On Sun, 11 Oct 2015 10:55:44 +0100
Dominic Benson <[email protected]> wrote:


> > I am against this change. I vote for a change for this. Nobody
> > expects firewall restart to stop traffic - ever.
> 
> I don’t think that this follows; with this change ‘reload’ does The
> Right Thing, and that is consistent with almost everything else. The
> restart action on a nameserver, webserver, database etc. would be
> expected to drop requests during the restart. Similarly a restart of
> a hardware firewall would drop traffic whereas a config commit
> wouldn’t.

That's very different. You think about one server situation. But as a
firewall/router that means all traffic routed will be halted and then
re-enabled, all natted connections are lost etc during restart. That is
NOT expected to happen during firewall restart.

There was a very big feature called shorewall-perl which was especially
for not to drop all traffic during restart - I'm very much against
crippling shorewall this way by introducing restart which by default
interrupts traffic.

My first suggestion was bad but I suggest config option for
shorewall.conf RESTART=restart|reload where reload is the default.

Especially naming proper behaviour LEGACY_RESTART is not a good idea.

-- 
Tuomo Soini <[email protected]>
Foobar Linux services
+358 40 5240030
Foobar Oy <http://foobar.fi/>

------------------------------------------------------------------------------
_______________________________________________
Shorewall-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/shorewall-users

Reply via email to