Open Stack

I'm also very interested to hear about this. From the looks of it, there
won't be an Active/Active solution for the L3 agent without intervention.

If you look back through the thread, you'll see a couple links to bugs/bps
which are in place for Havana. The DHCP agent is now Active/Active.



On Thu, Sep 12, 2013 at 6:00 AM, Kevin Jackson <kevin at linuxservices.co.uk>wrote:

> Hi all,
> What is the current development status of Neutron and HA.  We all should
> know by now some of the issues we have and have to workaround with the
> likes of the L3 agent, and it's routes being available on a single node
> that luckily got the request to create it.  If it fails, routing fails.
>
> In order for Neutron to be feature-parity with Nova Network, I expect this
> to be a solved solution in Havana?  Is that a fair assumption or would we
> still need to make decisions around Neutron to include/exclude L3 agents
> and put some workarounds in place still?
>
> Thanks!
>
> Kev
>
>
> On 16 August 2013 10:19, PREMSUDHAN CHRISTOPHER.J <premedge at gmail.com>wrote:
>
>> Hi Simon/Roman/Jacob,
>>      I was reading your thread about Network node HA and even i wanted
>> the same thing to setup in our environment,But Am little confused as you
>> guys were talking about only three components (i.e quantum-dhcp-agent,
>> quantum-ovs-plugin-agent (i believe this is referred as L2 agent in the
>> thread), Quantum-L3-agent). There is one more daemon running
>> "quantum-metadata-agent" in network node, How do we achieve HA for it?
>>
>> Apart from DHCP and L3 agent, can we have keepalived or pacemaker (just
>> switching the VIP or active/passive) for quantum-ovs-plugin-agent &
>> quantum-metadata-agent.?
>>
>> Also for "Quantum L3 agent" will the telekomcloud solution work with
>> Openvswitch along with Vlan ? Also apart from VM internet connectivity what
>> are the other impacts when we do not have HA for L3 agent (i hope VM
>> instance will be accessible over the LAN even if L3 agent is down and if we
>> have the gateway of VM instance pointing to physical router or firewall
>> still we can achieve internet connectivity  ) ?
>>
>> Thanks & Regards,
>> Premsudhan Christopher.
>>
>>
>> On Tue, Jul 23, 2013 at 5:24 PM, Jacob Godin <jacobgodin at gmail.com>wrote:
>>
>>> Not specifically to Neutron HA, but a great video none the less:
>>> http://www.youtube.com/watch?v=N9mPepfxm-4
>>>
>>>
>>> On Fri, Jul 19, 2013 at 11:09 AM, Roman Podolyaka <
>>> rpodolyaka at mirantis.com> wrote:
>>>
>>>> Hi all,
>>>>
>>>> Thanks, Akihiro and Simon! I didn't know we had such option.
>>>>
>>>> >> Regarding 1, the configuration option "dhcp_agents_per_network" is
>>>> introduced. This options specifies the number of DHCP agents scheduled to
>>>> host a network.
>>>> >> If you set this option to the number of dhcp agents, you can do this.
>>>>
>>>> >> It already exists in Havana by setting the dhcp_agents_per_network
>>>> parameter in neutron.conf .
>>>> >> Check this commit for details:
>>>> https://github.com/openstack/neutron/commit/c95e58b70796eae6bbc8bac283f81df9aa1d1894
>>>>
>>>>
>>>> On Fri, Jul 19, 2013 at 4:35 PM, Simon Pasquier <
>>>> simon.pasquier at bull.net> wrote:
>>>>
>>>>>  Le 19/07/2013 15:07, Roman Podolyaka a écrit :
>>>>>
>>>>> Hi Jacob, all
>>>>>
>>>>>  1. When a new Neutron network is created, it is automatically
>>>>> scheduled to one of DHCP-agents running in a cluster (actually, it's done
>>>>> lazily, when the first port on this network is created, but that's an
>>>>> implementation detail). Though, you have to manually schedule it to
>>>>> additional DHCP-agents to provide HA. I haven't seen any BPs adding such an
>>>>> option, but I agree, that it would be useful to have one.
>>>>>
>>>>>
>>>>> It already exists in Havana by setting the dhcp_agents_per_network
>>>>> parameter in neutron.conf .
>>>>> Check this commit for details:
>>>>>
>>>>> https://github.com/openstack/neutron/commit/c95e58b70796eae6bbc8bac283f81df9aa1d1894
>>>>>
>>>>>
>>>>>  2. Not sure we have one :( The solution from TelekomCloud guys seems
>>>>> to be the only way to go to provide HA for L3 agents in Active/Active mode
>>>>> currently.
>>>>>
>>>>>  3. Every network or compute node in a cluster runs an L2-agent,
>>>>> which allows one to create L2 tenant networks and plug
>>>>> VMs/routers/DHCP-servers into network ports (the underlying implementation
>>>>> depends on the core plugin you use, e.g. OVS can use VLANs or GRE tunnels
>>>>> to provide tenant isolation). So L2 networking is ready to use when you've
>>>>> started an L2-agent and provided it with a correct configuration file.
>>>>>
>>>>>  Thanks,
>>>>> Roman
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> OpenStack-operators mailing listOpenStack-operators at lists.openstack.orghttp://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> Simon Pasquier
>>>>> Software Engineer
>>>>> Bull, Architect of an Open World
>>>>> Phone: + 33 4 76 29 71 49http://www.bull.com
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> OpenStack-operators mailing list
>>>>> OpenStack-operators at lists.openstack.org
>>>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
>>>>>
>>>>>
>>>>
>>>> _______________________________________________
>>>> OpenStack-operators mailing list
>>>> OpenStack-operators at lists.openstack.org
>>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
>>>>
>>>>
>>>
>>> _______________________________________________
>>> OpenStack-operators mailing list
>>> OpenStack-operators at lists.openstack.org
>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
>>>
>>>
>>
>>
>> --
>> Warm Regards ,
>>
>> Premsudhan.J
>>
>> _______________________________________________
>> OpenStack-operators mailing list
>> OpenStack-operators at lists.openstack.org
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
>>
>>
>
>
> --
> Kevin Jackson
> http://about.me/kevjackson
> @itarchitectkev
>
> _______________________________________________
> OpenStack-operators mailing list
> OpenStack-operators at lists.openstack.org
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-operators/attachments/20130912/7f047ac1/attachment.html>