CNAME TTL (fwd)

[tmaestas at dnsconsultants.com]

	*GROAN* - I was the one who orignally reported this
	problem to Microsoft.  The hotfix they provided my
	company with did not truly, in my opinion, fix
	the problem.  It simply brought them into compliance
	with RFC2181 which states that a TTL is the maximum
	time to live, and that implementations can cache at a 
	value lower than the TTL if desired.  Now they cache both
	CNAME and A with the lower of the two TTLs.  They
	still disregard the actual TTL set by DNS administrators
	if the CNAME and A records have disparate TTLs.  In speaking
	with the authors of RFC2181, they agree that the only reason
	to cache CNAMEs and A's with the same TTL is lazy implementation.
	In any case, Microsoft has misquoted RFC2181 in this article by
	saying that the rfc defines that CNAMEs and their corresponding
	A's TTLs should be set at the lower of the two.  Nowhere in
	RFC 2181 is this stated.  CNAMES and their target A's do not
	form a RRset, so they do not *need* to have the same TTL.

-Tim


On Thu, 22 Mar 2001, Kerry M. Liles wrote:

> If this is really biting you on W2K, check out this KB article:
> 
> Q276324  and call MS to get the hotfix...
> 
> for the cut and pasters, it is at:
> 
> http://support.microsoft.com/support/kb/articles/Q276/3/24.ASP?LN=EN-US&SD=g
> n&FR=0&qry=dns%20cname&rnk=16&src=DHCS_MSPSS_gn_SRCH&SPR=WIN2000
> 
> 
> HTH
> 
> 
> 
> <tmaestas at dnsconsultants.com> wrote in message
> news:99dgnq$h6k at pub3.rc.vix.com...
> >
> >
> > Sorry, I had that backwards.  The bug is that the A record
> > is cached with the TTL of the CNAME.
> >
> > -Tim
> >
> >
> > ---------- Forwarded message ----------
> > Date: Thu, 22 Mar 2001 08:53:10 -0800 (PST)
> > From: tmaestas at maestas.dnsconsultants.com
> > To: Vincent DUQUESNE <v.duquesne at silicomp.fr>
> > Cc: bind-users at isc.org
> > Subject: Re: CNAME TTL
> >
> >
> > CNAMES and A records have distinct, often differing TTLs.
> > HOWEVER, if you are running Windows 2000, you can forget
> > about this until SP2.  There is a bug in the Win2k client
> > caching dns resolver where a CNAME is cached with the
> > same TTL as it's corresponding A record.
> >
> > -Tim
> >
> >
> > On Thu, 22 Mar 2001, Vincent DUQUESNE wrote:
> >
> > >
> > > Hi all,
> > >
> > > I have a question for you, masters :
> > > Does a DNS server send, in its answer, the TTL of the CNAME record or
> the
> > > TTL of the A record corresponding ?
> > >
> > > thanks in advance
> > >
> > > Vincent
> > >
> > >
> >
> >
> >
> >
> 
>