[Scottish] Debian: leafnode 1.9.27 keeps re-fetching all newsgroups from remote news-server(s)

David Marsh's list-reading hat scottish at mailman.lug.org.uk
Fri Jan 10 17:25:02 2003

[cc'ed to broonie, summary for his benefit: I run leafnode 1.9.27 under
Debian 'testing' (sarge) and it was seeming to keep re-fetching the
entire newsgroup list on each occasion it was run. Apologies if this is
a known issue or if it's user error somewhere at my end..]

On Fri, 10 Jan 2003 09:30:22 +0000
Colin McKinnon <colin@wew.co.uk> wrote:

> >>Use the source, Luke?
> >>    
> >>
> >Sorry, these days I'm just a user, really..
> >I wouldn't know where to start (don't think I even have the source).
> >
> By way of amends (and because its another _really_ quiet day here.....
> (from fetchnews.c)
> if ( active && !forceactive && ( stat( s, &st ) == 0 ) ) {
>     ----it will get NEWGROUPS
> else
>     ----it will get all groups
> It uses the mtime on 'active.read' as the time of the last full update
> - you might want to check the permissions and timestamp on this (on my
> box its in /var/spool/news, and is 0 bytes). 


I think that's it. I don't have an 'active.read' file, for some reason.

Touching and perming one into existence means that everything works as
it should on the next run.

If I then do a 'fetchnews -f' the active.read file is deleted while this
is taking place, but *doesn't* seem to be recreated afterwards, leading
to the re-fetching of the entire newsgroup list on the next run of
fetchnews. If I then recreate it by hand, all seems well again..

> If it's older than 
> timeout_active then 'forceactive' is set before the above bit of code.
> It is this file which is checked by the 'stat' statement in the above
> so if it is not found then it will get all groups too (stat() will
> return 0). 

Yep, that seems to be it.

David Marsh, <email in header is valid>  | http://web.viewport.co.uk/