| *** elo has quit IRC | 00:01 | |
| davidlenwell | I see what you are suggesting.. build our own plugin for neutron like octavia did anc capture the message.. format it the way we need for the rug driver to catch it and have enough information to act and push it back out to rpc | 00:06 |
|---|---|---|
| davidlenwell | but landing a driver in the neutron tree will take some effort.. what is the quick answer? | 00:07 |
| *** elo has joined #akanda | 00:17 | |
| *** elo has quit IRC | 00:54 | |
| *** puranamr has joined #akanda | 02:07 | |
| *** puranamr has quit IRC | 03:45 | |
| *** puranamr has joined #akanda | 04:16 | |
| *** puranamr has quit IRC | 04:20 | |
| *** stanchan has joined #akanda | 04:40 | |
| *** openstackgerrit has quit IRC | 05:31 | |
| *** openstackgerrit has joined #akanda | 05:31 | |
| *** puranamr has joined #akanda | 05:59 | |
| *** puranamr has quit IRC | 06:04 | |
| *** elo has joined #akanda | 07:28 | |
| *** elo has quit IRC | 07:38 | |
| *** elo has joined #akanda | 07:40 | |
| *** elo has quit IRC | 08:58 | |
| *** elo has joined #akanda | 11:01 | |
| *** elo has quit IRC | 11:02 | |
| *** elo has joined #akanda | 11:18 | |
| *** elo has quit IRC | 11:19 | |
| *** puranamr has joined #akanda | 12:02 | |
| *** puranamr has quit IRC | 12:06 | |
| *** puranamr has joined #akanda | 13:03 | |
| *** puranamr has quit IRC | 13:07 | |
| *** openstackgerrit has quit IRC | 13:46 | |
| *** openstackgerrit has joined #akanda | 13:46 | |
| *** elo has joined #akanda | 14:33 | |
| *** elo has quit IRC | 15:00 | |
| *** cleverdevil has joined #akanda | 16:01 | |
| *** puranamr has joined #akanda | 16:05 | |
| *** puranamr has quit IRC | 16:18 | |
| *** gsagie_ has joined #akanda | 16:22 | |
| *** gsagie_ has left #akanda | 16:40 | |
| drwahl | 2015-09-18 16:57:28.087 13371 TRACE nova ImportError: No module named akanda.nova.neutronv2.api | 16:57 |
| drwahl | seeing that in the nova-compute log | 16:58 |
| drwahl | what is network_api_class supposed to be (in kilo)? | 16:58 |
| rods | adam_g I opened https://bugs.launchpad.net/akanda/+bug/1496866 yesterday | 17:03 |
| openstack | Launchpad bug 1496866 in akanda "nameserver is not set correctly in the appliance" [Undecided,New] | 17:03 |
| rods | Unfortunately looks like DIB is not making it possible to fix without a hack | 17:05 |
| *** cleverdevil has quit IRC | 17:06 | |
| *** cleverdevil has joined #akanda | 17:06 | |
| rods | I fixed it in the DIB element I created for our custom conf, I can submit a pr if you are want | 17:08 |
| drwahl | i found what we neded for that nova config | 17:24 |
| adam_g | rods, thanks. i saw i havent had time to look closely yet. we might be able to fix it in DIB | 17:32 |
| drwahl | for kilo, does neutron still need some akanda patches? | 17:39 |
| drwahl | or is everything in upstream | 17:39 |
| drwahl | ? | 17:39 |
| adam_g | drwahl, no, should work wtih vanilla-everythign | 17:39 |
| drwahl | do we just need to drop all the "akanda" stuff from neutron.conf? | 17:42 |
| drwahl | also, why is devstack applying akanda-neutron? https://github.com/stackforge/akanda-rug/blob/master/devstack/plugin.sh#L236 | 17:42 |
| drwahl | err... wrong link | 17:43 |
| drwahl | https://github.com/stackforge/akanda-rug/blob/master/devstack/plugin.sh#L125 | 17:44 |
| adam_g | drwahl, you still need the akanda-neutron ML2 plugin loaded into neutron | 17:44 |
| adam_g | but that will load into a vanilla kilo+ neutron | 17:44 |
| *** jordantardif has joined #akanda | 17:45 | |
| drwahl | is the akanda-neutron ml2 plugin merged into kilo? | 17:46 |
| drwahl | or is that something we need to pull in separately? | 17:46 |
| adam_g | drwahl, it is its own akanda repository https://github.com/stackforge/akanda-neutron | 17:46 |
| adam_g | it has its own stable/kilo branch | 17:46 |
| adam_g | you can consider that repo 'upstream akanda' as well | 17:47 |
| drwahl | ok, so not everything is "upstream" it sounds like. we still need to include dependencies outside of neutron? | 17:49 |
| adam_g | drwahl, akanda-rug/akanda-neutron/akanda-horizon == upstream neutron | 17:52 |
| drwahl | so i can just clone the openstakc/neutron repo and have everything we need? | 17:53 |
| adam_g | .:adam_g:. drwahl, you still need the akanda-neutron ML2 plugin loaded into neutron | 17:53 |
| adam_g | derp | 17:54 |
| adam_g | akanda-rug/akanda-neutron/akanda-horizon == upstream *akanda* | 17:54 |
| drwahl | ah, ok | 17:54 |
| drwahl | that makes more sense now :) | 17:54 |
| adam_g | drwahl, we maintain the neutron ML2 driver out of tree, as its own upstream project a'la big tent | 17:54 |
| drwahl | is there a plan to get it into the mainline? | 17:55 |
| adam_g | drwahl, not that i know of. with the big tent stuff, maintaining it on our own is preferable. the neutron review queue would become a bottleneck for us making changes to it, and we would have a hard time testing it in their CI pipeline | 17:56 |
| jordantardif | adam_g: what about the router status stuff ? no plan on getting that upstreamed at least? | 17:58 |
| jordantardif | seems like there are a few things that are generic enough that they would be worth upstreaming | 17:58 |
| adam_g | jordantardif, im not sure about that. markmcclain might have a better idea | 18:02 |
| cleverdevil | yeah, I think over time, it will make sense to try and upstream things that are generally useful | 18:03 |
| cleverdevil | the ML2 driver makes sense out of tree | 18:04 |
| cleverdevil | but, enhancements and changes to neutron? | 18:04 |
| cleverdevil | those should go upstream, IMO :) | 18:04 |
| cleverdevil | (once they're baked and proven) | 18:04 |
| cleverdevil | bbiaf | 18:08 |
| *** cleverdevil has quit IRC | 18:08 | |
| adam_g | davidlenwell, so i was wrong yesterday and listening for *.create.start, when i shoul dbe looking for *.create.end which has all the context | 18:13 |
| adam_g | event_type: loadbalancer.create.end | 18:13 |
| adam_g | payload: {u'loadbalancer': {u'description': u'', u'provisioning_status': u'PENDING_CREATE', u'tenant_id': u'd22b149cee9b4eac8349c517eda00b89', u'admin_state_up': True, u'provider': u'akanda', u'listeners': [], u'vip_port_id': u'19fe1683-2eac-4722-bc6a-3ab30a1301bd', u'vip_address': u'192.168.0.10', u'vip_subnet_id': u'8c58b558-be54-45de-9873-169fe845bb80', u'id': u'779051d9-aaa1-45cb-805c-40ebc5077bd4', u'operating_status': u'OFFLINE', u'name': | 18:13 |
| adam_g | u''}} | 18:13 |
| davidlenwell | oh . sweet.. | 18:16 |
| davidlenwell | yeah . really thought it should have had | 18:16 |
| davidlenwell | I mean .. how else was any lbaas agent supposed to do its job | 18:17 |
| davidlenwell | adam_g: ^ | 18:17 |
| davidlenwell | well lets work together .. I want to contribute to this thing.. | 18:17 |
| adam_g | davidlenwell, so what of these are we trying to support with this thing? http://paste.ubuntu.com/12452474/ | 18:34 |
| adam_g | other than loadbalancer create/delete | 18:34 |
| davidlenwell | the member and pool stuff we'll need | 18:36 |
| adam_g | is there a TLDR on what those are and how they work? | 18:39 |
| markmcclain | jordantardif: the status extension will likely live as an out of tree extension for a bit | 18:39 |
| markmcclain | jordantardif, adam_g: the longer term view is that we should be able to leverage rpc and remove the need for the extension in M | 18:40 |
| davidlenwell | adam_g: looking for it now.. its in a wiki someplace and there is an lbaas v2 spec someplace | 18:40 |
| markmcclain | jordantardif, adam_g: for the ML2 driver a few changes have landed recently we might actually be able to do away with in M | 18:41 |
| adam_g | markmcclain, so just playing with the agent tsuff for lbaas it seems without an agent (ml2, lbaas) running somewhere, you only ever see the context-less *.start events and never the *.end ones | 18:42 |
| markmcclain | adam_g: interesting.. the start/end events are generated by teh rest framework | 18:43 |
| adam_g | markmcclain, it looks like the start is and the end might not get sent until an agent picks it up and returns. atm i have a noop driver loaded into the lbaas agent returning nothing | 18:44 |
| adam_g | i say that not having actually looked at the notification code in neutron, but just observing | 18:44 |
| markmcclain | hmmm.. and the rest call is returning? | 18:44 |
| markmcclain | or is it failing | 18:44 |
| adam_g | the rest call returns | 18:44 |
| markmcclain | interesting | 18:45 |
| adam_g | the *.create.end notification just reports that the async creation has happened (see paste above) | 18:45 |
| adam_g | s/has happened/has started | 18:45 |
| markmcclain | this should get called with ceilometer notifications are enabled: http://git.openstack.org/cgit/openstack/neutron/tree/neutron/api/v2/base.py#n465 | 18:46 |
| *** elo has joined #akanda | 18:57 | |
| drwahl | so, akanda dudes: we're running into a bit of an issue with what appears to be the auto_add_ipv6_subnet wrapper | 19:23 |
| drwahl | 2015-09-18 19:22:34.221 8271 TRACE neutron.api.v2.resource NoNetworkAvailable: Unable to create the network. No tenant network is available for allocation. | 19:23 |
| drwahl | any time we try to add a network | 19:24 |
| *** cleverdevil has joined #akanda | 19:39 | |
| *** elo has quit IRC | 19:53 | |
| *** elo has joined #akanda | 19:54 | |
| *** stanchan has quit IRC | 20:06 | |
| *** elo_ has joined #akanda | 20:19 | |
| *** elo has quit IRC | 20:22 | |
| *** elo_ has quit IRC | 20:34 | |
| markmcclain | drwahl: hmmm | 21:31 |
| markmcclain | kilo right? | 21:31 |
| drwahl | i think i got it | 21:34 |
| drwahl | we were missing some ml2 plugin configuration bits | 21:34 |
| *** stanchan has joined #akanda | 21:42 | |
| adam_g | davidlenwell, whats the plan for getting config pushed into the lbaas appliance? | 21:58 |
| davidlenwell | adam_g: was planning on dealing with it the same way the router does | 21:59 |
| adam_g | davidlenwell, how so? | 21:59 |
| adam_g | also, how does a lb's ports map to the booted appliance instance? | 22:01 |
| adam_g | the VIP port that gets created on the LB during its creation is marked as in-use and cant be used as the instance port | 22:01 |
| davidlenwell | https://github.com/stackforge/akanda-rug/blob/master/akanda/rug/api/akanda_client.py#L70 | 22:02 |
| adam_g | davidlenwell, so we run the akanda-appliance API service inside the VM with nginx? | 22:04 |
| davidlenwell | yeah .. but with some abstraction for drivers | 22:05 |
| davidlenwell | the alternitive is to use the nginx api for config and heartbeat .. but haveing the apliance api gives us more system level control | 22:06 |
| davidlenwell | figured it could act as a proxy | 22:06 |
| adam_g | davidlenwell, has the work thats needed for the API service to support config'ing loadbalancers been started somewhere? | 22:07 |
| davidlenwell | no | 22:11 |
| davidlenwell | but it should be fast | 22:11 |
| davidlenwell | I can make it this evening | 22:11 |
| davidlenwell | right now I have to go do dad things for a few hours .. but that was the general idea | 22:13 |
| *** openstackgerrit has quit IRC | 22:46 | |
| *** openstackgerrit has joined #akanda | 22:46 | |
| *** elo has joined #akanda | 23:10 | |
| *** cleverdevil has quit IRC | 23:19 | |
| *** elo has quit IRC | 23:24 | |
| *** cleverdevil has joined #akanda | 23:29 | |
| *** cleverdevil has quit IRC | 23:42 | |
| *** stanchan has quit IRC | 23:49 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!