15:16:31 <sc68cal> #startmeeting neutron_ipv6
15:16:32 <openstack> Meeting started Tue Sep  9 15:16:31 2014 UTC and is due to finish in 60 minutes.  The chair is sc68cal. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:16:34 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:16:37 <openstack> The meeting name has been set to 'neutron_ipv6'
15:17:16 <sc68cal> sorry, I'm sick in bed - probably should have sent an e-mail cancelling
15:17:45 <johnbelamaric> sorry to hear that
15:18:08 <sc68cal> xuhanp is not able to attend - but https://review.openstack.org/#/c/103811/ is in progress and will address the -1
15:18:47 <johnbelamaric> i am relatively new to openstack, and have been trying to determine the current status of IPv6 in juno - i hope you don't mind me asking here
15:19:50 <johnbelamaric> is neutron-l3-agent support implemented now, it sounded like most of the blueprints have been merged?
15:20:49 <sc68cal> johnbelamaric: we have support for ipv6 in Juno. Documentation needs to be written to help guide users
15:21:20 <sc68cal> in how to use
15:22:02 <johnbelamaric> great, thanks. are there still issues with dual stack? it sounded from previous meeting minutes there were some
15:22:43 <aveiga> dual-stack should work fine, however it's not full faeture parity
15:22:49 <aveiga> i.e. there are no floats yet, etc
15:23:04 <aveiga> security groups work though, and so do almost all of the provisioning methods.
15:23:30 <aveiga> there is also no config-drive support yet due to the hard-coding of a v4 addr
15:23:30 <johnbelamaric> no floats for ipv6 or just when running dual stack? just trying to understand current limitations
15:23:36 <aveiga> no floats for v6
15:23:42 <aveiga> I plan to file that one when K opens up
15:23:43 <johnbelamaric> ok
15:24:19 <SridharG> johnbelamaric: the following patch would also be a useful addition - https://review.openstack.org/#/c/77471/
15:24:49 <johnbelamaric> yeah, i saw that is abandoned
15:25:29 <SridharG> aveiga: I remember reading an email thread where it was suggested to use config-drive until metadata over IPv6 is supported.
15:25:48 <aveiga> SridharG: I get those backward all the time
15:25:49 <SridharG> aveiga: does config drive have issues with IPv6?
15:25:55 <aveiga> yes, it's the metadata IP that's wrong
15:26:06 <sc68cal> metadata IP is a v4 only API
15:26:16 <sc68cal> comes from Amazon, uses a link local ipv4 address
15:26:31 <SridharG> aveiga: sc68cal: yes you are right.
15:26:48 <sc68cal> if/when Amazon changes it to support IPv6, we can emulate
15:27:44 <HenryG> It should still work unless you completely turn off your IPv4 stack somehow.
15:27:46 <SridharG> sc68cal: any idea if Amazon has plans to support EC2 over IPv6?
15:28:09 <aveiga> lol
15:28:31 <SridharG> got the email thread - https://www.mail-archive.com/openstack-dev@lists.openstack.org/msg28671.html
15:29:02 <aveiga> HenryG: last I checked, nova won't build an instance without a v4?
15:29:46 <HenryG> aveiga: I haven't been keeping track. It doesn't surprise me though.
15:30:08 <sc68cal> yeah Nova requires a v4 address
15:30:37 <HenryG> We need to expand our scope to Nova and get that addressed.
15:30:46 <aveiga> +1
15:30:52 <HenryG> (Pun intended.)
15:30:58 <aveiga> it's mostly been a battle of getting basic working first, though
15:31:05 <sc68cal> Mark McClain has some experience with a hack to nova to make it work with v6 only
15:31:18 <aveiga> we can spread out now that full DS provisioning works in Juno
15:31:20 <sc68cal> it's somewhat nasty, but perhaps we can engage him to get a fix
15:32:23 <SridharG> Any idea if IPv6 is supported with Tenant Overlay networks (GRE/VxLANs)?
15:32:54 <aveiga> SridharG: I doubt it; we haven't tested any of this yet.  By a fluke some of them may work, depending on the driver
15:33:42 <SridharG> aveiga: ok.
15:33:59 <HenryG> Why wouldn't it work?
15:34:10 * HenryG needs to try it out
15:34:22 <SridharG> aveiga: I was looking at GRE/VxLANs support for IPv6 and found that IPv6 support in GRE is added in Linux Kernel 3.7 and VxLAN in 3.12 Kernel.
15:34:50 <aveiga> the same reason lots of things that should work dual-stacked may not: bad programming practices like accepting only INTs or not allowing a : in an address or letters
15:35:10 <aveiga> I'm not saying underlying GRE shouldn't work, just that implementations likely don't
15:35:42 <aveiga> for instance, anything relying on lpop is IPv4 only since they never implemented neighbor discovery
15:35:47 <aveiga> l2pop*
15:36:02 <HenryG> Yes, l2pop I can understand
15:36:33 <HenryG> Anyway, I should try it out for myself instead of asking silly questions, sorry. :)
15:36:54 <aveiga> no worries! I'd like to see some feedback on the ML if you make any progress with it
15:38:34 <HenryG> I have another silly question
15:39:20 <HenryG> Do instances work with GUAs in lieu of floating ips?
15:39:40 <aveiga> yes
15:40:02 <aveiga> that's how I'm running it now
15:40:36 <HenryG> OK, thanks, I think I am too, but my "global" network is not very global.
15:41:19 <HenryG> I.e. I can't seem to get out of my VirtualBox VM. Must be a VB thing.
15:41:57 <sc68cal> VB's networking is a little odd
15:42:15 <sc68cal> I've seen issues crossing the virt nics too
15:42:25 <sc68cal> on v4
15:48:24 <sc68cal> before I forget
15:48:44 <sc68cal> We have a bug for documenting the new ipv6 attributes. Currently it is assigned to me but I have not had capacity to address
15:49:00 <sc68cal> if someone wishes, they can take it up and I will be sure to help review
15:49:01 <sc68cal> https://webmail.comcast.com/owa/redir.aspx?C=jBAGQkQ45k6kh7NkmVeZ1SVs6gzZn9EIKlVJ4zoqOPO3dgV8w3YlMrt2w5lZoqURzNauGKfo5pw.&URL=https%3a%2f%2fbugs.launchpad.net%2fbugs%2f1293897
15:49:02 <sc68cal> doh
15:49:18 <sc68cal> https://bugs.launchpad.net/bugs/1293897
15:49:21 <uvirtbot> Launchpad bug 1293897 in openstack-manuals "Create new IPv6 attributes for Subnets" [Medium,Confirmed]
15:54:42 <sc68cal> any other business to discuss?
15:55:09 <sc68cal> otherwise we'll adjourn until next wek
15:59:11 <sc68cal> #endmeeting