This is the mail archive of the
libc-alpha@sourceware.cygnus.com
mailing list for the glibc project.
Re: gethostbyname2_r buggy?
- To: Horst von Brand <vonbrand@sleipnir.valparaiso.cl>
- Subject: Re: gethostbyname2_r buggy?
- From: Franz Sirl <Franz.Sirl-kernel@lauterbach.com>
- Date: Mon, 06 Sep 1999 10:43:09 +0200
- Cc: Pawel Krawczyk <kravietz@ceti.pl>,libc-alpha@sourceware.cygnus.com
- References: <Message from Pawel Krawczyk <kravietz@ceti.pl><19990905173211.C30974@ceti.pl>
At 22:21 05.09.99 , Horst von Brand wrote:
>Pawel Krawczyk <kravietz@ceti.pl> said:
> > Recently I've been tracking the reason why nscd has failed on some
> > type of hosts and discovered that it was caused by failure of
> > gethostbyname2_r() (called from addhstbyname() in nscd/hstcache.c).
> > This function fails returning -1 when resolved host is a recursive
> > CNAME, e.g:
> >
> > [root@ns1 named]# host test1
> > test1.ceti.com.pl is a nickname for test2.ceti.com.pl
> > test2.ceti.com.pl is a nickname for test3.ceti.com.pl
> > test3.ceti.com.pl has address 195.116.211.2
>
>I believe this is illegal...
You are probably thinking about the restrictions on what can point to a
CNAME, eg. the usual suspects MX and NS are not allowed to point to a
CNAME. However the CNAME chain Pawel posted is perfectly legal.
The simple rule is: Only a CNAME can point to another CNAME.
Well, unless this has changed recently :-).
Franz.