Openstack IPv6 neutron confiuraton
Hi, We already have IPv4 vxLAN in our openstack. Now we want to add IPv6 for tenant network. Looking for help how to configure ipv6 tenant network in openstack neutron. Kindly help me to understand how ipv6 packet flow. Please give me some documents with network diagram. Thanks & B'Rgds, Rony
Hi, In tenant networks IPv6 packets are going same way as IPv4 packets. There is no differences between IPv4 and IPv6 AFAIK. In https://docs.openstack.org/neutron/latest/admin/deploy-ovs.html You can find some deployment examples and explanation when ovs mechanism driver is used and in https://docs.openstack.org/neutron/latest/admin/deploy-lb.html there is similar doc for linuxbridge driver. There are differences with e.g. how DHCP is handled for IPv6. Please check https://docs.openstack.org/neutron/latest/admin/config-ipv6.html for details.
On 2 Aug 2019, at 07:01, Md. Farhad Hasan Khan <rony.khan@brilliant.com.bd> wrote:
Hi, We already have IPv4 vxLAN in our openstack. Now we want to add IPv6 for tenant network. Looking for help how to configure ipv6 tenant network in openstack neutron. Kindly help me to understand how ipv6 packet flow. Please give me some documents with network diagram.
Thanks & B’Rgds, Rony
— Slawek Kaplonski Senior software engineer Red Hat
On Friday, August 02, 2019 09:16 CEST, Slawek Kaplonski <skaplons@redhat.com> wrote:
Hi,
In tenant networks IPv6 packets are going same way as IPv4 packets. There is no differences between IPv4 and IPv6 AFAIK. In https://docs.openstack.org/neutron/latest/admin/deploy-ovs.html You can find some deployment examples and explanation when ovs mechanism driver is used and in https://docs.openstack.org/neutron/latest/admin/deploy-lb.html there is similar doc for linuxbridge driver.
For private networking this is true, if you want public connectivity with IPv6, you need to be aware that there is no SNAT and no floating IPs with IPv6. Instead you need to assign globally routable IPv6 addresses directly to your tenant subnets and use address-scopes plus neutron-dynamic-routing in order to make sure that these addresses get indeed routed to the internet. I have written a small guide how to do this[1], feedback is welcome. [1] https://cloudbau.github.io/openstack/neutron/networking/ipv6/2017/09/11/neut...
There are differences with e.g. how DHCP is handled for IPv6. Please check https://docs.openstack.org/neutron/latest/admin/config-ipv6.html for details.
Also noting that the good reference article at the end of this doc sadly has disappeared, though you can still find it via the web archives. See also https://review.opendev.org/674018
At a quick look through Jens's guide, it looks like a really nice tutorial to follow. External connectivity is really an important point on IPv6 networking with neutron. I think my presentation at Boston summit two years ago still works [1]. This is based on my experience when I was involved in IPv6 POC with our customers (around Autumn in 2016). While Jens's guide covers detail commands for all, mine helps understanding some backgrounds on neutron with IPv6. [1] https://www.slideshare.net/ritchey98/openstack-neutron-ipv6-lessons Thanks, Akihiro Motoki (irc: amotoki) On Fri, Aug 2, 2019 at 5:14 PM Jens Harbott <frickler@x-ion.de> wrote:
On Friday, August 02, 2019 09:16 CEST, Slawek Kaplonski <skaplons@redhat.com> wrote:
Hi,
In tenant networks IPv6 packets are going same way as IPv4 packets. There is no differences between IPv4 and IPv6 AFAIK. In https://docs.openstack.org/neutron/latest/admin/deploy-ovs.html You can find some deployment examples and explanation when ovs mechanism driver is used and in https://docs.openstack.org/neutron/latest/admin/deploy-lb.html there is similar doc for linuxbridge driver.
For private networking this is true, if you want public connectivity with IPv6, you need to be aware that there is no SNAT and no floating IPs with IPv6. Instead you need to assign globally routable IPv6 addresses directly to your tenant subnets and use address-scopes plus neutron-dynamic-routing in order to make sure that these addresses get indeed routed to the internet. I have written a small guide how to do this[1], feedback is welcome.
[1] https://cloudbau.github.io/openstack/neutron/networking/ipv6/2017/09/11/neut...
There are differences with e.g. how DHCP is handled for IPv6. Please check https://docs.openstack.org/neutron/latest/admin/config-ipv6.html for details.
Also noting that the good reference article at the end of this doc sadly has disappeared, though you can still find it via the web archives. See also https://review.opendev.org/674018
Hi Motoki, Thanks. I shall check the slide. Thanks & B'Rgds, Rony -----Original Message----- From: Akihiro Motoki [mailto:amotoki@gmail.com] Sent: Friday, August 2, 2019 6:29 PM To: rony.khan@brilliant.com.bd; OpenStack Discuss Subject: Re: Openstack IPv6 neutron confiuraton At a quick look through Jens's guide, it looks like a really nice tutorial to follow. External connectivity is really an important point on IPv6 networking with neutron. I think my presentation at Boston summit two years ago still works [1]. This is based on my experience when I was involved in IPv6 POC with our customers (around Autumn in 2016). While Jens's guide covers detail commands for all, mine helps understanding some backgrounds on neutron with IPv6. [1] https://www.slideshare.net/ritchey98/openstack-neutron-ipv6-lessons Thanks, Akihiro Motoki (irc: amotoki) On Fri, Aug 2, 2019 at 5:14 PM Jens Harbott <frickler@x-ion.de> wrote:
On Friday, August 02, 2019 09:16 CEST, Slawek Kaplonski <skaplons@redhat.com> wrote:
Hi,
In tenant networks IPv6 packets are going same way as IPv4 packets. There is no differences between IPv4 and IPv6 AFAIK. In https://docs.openstack.org/neutron/latest/admin/deploy-ovs.html You can find some deployment examples and explanation when ovs mechanism driver is used and in https://docs.openstack.org/neutron/latest/admin/deploy-lb.html there is similar doc for linuxbridge driver.
For private networking this is true, if you want public connectivity with IPv6, you need to be aware that there is no SNAT and no floating IPs with IPv6. Instead you need to assign globally routable IPv6 addresses directly to your tenant subnets and use address-scopes plus neutron-dynamic-routing in order to make sure that these addresses get indeed routed to the internet. I have written a small guide how to do this[1], feedback is welcome.
[1] https://cloudbau.github.io/openstack/neutron/networking/ipv6/2017/09/1 1/neutron-pike-ipv6.html
There are differences with e.g. how DHCP is handled for IPv6. Please check https://docs.openstack.org/neutron/latest/admin/config-ipv6.html for details.
Also noting that the good reference article at the end of this doc sadly has disappeared, though you can still find it via the web archives. See also https://review.opendev.org/674018
Hi Slawek, Thanks. I shall check documents links. Thanks & B'Rgds, Rony -----Original Message----- From: Slawek Kaplonski [mailto:skaplons@redhat.com] Sent: Friday, August 2, 2019 1:16 PM To: rony.khan@brilliant.com.bd Cc: OpenStack Discuss Subject: Re: Openstack IPv6 neutron confiuraton Hi, In tenant networks IPv6 packets are going same way as IPv4 packets. There is no differences between IPv4 and IPv6 AFAIK. In https://docs.openstack.org/neutron/latest/admin/deploy-ovs.html You can find some deployment examples and explanation when ovs mechanism driver is used and in https://docs.openstack.org/neutron/latest/admin/deploy-lb.html there is similar doc for linuxbridge driver. There are differences with e.g. how DHCP is handled for IPv6. Please check https://docs.openstack.org/neutron/latest/admin/config-ipv6.html for details.
On 2 Aug 2019, at 07:01, Md. Farhad Hasan Khan <rony.khan@brilliant.com.bd> wrote:
Hi, We already have IPv4 vxLAN in our openstack. Now we want to add IPv6 for tenant network. Looking for help how to configure ipv6 tenant network in openstack neutron. Kindly help me to understand how ipv6 packet flow. Please give me some documents with network diagram.
Thanks & B’Rgds, Rony
— Slawek Kaplonski Senior software engineer Red Hat
participants (4)
-
Akihiro Motoki
-
Jens Harbott
-
Md. Farhad Hasan Khan
-
Slawek Kaplonski