10:32 p.m.
HI,
Some time ago Michele Baldessari wrote about deploying OpenStack
TripleO on a bunch of Intel NUCs. To enable this he needed to make
some minor changes [1] to os-cloud-config and python-tripleoclient to
add basic support for AMT (utilizing wsmancli).
I have tried this out myself with a set of Lenovo M93p's and it does
work. I believe basic support of AMT can be very helpful to allow
tests on more COTS products, when IPMI is not available. Is this
something that can be considered for inclusion?
regards,
Gerard
[1] http://acksyn.org/files/tripleo/nuc.patch
--
Gerard Braad | http://gbraad.nl
[ Doing Open Source Matters ]
10:39 p.m.
On 06/06/16 13:32, Gerard Braad wrote:
HI,
Some time ago Michele Baldessari wrote about deploying OpenStack
TripleO on a bunch of Intel NUCs. To enable this he needed to make
some minor changes [1] to os-cloud-config and python-tripleoclient to
add basic support for AMT (utilizing wsmancli).
I have tried this out myself with a set of Lenovo M93p's and it does
work. I believe basic support of AMT can be very helpful to allow
tests on more COTS products, when IPMI is not available. Is this
something that can be considered for inclusion?
regards,
Gerard
[1] http://acksyn.org/files/tripleo/nuc.patch
Full support for Intel AMT and NUCs has been in upstream TripleO for a
while (and I believe downstream as of RHOS 8). For NUCs without AMT you
can use the wol driver which I use for all testing in my NUC lab.
Regards,
Graeme
--
Graeme Gillies
Principal Systems Administrator
Openstack Infrastructure
Red Hat Australia
11:13 p.m.
Hi.
On Mon, Jun 6, 2016 at 11:39 AM, Graeme Gillies <ggillies(a)redhat.com> wrote:
Full support for Intel AMT and NUCs has been in upstream TripleO for
a
while (and I believe downstream as of RHOS 8). For NUCs without AMT you
can use the wol driver which I use for all testing in my NUC lab.
Is support available because of the following?
/tripleo-common/tripleo_common/utils/nodes.py:
187: '.*_amt': PrefixedDriverInfo('amt'),
If so, I was unable to find documentation that describes this is possible.
It would have been sufficient to describe a node in instackenv.json as follows
"pm_type":"pxe_amt",
"pm_user":"admin",
"pm_password":"admin",
"pm_addr":"192.168.1.101",
"mac": ["de:8a:ad:be:3c:ef:"]
without any additional patches or software? The article [1] I refer to
was written last March. And this is also what I needed to make it
work.
regards,
Gerard
[1] http://acksyn.org/posts/2016/03/tripleo-on-nucs/
--
Gerard Braad | http://gbraad.nl
[ Doing Open Source Matters ]
11:30 p.m.
On 06/06/2016 06:13 AM, Gerard Braad wrote:
Hi.
Hi Gerard.
On Mon, Jun 6, 2016 at 11:39 AM, Graeme Gillies <ggillies(a)redhat.com> wrote:
> Full support for Intel AMT and NUCs has been in upstream TripleO for a
> while (and I believe downstream as of RHOS 8). For NUCs without AMT you
> can use the wol driver which I use for all testing in my NUC lab.
Is support available because of the following?
/tripleo-common/tripleo_common/utils/nodes.py:
187: '.*_amt': PrefixedDriverInfo('amt'),
Yes.
If so, I was unable to find documentation that describes this is possible.
It would have been sufficient to describe a node in instackenv.json as follows
"pm_type":"pxe_amt",
"pm_user":"admin",
"pm_password":"admin",
"pm_addr":"192.168.1.101",
"mac": ["de:8a:ad:be:3c:ef:"]
without any additional patches or software? The article [1] I refer to
was written last March. And this is also what I needed to make it
work.
Should be, the patch enabling it was merged post-March.
regards,
Gerard
[1] http://acksyn.org/posts/2016/03/tripleo-on-nucs/
11:57 p.m.
Hi,
On Mon, Jun 6, 2016 at 12:30 PM, Dmitry Tantsur <dtantsur(a)redhat.com> wrote:
On 06/06/2016 06:13 AM, Gerard Braad wrote:
> Is support available because of the following?
> /tripleo-common/tripleo_common/utils/nodes.py:
> 187: '.*_amt': PrefixedDriverInfo('amt'),
Yes.
So, is this in Liberty packages? If so, I will retest this later this week.
regards,
Gerard
--
Gerard Braad | http://gbraad.nl
[ Doing Open Source Matters ]
12:52 a.m.
On 06/06/2016 06:57 AM, Gerard Braad wrote:
Hi,
On Mon, Jun 6, 2016 at 12:30 PM, Dmitry Tantsur <dtantsur(a)redhat.com> wrote:
> On 06/06/2016 06:13 AM, Gerard Braad wrote:
>> Is support available because of the following?
>> /tripleo-common/tripleo_common/utils/nodes.py:
>> 187: '.*_amt': PrefixedDriverInfo('amt'),
> Yes.
So, is this in Liberty packages? If so, I will retest this later this week.
Sorry, I didn't realize you're asking about Liberty. I'm afraid it's
only available in Mitaka.
You can always enroll nodes manually, something like:
ironic node-create -d pxe_amt -i amt_username=user -i
amt_password=password -i amt_address=address
ironic port-create -n NODE_UUID -a MAC
and then proceed with 'configure boot' and inspection.
regards,
Gerard
12:58 a.m.
Hi Gerard,
On Mon, Jun 06, 2016 at 11:32:34AM +0800, Gerard Braad wrote:
Some time ago Michele Baldessari wrote about deploying OpenStack
TripleO on a bunch of Intel NUCs. To enable this he needed to make
some minor changes [1] to os-cloud-config and python-tripleoclient to
add basic support for AMT (utilizing wsmancli).
I have tried this out myself with a set of Lenovo M93p's and it does
work. I believe basic support of AMT can be very helpful to allow
tests on more COTS products, when IPMI is not available. Is this
something that can be considered for inclusion?
As Dmitry mentioned, most of the stuff has landed already. I
haven't yet played with it, because the disk hosting my
undercloud vm died and I had no time to reinstall. Once I get
to it, I will drop you a line (unless you beat me to it ;)
cheers,
Michele
--
Michele Baldessari <michele(a)acksyn.org>
C2A5 9DA3 9961 4FFB E01B D0BC DDD4 DCCB 7515 5C6D
1:02 a.m.
Hi,
On Mon, Jun 6, 2016 at 1:58 PM, Michele Baldessari <michele(a)acksyn.org> wrote:
As Dmitry mentioned, most of the stuff has landed already. I
haven't yet played with it, because the disk hosting my
undercloud vm died and I had no time to reinstall. Once I get
to it, I will drop you a line (unless you beat me to it ;)
Thanks for the input...
I also have to make some time for this... I was lucky our home systems
had to be reinstalled, so I was able to do a whole flow... but I will
try as what Dmitry just suggested. I will try to manually enroll a
system.
cheers,
Gerard
--
Gerard Braad | http://gbraad.nl
[ Doing Open Source Matters ]
2880
days inactive
2880
days old
7 comments
4 participants
participants (4)
-
Dmitry Tantsur -
Gerard Braad -
Graeme Gillies -
Michele Baldessari