Re: [leafnode-list] Fetchnews: Falling back and regrouping.

["Michael O'Quinn" <michael@xxxxxxxxxxx>]

On Tue, 2 Apr 2002, Matthias Andree wrote:

> Joerg Dietrich schrieb am Montag, den 01. April 2002:
> 
> > On Sat, Mar 30, 2002 at 08:40:38PM +0100, Matthias Andree wrote:
> > > How do we fix the "downloaded again" then without introducing additional
> > > dependencies on Berkeley DB or gdbm? I cannot rewrite SERVERINFO for
> > > each article downloaded.
> > 
> > Write SERVERINFO after any group is fetched. Or am I missing
> > something?
> 
> I'm doing that in 1.9.21pre1, but it does not help if fetchnews aborts
> before the group fetch is complete and articles on the leafnode machine
> expire earlier than on the upstream, because we won't detect duplicates
> in that case and refetch. That's why I am currently clamping maxage to
> the groupexpire or expire if they are lower than the actual maxage.

IHMO refetching is the lesser of two evils.  

The only positive way to catch it every time (if there are short expire
times on the leafnode machine) is to use a history database or to update
SERVERNAME~ after each article.  We've already hashed those out, and
they've both been ruled out for 1.9.x.

Michael O'QUinn



-- 
leafnode-list@xxxxxxxxxxxxxxxxxxxxxxxxxxxx -- mailing list for leafnode
To unsubscribe, send mail with "unsubscribe" in the subject to the list