[oe] [meta-networking][PATCH 2/2] dnsmasq: disable the service by default

[Joe MacDonald]

[Re: [oe] [meta-networking][PATCH 2/2] dnsmasq: disable the service by default] On 17.02.28 (Tue 01:12) Huang, Jie (Jackie) wrote:

> It can be overridden to enable the service in the conf file of the
> distro with witch only dnsmasq is used in the images.
> 
> We can also disable the service in the distro conf file instead of in
> the recipe if it makes more sense.

I've been thinking about this and I think it would make more sense for
dnsmasq to have a PROVIDE/CONFLICT option that would eliminate the risk,
though I guess I can see a scenario where you might want to use bind for
DNS services and use dnsmasq to provide just DHCP or something.  It
would be a weird setup, though.

Seems to me that your patch is to cover a case where you want to have
both on the system, though, so can you explain why we wouldn't make
dnsmasq incompatible with bind?

In that scenario, why not check for a PACKAGECONFIG option or for the
presence of bind and only disable dnsmasq if you find bind as well?

-J.

> 
> Thanks,
> Jackie
> 
> From: Bruce Ashfield [mailto:bruce.ashfield at gmail.com]
> Sent: Monday, February 27, 2017 10:39 PM
> To: Huang, Jie (Jackie)
> Cc: openembedded-devel
> Subject: Re: [oe] [meta-networking][PATCH 2/2] dnsmasq: disable the service by default
> 
> 
> 
> On Mon, Feb 27, 2017 at 3:09 AM, <jackie.huang at windriver.com<mailto:jackie.huang at windriver.com>> wrote:
> From: Jackie Huang <jackie.huang at windriver.com<mailto:jackie.huang at windriver.com>>
> 
> The service will fail to start/stop when dnsmasq and
> bind (provides named) are both installed since they
> are using the same port 53:
> 
> | dnsmasq: failed to create listening socket for port 53: Address already in use
> 
> So disable the service by default, if the user really
> want to use two dns servives, the config needs to be
> changed to use different interfaces or ports.
> 
> But what if the user is only using dnsmasq ? (which is the case in the images I use).
> This would be a behaviour change that they add dnsmasq, but don't actually get the
> service started.
> IMHO if someone is installing two packages that provide the same service, they
> are more aware of what is going on and can disable one (or the other) .. versus
> someone that adds a single package that provides a service .. and then find out that
> it doesn't actually start.
> Bruce
> 
> 
> As reference, other distro like centos/fedora also
> disables the services by default on installation.
> 
> Signed-off-by: Jackie Huang <jackie.huang at windriver.com<mailto:jackie.huang at windriver.com>>
> ---
>  meta-networking/recipes-support/dnsmasq/dnsmasq.inc | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/meta-networking/recipes-support/dnsmasq/dnsmasq.inc b/meta-networking/recipes-support/dnsmasq/dnsmasq.inc
> index 424a06b..9c2276f 100644
> --- a/meta-networking/recipes-support/dnsmasq/dnsmasq.inc
> +++ b/meta-networking/recipes-support/dnsmasq/dnsmasq.inc
> @@ -81,3 +81,4 @@ RPROVIDES_${PN} += "${PN}-systemd"
>  RREPLACES_${PN} += "${PN}-systemd"
>  RCONFLICTS_${PN} += "${PN}-systemd"
>  SYSTEMD_SERVICE_${PN} = "dnsmasq.service"
> +SYSTEMD_AUTO_ENABLE ?= "disable"
> --
> 2.8.3
> 
> --
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel at lists.openembedded.org<mailto:Openembedded-devel at lists.openembedded.org>
> http://lists.openembedded.org/mailman/listinfo/openembedded-devel
> 
> 
> 
> --
> "Thou shalt not follow the NULL pointer, for chaos and madness await thee at its end"
-- 
-Joe MacDonald.
:wq
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 473 bytes
Desc: Digital signature
URL: <http://lists.openembedded.org/pipermail/openembedded-devel/attachments/20170227/ad50eed5/attachment-0002.sig>