Re: [leafnode-list] timeout connecting to (null) + Segmentation

[Ralf Wildenhues <wildenhu@xxxxxxxxxxxxxxx>]

Matthias Andree wrote:
> 
> That's sufficient to figure what happens, however I'm not quite sure how
> to fix this. I've currently added a workaround to fixxover to allow for
> a graceful shutdown, but that's not proper.
> 
> Here's what happens:
> 
> fetchnews -f flag (forceactive <- 1) lets fetchnews ignore the existing
> groupinfo file, and discard any relevant markers. When all servers are
> unreachable, no active is present at the time that fetchnews tries to
> update the overview files of the "interesting groups", assuming that no
> groups outside the interesting groups are changed.

Only things I could find out (not too much tested, sorry):

2 upstreams, A and B. group.from.A only served by A.
'fetchnews -f' and unavailable A
--> leafnode does not know group group.from.A any more ("no such group")
After another successful connect to A with 'fetchnews' this problem is
gone, pending outgoing messages to group.from.A are then sent.

Iff at least one upstream is available, active built (with the list of
that/those servers).  Groups only served by unreachable servers are
unknown until next successful connect.

HTH Ralf

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