Re: [PATCH 1/1] nm-dns-dnsmasq: clear cache when dns server changes

[Dan Williams <dcbw redhat com>]

On Tue, 2016-11-29 at 23:20 +0100, Beniamino Galvani wrote:
> On Tue, Nov 29, 2016 at 10:10:21PM +0100, Thomas Haller wrote:
> > On Tue, 2016-11-29 at 20:23 +0100, Christian Hesse wrote:
> > > Thomas Haller <thaller redhat com> on Tue, 2016/11/29 17:51:
> > > > On Tue, 2016-11-29 at 15:03 +0100, Christian Hesse wrote:
> > > > > From: Christian Hesse <mail eworm de>
> > > > >
> > > > > A new DNS may have different data for us. So tell dnsmasq to
> > > > > clear
> > > > > the
> > > > > cache when servers change.
> > > >
> > > > we currently always do a ClearCache call after SetServerEx.
> > > >
> > > > How does this relate to your patch? Does --clear-on-reload make
> > > > the
> > > > ClearCache call unnecessary? Why is ClearCache not sufficient?
> > >
> > > Oops, missed that in my logs (and did not find the code)... You
> > > are
> > > right,
> > > cache is cleared.
> > >
> > > Either of both is just fine and it works as-is. So ignore my
> > > patch.
> >
> > Hi,
> >
> > clear-on-reload might be better:
> >
> >   - it saves the explicit call for ClearCache (less code, and
> > possibly
> >     less runtime-overhead).
> >   - depending on how --clear-on-reload is implemented, it might be
> >     race-free and clear the cache atomically when setting new
> > servers.
>
> I agree and wish I had noticed such option before :)
>
> > IMO, i would apply the patch, and possibly a follow-up to drop the
> > ClearCache call. But I don't know dnsmasq well enough to be sure
> > that
> > that is indeed correct...
>
> I tried the patch with the ClearCache call reverted and the behavior
> seems equivalent to master. How about branch bg/dnsmasq-clear-cache-
> option?

LGTM

Dan