[OpenSER-Devel] serial forking and nat
Bogdan-Andrei Iancu
bogdan at voice-system.ro
Tue Sep 11 18:06:06 CEST 2007
Hi Juha,
indeed, there is the problem with storing the serial forking information.
I will review your idea and comment it in the next days.
Thanks and regards,
Bogdan
Juha Heinanen wrote:
> Juha Heinanen writes:
>
> > each time when t_relay is called (perhaps with a new "serial fork"
> > parameter), it would serve the
> > next set of branches that have the highest q value and then mark those
> > as served in dset.c branches array.
>
> if that is too much to ask, then how about adding an int "status" field
> to branch struct and include it also as a new get_branch parameter. in
> addition, there would be a new dset.h function that can be used to set
> status value of a branch.
>
> initial value of status field for each branch would be zero. when
> t_relay is called, it always serves those branches whose status field
> value is zero thus retaining backwards compatibility.
>
> next_contacts/next_branches would then be able to do their job by
> manipulating status field values. for example, when
> load_contacts/serialize_branches is called, it would set "status" of
> each branch to 1 (unused). each time next_contacts/next_branches is
> called, it would set each zero valued branch to -1 (used) and then
> select the highest unused q value branches and set their status field
> value to zero. if there is no unused branches left,
> next_contacts/next_branches would return -1.
>
> i don't know the internals of tm module, but i would think that adding a
> status==0 test before a branch is served, would not be a big deal.
>
> comments?
>
> -- juha
>
> _______________________________________________
> Devel mailing list
> Devel at openser.org
> http://openser.org/cgi-bin/mailman/listinfo/devel
>
>
More information about the Devel
mailing list