From: Nick Murtagh (murtaghn at domain tcd.ie)
Date: Thu 11 Apr 2002 - 15:02:18 IST
On Thursday 11 April 2002 15:50, you wrote:
> On Thu, Apr 11, 2002 at 02:27:32PM +0100, Nick Murtagh wrote:
> > That's what it does. It the server knows about the A record, it might
> > return that as well. It's not the only aspect of DNS that's dumb. I want
> > a way to invalidate the caches for records that get updated :)
>
> Why that's easy - you just gradually reduce the TTL of the record, so that
> N time units before the planned change, the TTL is reduced to < N time
> units.
>
> Oh - sorry - you want something to cope with improperly planned changes ?
> :-)
Of course :)
So you get around invalid caching by effectively disabling the caching, and
any performance benefits that may bring (albeit temporarily). That's not what
I'd call a real solution.
And yes, there are broken caches out there. Also, aren't there minimum legal
TTLs in the specs?
This archive was generated by hypermail 2.1.6 : Thu 06 Feb 2003 - 13:16:00 GMT