From mboxrd@z Thu Jan  1 00:00:00 1970
From: Ian Jackson <Ian.Jackson@eu.citrix.com>
Subject: Re: [OSSTEST Nested PATCH v8 3/7] Edit some APIs in
 TestSupport.pm for nested test
Date: Tue, 21 Apr 2015 15:30:09 +0100
Message-ID: <21814.24305.129814.960644@mariner.uk.xensource.com>
References: <1428959990-24199-1-git-send-email-longtaox.pang@intel.com>
	<1428959990-24199-4-git-send-email-longtaox.pang@intel.com>
	<1429611542.4743.22.camel@citrix.com>
	<21814.17314.479011.970684@mariner.uk.xensource.com>
	<1429620780.4743.67.camel@citrix.com>
	<21814.20580.723370.301076@mariner.uk.xensource.com>
	<1429623708.4743.97.camel@citrix.com>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Return-path: <xen-devel-bounces@lists.xen.org>
In-Reply-To: <1429623708.4743.97.camel@citrix.com>
List-Unsubscribe: <http://lists.xen.org/cgi-bin/mailman/options/xen-devel>,
	<mailto:xen-devel-request@lists.xen.org?subject=unsubscribe>
List-Post: <mailto:xen-devel@lists.xen.org>
List-Help: <mailto:xen-devel-request@lists.xen.org?subject=help>
List-Subscribe: <http://lists.xen.org/cgi-bin/mailman/listinfo/xen-devel>,
	<mailto:xen-devel-request@lists.xen.org?subject=subscribe>
Sender: xen-devel-bounces@lists.xen.org
Errors-To: xen-devel-bounces@lists.xen.org
To: Ian Campbell <ian.campbell@citrix.com>
Cc: robert.hu@intel.com, "longtao.pang" <longtaox.pang@intel.com>, wei.liu2@citrix.com, xen-devel@lists.xen.org
List-Id: xen-devel@lists.xenproject.org

Ian Campbell writes ("Re: [OSSTEST Nested PATCH v8 3/7] Edit some APIs in TestSupport.pm for nested test"):
> On Tue, 2015-04-21 at 14:28 +0100, Ian Jackson wrote:
> > I'm only objecting to the involvement of the host property machinery.
> 
> I'm afraid I'm not 100% sure I understand, but do you just mean that
> longtao should assign to $ho->{Ip} directly within selecthost instead of
> using $setprop to set the IpAddr hostprop and therefore indirectly
> ->{Ip}?

Yes.

> IOW the existing code in selecthost which looks up the IpAddr host prop
> into $ho->{IpStatic} and then into $ho->{Ip} should prefer the runvar to
> the host db if it is set?

Yes.

> BTW, what is the difference between ->{Ip} and ->{IpStatic} and should
> the runvar be reflected in both or just in ->{Ip}?

AFAICT the only difference is that only IpStatic is used for expanding
`ipaddr' and `ipaddrhex' in the patterns for dhcp pxe configurations.

There is an incipient problem here: our existing setup does not
require that osstest has complete control of the pxe config file for
every host on the network.  And indeed the Citrix (Cambridge) instance
lacks such control.

Instead there are some symlinks etc.  The mg-hosts mkpxedir subcommand
arranges to delegate the DHCP from root to the osstest user (and
group), by making an appropriate subdirectory with the right
permissions, and a suitable symlink.

It's not quite clear to me how this should work for nested HVM hosts.
I don't think making 2^24 symlinks "pxelinux.cfg/5a:36:0e:??:??:??"
is a practical proposition.  Perhaps nested HVM hosts should choose
their MAC addresses from a smaller set, but then there might have to
be some inter-flight coordination.

Alternatively we could provide a service for making the right setup
(well, really, just a symlink) on demand.  In practice the number of
required symlinks is more likely to be around 2^8 * <total number of
different guests in a flight>, which will be thousands rather than
millions.

Ian.