idea to rebuild
>BIND9 from scratch in case there's junk from an earlier, broken
>compilation attempt lying around. It could well be that your problems
>are related to the compilation problems you referred to earlier. I'd
>get them fixed first obviously.



More information about the bind-users mailing list