Hi Hisaharu,
Thanks for sharing this design proposal and the POC code.
I will have a look at the code as soon as possible.
At a first glance, I think the design that you are proposing is in line with
the goals of the network service blueprint
(http://wiki.openstack.org/NetworkService)
If I got
Is there at least a way for not showing branches which are not proposed for
merge into lp:nova?
Salvatore
-Original Message-
From: openstack-bounces+salvatore.orlando=eu.citrix@lists.launchpad.net
[mailto:openstack-bounces+salvatore.orlando=eu.citrix@lists.launchpad.net]
On
+salvatore.orlando=eu.citrix@lists.launchpad.net]
On Behalf Of Salvatore Orlando
Sent: 25 February 2011 18:23
To: Jay Pipes; Justin Santa Barbara
Cc: openstack@lists.launchpad.net
Subject: Re: [Openstack] Working on fixing code after a review? Please mark
merge proposal Work In Progress
Hi Thierry,
Does the branch merge proposal freeze also affect branches On hold for
merging? (i.e.: branches which are being fixed after a review)
Thanks,
Salvatore
-Original Message-
From: openstack-bounces+salvatore.orlando=eu.citrix@lists.launchpad.net
Hi Thierry,
Will you consider some help from non-core team members?
Salvatore
-Original Message-
From: openstack-bounces+salvatore.orlando=eu.citrix@lists.launchpad.net
[mailto:openstack-bounces+salvatore.orlando=eu.citrix@lists.launchpad.net]
On Behalf Of Thierry Carrez
Hi Rick,
Wednesday 22:00 UTC would be fine for me.
Salvatore
-Original Message-
From: Rick Clark [mailto:r...@openstack.org]
Sent: 10 May 2011 10:12
To: Josh Wilmes
Cc: Dan Wendlandt; James Urquhart; Erik Carlin; Salvatore Orlando;
radur...@cisco.com; Ewan Mellor; Youcef Laribi
Hi,
The quantum API draft document has been converted into wiki pages:
API specification: http://wiki.openstack.org/QuantumAPISpec
Use Cases: http://wiki.openstack.org/QuantumAPIUseCases
There are anchors for each operation and use case, which can be used in the
I'm not sure I will be able to make it for tomorrow as I will be travelling as
well.
If that is the case, please accept my apologies for the absence.
Salvatore
-Original Message-
From: openstack-bounces+salvatore.orlando=eu.citrix@lists.launchpad.net
Hi Sumit,
It might be either a problem related with the routes version in RHEL or a
problem with the API implementation, which has not been tested on RHEL.
For instance, there might be a way for configuring the same routes without
causing the error you have encountered.
I have filed a bug
Hi,
I'm jumping in this thread to understand whether there's a chance KeyStone can
implement some use cases around the concept of object ownership.
I apologise if this email turns out to be a bunch of nonsense; unfortunately I
don't have a strong AAA background :).
As pointed out several times
Hi Vish,
Thanks a lot for this very clear and simple explanation of the ha-networking
option.
At a first glance, it seems this feature should work without any modification
even on XenServer deployments. I'll give it a try and I will let you know!
It seems that the approach for attributing IPs
Hi a quick question,
I just branched from lp:nova and fetched revision number 1500. Whereas
lp:nova/Diablo is at revno 1617.
Is this just because of the fork which occurred at rbp?
I'm asking because it looks like some bits of code, such as the Quantum
Manager, which was included in Diablo-rbp,
@lists.launchpad.net]
On Behalf Of Salvatore Orlando
Sent: 23 September 2011 00:04
To: openstack@lists.launchpad.net (openstack@lists.launchpad.net)
Subject: [Openstack] Current rev of lp:nova
Hi a quick question,
I just branched from lp:nova and fetched revision number 1500. Whereas
lp:nova/Diablo
One of those folks just filled the details he could fill!
Salvatore
From: openstack-bounces+salvatore.orlando=eu.citrix@lists.launchpad.net
[mailto:openstack-bounces+salvatore.orlando=eu.citrix@lists.launchpad.net]
On Behalf Of Dan Wendlandt
Sent: 10 October 2011 18:52
To: Debo Dutta
@lists.launchpad.net; Salvatore Orlando
Subject: Re: [Openstack] Compute with Xen
On Wed, Mar 14, 2012 at 10:29 AM, Eduardo Nunes
eduardo.ke...@gmail.com wrote:
i'e installed the xcp with this notes,
http://wiki.xen.org/wiki/XAPI_on_Ubuntu
and instaled the compute, but it's now working
Hi Thomas,
I can probably help you somehow with Openstack networking on XenServer, as I
did some work on it in the past.
I see you are trying to use the VLAN manager, but the behaviour is not the
expected one.
However, since you can spin up instances, and they appear to be attached to the
Of Todd Deshane
Sent: 21 March 2012 03:36
To: Salvatore Orlando
Cc: John Garbutt; Thomas Goirand; xen-api; PKG OpenStack; Dave Scott;
openstack@lists.launchpad.net
Subject: Re: [Openstack] [Xen-API] nova-xcp-network plugin searching for a
xapi0 interface
On Tue, Mar 20, 2012 at 8:51 PM
I recently had a (probably) similar issue, with xcp-squeezed consuming all the
memory and then causing a reboot.
I was using a pae kernel (3.0.0-16.generic.pae), and the workaround, as
suggested by the Kronos team, was to increase dom0 memory with a boot parameter
to the hypervisor. I used
Hi,
During our last project meeting there was some interest in introducing
review days for Quantum core devs.
I see multiple benefits in this approach:
1) Contributors would know which core member the need to contact on a given
days for soliciting a review for their patches
2) Predictable code
As the link to the wiki page is now buried deep in this thread, here it is
again: http://wiki.openstack.org/Quantum/ReviewDays
Weekend days have been added on GaryK's request!
Salvatore
On 11 June 2012 16:51, Dan Wendlandt d...@nicira.com wrote:
On Mon, Jun 11, 2012 at 2:25 AM, Salvatore
While I understand the reasons for which the RequestExtensions scheme might
be preferred over the resource extension one, I was wondering whether
assigning namespaces to attributes could be a solution to the XML
issue. This would clearly separate core attributes from extended ones.
Bob, can you
+111 from me.
This would make a lot easier extracting info from logs and correlating
events in different logs. In the long run these association might also be
used by metering/billing applications.
This is probably out of scope, but I was wondering whether we can also add
a concept of parent
Thanks a lot for sharing this information.
Even if HACKING.rst has an example along the same lines, it does not
state clearly that ordering should include also the name of the
module/class being imported.
Quoting LZY's example, I think it might be worth adding something like
the following to
Hi Yong,
I do not see any obvious reason for these failures, especially as they
appear to occur when the vdi is created. If I recall it correctly, that
code is stubbed out for unit tests, and it does not seem your patch
un-stubs it.
Do you see the failures also on your dev machine?
Salvatore
On
something to do with mox, but I will let you know soon - unfortunately
each test run takes almost 10 minutes!
Salvatore
On 2 July 2012 19:18, Salvatore Orlando sorla...@nicira.com wrote:
Hi Yong,
I do not see any obvious reason for these failures, especially as they
appear to occur when the vdi
://paste.openstack.org/show/19204/
Regards,
Salvatore
On 2 July 2012 22:53, Salvatore Orlando sorla...@nicira.com wrote:
Hi Yong,
I have been able to reproduce the error on my dev machine - I couldn't
earlier on because I was tesyinh the test_xenapi module only, and no
failure occured.
It seems
Hi Phil,
Thanks for taking this initiative. I think there has not been any Openstack
event in the UK since the EMEA day back in July 2011.
I am definitely interested; ideally, I would prefer East Anglia or London,
but anywhere in southern England or East Midlands will work for me.
Regards,
Hi,
A proposal for the implementation of the public networks feature has been
published.
It can be reached from the quantum-v2-public-networks blueprint page [1].
Feedback is more than welcome!
Regards,
Salvatore
[1]:
https://blueprints.launchpad.net/quantum/+spec/quantum-v2-public-networks
] Public Network spec proposal
To: Salvatore Orlando sorla...@nicira.com
Cc: openstack-...@lists.openstack.org
See inline comments.
Thanks
-Salvatore Orlando sorla...@nicira.com sorla...@nicira.com wrote:
-
To: Yong Sheng Gong/China/IBM@IBMCN
From: Salvatore Orlando sorla...@nicira.com
#3. This seems to be a cleaner approach for the user
interface.
Thanks
Gary
On 07/12/2012 01:52 AM, Salvatore Orlando wrote:
Hi,
A proposal for the implementation of the public networks feature has
been published.
It can be reached from the quantum-v2-public-networks blueprint page [1
I don't think either of you is wrong. I too think that in cases where it's
not easy to find a majority, it might make sense to just do what the other
projects are doing.
Unfortunately for us, Keystone adopts the name is unique phylosophy,
whereas nova adopts name is a label.
Is it worth
blueprint.
Salvatore
On 14 July 2012 23:10, Gary Kotton gkot...@redhat.com wrote:
**
On 07/12/2012 06:39 PM, Salvatore Orlando wrote:
Thank you again for your feedback.
On the discussion about two or three-way logic, I understand Yong's
point of being able to fetch public and private networks
particular users, say power users the power of creating public networks.
Regards,
Salvatore
[fwd to openstack-dev - please ensure it is kept in the recipient list]
On 17 July 2012 15:04, Salvatore Orlando sorla...@nicira.com wrote:
Gary,
I think your are making a very good point here
On 23 July 2012 09:02, Dan Wendlandt d...@nicira.com wrote:
On Sun, Jul 22, 2012 at 5:51 AM, Gary Kotton gkot...@redhat.com wrote:
**
This is an interesting idea. In addition to the creation we will also
need the update. I would prefer that the agents would have one topic - that
is for
Aaron,
it would be great if you can update the wiki page for running Quantum with
devstack with the info in http://wiki.openstack.org/RunningQuantumV2Api that
you created.
I also had other people asking me about how to run devstack with Quantum
v2, and I pointed them to your page.
On the issue
Hi,
Folsom-3 is going to be released this evening GMT. Expect it to be
available after 1700 GMT.
You'll find both a folsom-3 tag on github, as well as tarballs downloadable
from launchpad.
Regards,
Salvatore
On 17 August 2012 08:50, 延生 付 willfly0...@yahoo.com.cn wrote:
Seems not release yet.
Hi Alessandro,
When I get 400 errors from nova command line it is usually because of
invalid values in the cache.
Try executing your commands with the --no_cache option or destroying the
cache altogether by removing the .novaclient directory in your home dir.
Salvatore
On 22 August 2012 11:45,
...
The tests for each plugin could then be executed via an independent tox
run.
Is there any merit to this, now or in the future?
Thanks,
Maru
On 2012-08-24, at 2:56 PM, Salvatore Orlando (Code Review) wrote:
Salvatore Orlando has uploaded a new change for review.
Change subject: Enable tox
Hi Naveen,
That is almost correct. You can associate a provider network (either a
physical network or a vlan) with a Quantum network. The association is at
the network level, not the subnet level.
As the 'provider networks' is an extension in quantum, there is no direct
CLI option available.
On 4 September 2012 22:16, Trey Morris trey.mor...@rackspace.com wrote:
The transition is going to be difficult either way when you consider data
migrations..
This is one huge aspect. The other concerns deployment, as Quantum
interacts with nova-compute in a quite different way and roll out
It seems that quantum is rejecting authentication credentials supplied by
nova; please verify the following nova flags are set to the correct value:
- quantum_admin_username
- quantum_admin_tenant_name
- quantum_admin_auth_url (should he http://keystone_endpoint/v2.0
- quantum_url_timeout
-
Hi Balaji,
Quantum now has a dhcp agent which serves addresses to subnets using
dnsmasq.
The dhcp agent is a separate binary (bin/dhcp-agent). per-net-dhcp-enable
has been translated into a simple feature that enables or disables the dhcp
agent for specific subnets.
This is controlled by the
Hi Lingala,
the dhcp agent currently only works with dnsmasq. Even if this explain why
you do not see dhcpd, we still need to investigate why dnsmasq is not
running.
From your description, it seems that this is the root cause of your issue.
If no dnsmasq instance is seen on the node where
Hi Janis,
If your provider network maps directly to a publicly accessible network
then you should be able to assign IPs to VMs from a public range without
using Floating IPs and NAT.
With Quantum, you just specify a physical network (e.g.: eth0) and a
segmentation id (e.g.: vlan id 999) for
The information you found in the guide you mentioned indicate you want to
use br-eth1 as the uplink for the VMs attached to your network.
However, creating br-eth1 (or whatever is the name of the bridge) is beyond
the scope of Quantum. It is up to who deploys Quantum configure
connectivity on the
Hi Vinay,
I understand your concerns about conflicts with already assigned OUIs.
It is however my opinion that it is not up to the Openstack Foundation, but
to entities deploying Openstack, to buy MAC OUIs.
As regards Quantum, we should ensure the default MAC range we use is
locally assigned;
Srikanth,
from your analysis it seems that L2 connectivity between the compute and
the controller node is working as expected.
Before looking further, it is maybe worth ruling out the obvious problems.
Hence:
1) is the dhcp-agent service running (or is it stuck in some error state?)
2) Can you
Hi Trinath,
Even if is perfectly reasonable to run the DHCP/L3 agents in the controller
node, the advice we give in the administration guide is slightly different.
As suggested in [1], the only Quantum component running on the controller
node should be the API server.
The DHCP and L3 agents might
Hi Johanna,
are you running Openstack with Quantum or nova-network? In the former case,
which plugin are you using?
If you are using nova-network, which network manager are you using?
Salvatore
On 2 November 2012 06:32, Heinonen, Johanna (NSN - FI/Espoo)
johanna.heino...@nsn.com wrote:
Hi
Hi Veera,
Is your setup using nova-network or Quantum?
what network services are you running in the controller node?
If you're using quantum you might provide information about the plugin
you're using; in case of nova-network, knowing which network manager you're
running might help the community
Hi Balaji,
I am not sure I understand your questions. I think that with clients your
referring to python-novaclient and/or python-quantumclient.
If that is correct, those are merely applications that provide users with
tools for access the respective endpoints. These applications are usually
not
Kevin is correct; as your're running IP namespaces, you'll need to look for
iptables in that namespace.
The metadata settings however are visible in the log extract you posted.
It seems your metadata server is sitting at 172.16.0.201:8775
I guess you've already ensured that endpoint is reachable
Hi Ilkka,
you can associate host routes with a subnet. This will cause the dhcpagent
to send an option 121 (classless-host-route) in the DHCPOFFER packet, thus
configuring VM's route table.
However, we do not support at the moment configuration of custom routes
inside a logical router. The
Hi Paras,
If your goal is to delete namespaces, have you tried the
quantum-netns-cleanup utility?
If the quantum network 20554b0b-dc5f-48c5-87fa-47b90dc9242f and the
quantum router 39d5fa21-c604-4d3b-a37b-90457c9b11fe have not beed
deleted it's unlikely however the namespaces will be deleted.
Hi Trevor,
sorry for the delay in the response.
It seems you want a L2-only network.
In theory, all you should do to achieve this is to not create a subnet
on your network.
Disabling DHCP on the subnet won't make any difference, since the
--enable-dhcp option only controls the IP address delivery
Hi Greg,
I would like to understand a little bit better your setup.
I scrolled through the thread, but I'm not sure if you've already
provided the information I'm asking.
On the compute node, the pair phy-br-eth1/int-br-eth1 is a veth pair
or a OVS patch port?
I'm assuming you're using the OVS
It's great to see that things are starting to run properly!
I'm sorry I did not read you were running a provider network. That
would have been the typical symptom of a missing mapping.
Some more comments inline on the 'new' issues.
Salvatore
On 20 February 2013 21:57, Greg Chavez
The other thread currently active at the moment is titled [Openstack]
VM doesnt get IP.
Aaron Rosen, Quantum core developer, is assisting a member of the
community who's apparently having exactly the same problem as you are.
Please note that 169.254.169.254 is the metadata address. In order for
I'm not sure I followed the thread correctly from the beginning, but I
read that you have configured you NIC for private VM networking, in
VLAN mode, on VLAN 105.
Is that correct?
In general trunking all your switch ports used for VM networking will
save you the hassle of adding the VLANs you are
The logs should provide also the reason for the failure you are reporting.
If you're using rootwrap, one thing I'd very is that ovs-vsctl is in
your filter list.
You should have something like the following:
ovs-vsctl: CommandFilter, /bin/ovs-vsctl, root
ovs-vsctl_usr: CommandFilter,
Hi Zhidong,
Please look inline for answers to your questions.
On 15 March 2013 08:05, Zd Yu zdyu2...@gmail.com wrote:
I am confused by the existence of both 'network' and 'subnet'. According to
v2 API spec[1], subnet is introduced to support IPAM capabilities. Can
anyone elaborate it to me?
Hi Samuel,
in your logs you should have another traceback related to the failure
in creating the qg-xx interface.
This might help us shed some light on your issue. The only likely
cause I can think of at the moment is that the external bridge
(usually br-ex) is not correctly configured.
Hi Ankush,
unfortunately I am not aware of the possibility of using an external dhcp
server for quantum, neither in folsom or grizzly).
Currently, the dnsmasq option appears to be the only available one.
Salvatore
On 16 April 2013 08:02, ankush grover ankushcen...@gmail.com wrote:
Hi
Quantum's metadata solution for Grizzly can run either with or without the
l3 agent.
When running within the l3 agent, packets directed to 169.254.169.254 are
sent to the default gateway; the l3 agent will spawn a metadata proxy for
each router; the metadata proxy forwards them to the metadata
The dhcp agent will set a route to 169.254.0.0/16 if
enable_isolated_metadata_proxy=True.
In that case the dhcp port ip will be the nexthop for that route.
Otherwise, it might be your image might have a 'builtin' route to such
cidr.
What's your nexthop for the link-local address?
Salvatore
On
Hi Aliki,
You are correct, since you won't need the network node as host of Quantum
services as described in the guide you linked. However you might still want
to run nova-network on a dedicated network node.
Also, without quantum of course you won't need to run the l2 agent on each
compute
Hi Joe,
are you using the OVS plugin with GRE overlays?
In that case your problem might be the fact that the plugin pushes a OVS
flow entry which applies the 'local' vlan tag only to packet directed to
the VM's mac [1]
To me, this does not look like a bug; it's probably intended behaviour, as
it
bridge and add ethX to it you'll
also have to set the mac on your bridge to be the same as ethX (which is
the mac that quantum handed out).
Aaron
On Tue, Apr 30, 2013 at 4:25 PM, Salvatore Orlando sorla...@nicira.comwrote:
Hi Joe,
are you using the OVS plugin with GRE overlays?
In that case
On 21 May 2013 05:01, Zhidong Yu zdyu2...@gmail.com wrote:
It seems there are two different meaning for this term. One is about a
sort of usage scenario where the virtual network is *directly *mapped to
a physical and instance is *directly *connected to the outside world
without floating IP
On 23 May 2013 13:15, Ashok Kumaran ashokkumara...@gmail.com wrote:
Since you have a single node installation , for connecting to the VM's
from the base host you won't need router/l3 stuffs.if its not happening
then you might need to check the secgroup rules.
This is true only if you're
Nova does not have an option for creating 'unbound NICs', which would be
tantamount to unplugged network cards, if this is what you want to achieve.
Using Quantum you can create a few ports on some network, set these ports
administratively down, and boot the VM with this ports (--nic port-id).
:06.007 3232 TRACE nova.api.openstack
On Sat, May 25, 2013 at 5:33 AM, Salvatore Orlando sorla...@nicira.comwrote:
Nova does not have an option for creating 'unbound NICs', which would be
tantamount to unplugged network cards, if this is what you want to achieve.
Using Quantum you can create
Hi Dhanasekaran,
It seems to me that the 'shared nothing' architecture [1] probably has
little to do with your need, as it refers to the various nova nodes, rather
than instances.
It looks like you want to start an instance which is distributed across
several nodes. Is this your goal?
Salvatore
We will need to look at iptables on your network node.
If you run iptables -n -tnat --list you should see a couple of DNAT/SNAT
rules for forwarding traffic netween 9.126.108.127.
In any case, bear in mind that the default security group does not allow
ICMP. If you have not enabled it, it might
Hi again,
You would need a compute driver for a hypervisor supporting a distributed
virtual machine.
I have a very limited knowledge of server virtualization and nova drivers,
but I don't think such driver exists for nova.
In fact, I don't think a hypervisor which does that is generally
Good job guys.
I reckon we might make users' life easier if we change naming strategy for
default security groups to 'default-$tenant_id'
On the other hand this is not a priority since as an admin user I guess you
can already get that information properly choosing the fields to display.
Salvatore
The reason is merely legal.
More information are available in this thread:
https://lists.launchpad.net/openstack/msg22544.html
Salvatore
On 21 June 2013 00:38, tudor tu...@tudorholton.com wrote:
**
Can you give the reasoning behind the name change? It seems strange to
make this change
Yes, it is a very common use case.
Nowadays, everybody who deploys Xen, KVM, or LXC without a Cloud Management
System is actually using Open vSwitch.
In that case you can program Open vSwitch using its own interface to
provide virtual networks.
When deploying Openstack, you can use Open vSwitch
Comments inline.
Salvatore
On 22 June 2013 14:09, Dennis Jacobfeuerborn denni...@conversis.de wrote:
On 22.06.2013 12:14, Salvatore Orlando wrote:
Yes, it is a very common use case.
Nowadays, everybody who deploys Xen, KVM, or LXC without a Cloud
Management System is actually using Open
Comments inline.
I do apologise but I do not know what Pandora is.
I hope you will find my answers still useful.
Regards,
Salvatore
On 22 June 2013 12:23, Peter Cheung mcheun...@hotmail.com wrote:
Thanks Salvatore
I am thinking these:
1) openvswitch has more features than quantum, so i want
A few comments inline from my side.
On 24 June 2013 14:53, Peter Cheung mcheun...@hotmail.com wrote:
Hi Kyle
Is Neutron support this
1) create vlan, adding a port(tap) for a vm?
I think this is one of Neutron's basic use cases, unless I'm missing
something.
2) create vlan tag and
Hi Peter,
So far I think this is your best bet:
http://docs.openstack.org/developer/quantum/
There might be other documents, but none that I'm aware of.
There's plenty of awesome people writing on their own blogs about
openstack, and possibly about developing for Neutron.
planet.openstack.org
Hi,
you don't have to create networks in nova too - the ones you created with
quantum should be enough.
First, it might perhpaps good to exclude possible issues with your image.
If you're using KVM you could go and check the libvirt XML which is
generated for your instance. If you see a populated
Hav you tried checking the credentials that glance uses for validating
tokens with keystone?
They are defined in glance's conf files in the section:
[keystone_authtoken]
signing_dir = /var/cache/glance/api
auth_uri = http://127.0.0.1:5000/
auth_host = 127.0.0.1
auth_port = 35357
auth_protocol =
84 matches
Mail list logo