[openstack-dev] [Fuel][QA] What is the preferred way to bootstrap a baremetal node with Fuel on product CI?
Dennis Dmitriev
ddmitriev at mirantis.com
Thu Feb 11 12:45:43 UTC 2016
Thanks to all for answers!
We will leave Fuel master node on a VM for our testing until some
specific cases will require it on a baremetal.
Ironic looks like a good tool for PXE provisioning and manage other
baremetal slaves via IPMI, we will investigate how it could be used in
our testing tools later.
On 02/10/2016 12:43 PM, Vladimir Kuklin wrote:
> Folks
>
> I think the easiest and the best option here is to boot iPXE or
> pxelinux with NFS and put master node image onto an NFS mount. This
> one should work seamlessly.
>
> On Wed, Feb 10, 2016 at 1:36 AM, Andrew Woodward
> <awoodward at mirantis.com <mailto:awoodward at mirantis.com>> wrote:
>
> Unless we hope to gain some insight and specific testing by
> installing the ISO on a bare-metal node (like UEFI), I'd propose
> that we stop testing things that are well tested elsewhere (a
> given ISO produces a working fuel master) and just focus on what
> we want to test in this environment.
>
> Along this line, we cold
>
> a) keep fuel masternode as a VM that is set up with access to the
> networks with the BM nodes. We have a good set of tools to build
> the master node in a VM already we can just re-use time
>
> b) use cobbler to control PXE based ISO boot/install, then either
> create new profiles in cobbler for various fuel nodes with
> different ISO or replace the single download link. (Make sure you
> transfer the image over HTTP as TFTP will be slow for such size.
> We have some tools and knowledge around using cobbler as this is
> effectively what fuel does its self.
>
> c) fuel on fuel, as an extension of b, we can just use cobbler on
> an existing fuel node to provision another fuel node, either from
> ISO or even it's own repo's (we just need to send a kickstart)
>
> d) you can find servers with good BMC or DRAC that we can issue
> remote mount commands to the virtual cd-rom
>
> e) consider using live-cd approach (long implmentation). I've been
> asked about supporting this in product where we start an
> environment with live-cd, the master node may make it's own home
> and then it can be moved off the live-cd when it's ready
>
>
> On Tue, Feb 9, 2016 at 10:25 AM Pavlo Shchelokovskyy
> <pshchelokovskyy at mirantis.com
> <mailto:pshchelokovskyy at mirantis.com>> wrote:
>
> Hi,
>
> Ironic also supports running it as standalone service, w/o
> Keystone/Glance/Neutron/Nova etc integration, deploying images
> from HTTP links. Could that be an option too?
>
> BTW, there is already an official project under OpenStack
> Baremetal program called Bifrost [0] that, quoting, "automates
> the task of deploying a base image onto a set of known
> hardware using Ironic" by installing and configuring Ironic in
> standalone mode.
>
> [0] https://github.com/openstack/bifrost
>
> Cheers,
>
>
> On Tue, Feb 9, 2016 at 6:46 PM Dennis Dmitriev
> <ddmitriev at mirantis.com <mailto:ddmitriev at mirantis.com>> wrote:
>
> Hi all!
>
> To run system tests on CI on a daily basis using baremetal
> servers
> instead of VMs, Fuel admin node also should be bootstrapped.
>
> There is no a simple way to mount an ISO with Fuel as a
> CDROM or USB
> device to a baremetal server, so we choose the
> provisioning with PXE.
>
> It could be done in different ways:
>
> - Configure a libvirt bridge as dnsmasq/tftp server for
> admin/PXE network.
> Benefits: no additional services to be configured.
> Doubts: ISO should be mounted on the CI host (via
> fusefs?); a HTTP
> or NFS server for basic provisioning should be started in
> the admin/PXE
> network (on the CI host);
>
> - Start a VM that is connected to admin/PXE network, and
> configure
> dnsmasq/tftp there.
> Benefits: no additional configuration on the CI host
> should be
> performed
> Doubts: starting the PXE service becomes a little
> complicated
>
> - Use Ironic for manage baremetal nodes.
> Benefits: good support for different hardware,
> support for
> provisioning from ISO 'out of the box'.
> Doubts: support for Ironic cannot be implemented in
> short terms,
> and there should be performed additional investigations.
>
> My question is: what other benefits or doubts I missed
> for first two
> ways? Is there other ways to provision baremetal with Fuel
> that can be
> automated in short terms?
>
> Thanks for any suggestions!
>
>
> --
> Regards,
> Dennis Dmitriev
> QA Engineer,
> Mirantis Inc. http://www.mirantis.com
> e-mail/jabber: dis.xcom at gmail.com <mailto:dis.xcom at gmail.com>
>
>
> __________________________________________________________________________
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe:
> OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
> <http://OpenStack-dev-request@lists.openstack.org?subject:unsubscribe>
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>
> --
> Dr. Pavlo Shchelokovskyy
> Senior Software Engineer
> Mirantis Inc
> www.mirantis.com
> __________________________________________________________________________
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe:
> OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
> <http://OpenStack-dev-request@lists.openstack.org?subject:unsubscribe>
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>
> --
> --
> Andrew Woodward
> Mirantis
> Fuel Community Ambassador
> Ceph Community
>
> __________________________________________________________________________
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe:
> OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
> <http://OpenStack-dev-request@lists.openstack.org?subject:unsubscribe>
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>
>
>
>
> --
> Yours Faithfully,
> Vladimir Kuklin,
> Fuel Library Tech Lead,
> Mirantis, Inc.
> +7 (495) 640-49-04
> +7 (926) 702-39-68
> Skype kuklinvv
> 35bk3, Vorontsovskaya Str.
> Moscow, Russia,
> www.mirantis.com <http://www.mirantis.ru/>
> www.mirantis.ru <http://www.mirantis.ru/>
> vkuklin at mirantis.com <mailto:vkuklin at mirantis.com>
>
>
> __________________________________________________________________________
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
--
Regards,
Dennis Dmitriev
QA Engineer,
Mirantis Inc. http://www.mirantis.com
e-mail/jabber: dis.xcom at gmail.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20160211/a4d88bff/attachment.html>
More information about the OpenStack-dev
mailing list