|
On Tue, Dec 06, 2005 at 10:39:43AM +0200, Julian Anastasov wrote:
>
> Hello,
>
> On Tue, 6 Dec 2005, Horms wrote:
>
> > Yesterdays' patch + changing timeout to 10*HZ instead of 60*HZ,
> > which I am in favour of.
>
> OK. It works for short connections but not for long. One
> can require the persistence timeout to work after the last connection
> is expired, not from its start. It can be solved only if we can
> control this 10-second assignment. There can be 2 options: set to
> 10secs or preserve timeout. But this patch is a good start, we can
> modify it later.
I don't follow this. The persistance template will stay around until
the last controlled connection expires. If during that time
a new connection arrives, then the template will be used, and,
furthermore its timeout will be restored to the user-defined value.
The 10s only comes into play if the template expired but still
had controled connections, and those controlled connections
subsequently expired without any new connections for the template
arriving.
--
Horms
|
