From mboxrd@z Thu Jan 1 00:00:00 1970 From: Boris Derzhavets Subject: Re: Ongoing issue with libvirt-xen driver for Openstack Nova Date: Thu, 16 Apr 2015 12:08:56 -0400 Message-ID: References: , <20150415154854.GF1707@perard.uk.xensource.com>, , <1429176429.25195.55.camel@citrix.com>, , <1429180538.25195.66.camel@citrix.com>, , <1429185944.25195.121.camel@citrix.com>, <20150416151249.GG1707@perard.uk.xensource.com> Mime-Version: 1.0 Content-Type: multipart/mixed; boundary="===============5226721908353060865==" Return-path: In-Reply-To: <20150416151249.GG1707@perard.uk.xensource.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: Anthony PERARD Cc: "xen-devel@lists.xen.org" List-Id: xen-devel@lists.xenproject.org --===============5226721908353060865== Content-Type: multipart/alternative; boundary="_07faf083-c6b8-4dd4-a1fc-9d49f241637a_" --_07faf083-c6b8-4dd4-a1fc-9d49f241637a_ Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable > Date: Thu=2C 16 Apr 2015 16:12:49 +0100 > From: anthony.perard@citrix.com > To: bderzhavets@hotmail.com > CC: xen-devel@lists.xen.org > Subject: Re: [Xen-devel] Ongoing issue with libvirt-xen driver for Openst= ack Nova >=20 > On Thu=2C Apr 16=2C 2015 at 01:05:44PM +0100=2C Ian Campbell wrote: > > On Thu=2C 2015-04-16 at 06:46 -0400=2C Boris Derzhavets wrote: > >=20 > > > > Was this just the tail of the log or the complete thing? > > >=20 > > >=20 > > > No it was in the middle . I am attaching xen-hotplug.log.gz > >=20 > > Thanks=2C I think this shows that as expected our hotplug scripts have = not > > been responsible for bringing vif1.0 up before renaming=2C so something > > else must be doing it. > >=20 > > If not some openstack component perhaps something like network-manager? > > Or some other udev rules perhaps? >=20 > It is likely to be something done in Neutron. You can always try to read > the logs of the different services (q-svc q-agt q-dhcp q-l3 q-meta) to > check what is happening to the network interface (vif1.0). There also cou= ld > be something wrong with Nova that will not set network properly in this > case. Or some other compatibility issue... >=20 > In anycase=2C the combo openstack-libvirt-xen works better with the defau= lt > set of services runned by devstack (as long as they don't switch yet to > Neutron). My core stuff is supposed to be RDO Kilo (RH) on Fedoras=2C like now it i= s RDO Juno. RH deprecated nova-networking a while ago =2C I guess in Havana release. Devstack&&stack.sh on Ubuntu 14.04 is just a test-bed for the most recent= nova.=20 =20 Thank you very much for your support and getting in touch with explanation= . I couldn't imagine =2C that slide show http://www.slideshare.net/xen_com_mg= r/openstack-xenfinal is for nova networking Boris >=20 > I will try to setup devstack with your local.conf to reproduce the issue. >=20 > --=20 > Anthony PERARD = --_07faf083-c6b8-4dd4-a1fc-9d49f241637a_ Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable

>=3B Date: Thu=2C= 16 Apr 2015 16:12:49 +0100
>=3B From: anthony.perard@citrix.com
&g= t=3B To: bderzhavets@hotmail.com
>=3B CC: xen-devel@lists.xen.org
&= gt=3B Subject: Re: [Xen-devel] Ongoing issue with libvirt-xen driver for Op= enstack Nova
>=3B
>=3B On Thu=2C Apr 16=2C 2015 at 01:05:44PM +0= 100=2C Ian Campbell wrote:
>=3B >=3B On Thu=2C 2015-04-16 at 06:46 -= 0400=2C Boris Derzhavets wrote:
>=3B >=3B
>=3B >=3B >=3B &= gt=3B Was this just the tail of the log or the complete thing?
>=3B &g= t=3B >=3B
>=3B >=3B >=3B
>=3B >=3B >=3B No it was in = the middle . I am attaching xen-hotplug.log.gz
>=3B >=3B
>=3B = >=3B Thanks=2C I think this shows that as expected our hotplug scripts ha= ve not
>=3B >=3B been responsible for bringing vif1.0 up before rena= ming=2C so something
>=3B >=3B else must be doing it.
>=3B >= =3B
>=3B >=3B If not some openstack component perhaps something lik= e network-manager?
>=3B >=3B Or some other udev rules perhaps?
&g= t=3B
>=3B It is likely to be something done in Neutron. You can alway= s try to read
>=3B the logs of the different services (q-svc q-agt q-d= hcp q-l3 q-meta) to
>=3B check what is happening to the network interf= ace (vif1.0). There also could
>=3B be something wrong with Nova that = will not set network properly in this
>=3B case. Or some other compati= bility issue...
>=3B
>=3B In anycase=2C the combo openstack-libv= irt-xen works better with the default
>=3B set of services runned by d= evstack (as long as they don't switch yet to
>=3B Neutron).

&nb= sp=3BMy core stuff =3B is supposed to be RDO Kilo (RH) on Fedoras=2C li= ke now it is RDO Juno.
=3BRH deprecated nova-networking a while ago= =2C I guess =3B in Havana release.
=3BDevstack&=3B&=3Bst= ack.sh on Ubuntu 14.04 =3B is just a test-bed for the most recent nova.=
=3B
Thank you =3B very much for your support and getting i= n touch with explanation.
I couldn't imagine =2C that slide show http://www.slideshare.net/xen_com_mgr/openstack-xenfinal
is for= nova networking

Boris

>=3B
>=3B I will try to setup = devstack with your local.conf to reproduce the issue.
>=3B
>=3B = --
>=3B Anthony PERARD

= --_07faf083-c6b8-4dd4-a1fc-9d49f241637a_-- --===============5226721908353060865== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline _______________________________________________ Xen-devel mailing list Xen-devel@lists.xen.org http://lists.xen.org/xen-devel --===============5226721908353060865==--