| *** puranamr has quit IRC | 00:01 | |
| *** cleverde_ has quit IRC | 00:08 | |
| *** puranamr has joined #akanda | 00:10 | |
| davidlenwell | adam_g: is rug/openstack/common/rpc/impl_zmq.py still relevant after your patches to add oslo? | 00:19 |
|---|---|---|
| davidlenwell | oslo messaging? | 00:19 |
| adam_g | davidlenwell, nope, all of rug.openstack.common.rpc will be dropped | 00:20 |
| davidlenwell | ok .. thanks | 00:20 |
| davidlenwell | those patches are up and pending review right? or do you still have a ways to go on that ? | 00:21 |
| *** puranamr has quit IRC | 00:22 | |
| *** puranamr has joined #akanda | 00:31 | |
| *** puranamr has quit IRC | 00:37 | |
| openstackgerrit | Adam Gandelman proposed stackforge/akanda-rug: Migrate to oslo.i18n https://review.openstack.org/192395 | 00:38 |
| openstackgerrit | Adam Gandelman proposed stackforge/akanda-rug: Detangle config option registration https://review.openstack.org/191993 | 00:38 |
| openstackgerrit | Adam Gandelman proposed stackforge/akanda-rug: Migrate to oslo.log https://review.openstack.org/191241 | 00:38 |
| openstackgerrit | Adam Gandelman proposed stackforge/akanda-rug: Rework messaging layer for oslo.messaging https://review.openstack.org/190401 | 00:38 |
| openstackgerrit | Adam Gandelman proposed stackforge/akanda-rug: Add a log shim to avoid a dependency on openstack.common.log https://review.openstack.org/190400 | 00:38 |
| openstackgerrit | Adam Gandelman proposed stackforge/akanda-rug: Add support for hash-based RUG scale out https://review.openstack.org/195366 | 00:38 |
| davidlenwell | woa | 00:38 |
| sarob | hit me | 00:38 |
| adam_g | huh the ordering is wrong there | 00:39 |
| *** puranamr has joined #akanda | 00:39 | |
| adam_g | i had to rebase that whole stack on that oslo_config thing from doug | 00:39 |
| davidlenwell | which one should we start with | 00:39 |
| adam_g | https://review.openstack.org/#/c/192395/ is the top of the stack | 00:39 |
| adam_g | the RUG scale out is a WIP | 00:40 |
| adam_g | but, unless im overlooking something obvious, might be pretty easy. that implements something that mostly works tho i haven't tested much | 00:40 |
| adam_g | markmcclain, ^ | 00:40 |
| adam_g | actually maybe i can break that off that stack, doesnt really depend on anything there | 00:41 |
| adam_g | oh wait, yes it does (oslo_log) | 00:42 |
| *** puranamr has quit IRC | 00:42 | |
| adam_g | gotta run to the butcher for BBQ meats before the butcher goes home | 00:42 |
| davidlenwell | looks like it actually starts with https://review.openstack.org/#/c/191241/7 | 01:12 |
| *** puranamr has joined #akanda | 01:23 | |
| davidlenwell | adam_g: https://review.openstack.org/#/c/190400/ <-- still want to block this from merging? | 01:23 |
| adam_g | davidlenwell, so https://review.openstack.org/#/c/192395/ is the top of the oslo stack | 01:40 |
| adam_g | if you want to base on that, use that as the top | 01:40 |
| davidlenwell | sweet | 01:40 |
| adam_g | actually | 01:41 |
| adam_g | ill squash the log + messaging patches together now | 01:41 |
| davidlenwell | well I hope you were done with it.. sarob just merged it | 01:41 |
| adam_g | and get rid of that log shim thing | 01:41 |
| davidlenwell | k | 01:41 |
| adam_g | its not going to merge till that log shim one goes away | 01:41 |
| davidlenwell | ahh .. I see now | 01:42 |
| openstackgerrit | Adam Gandelman proposed stackforge/akanda-rug: Migrate to oslo.i18n https://review.openstack.org/192395 | 01:49 |
| openstackgerrit | Adam Gandelman proposed stackforge/akanda-rug: Add support for hash-based RUG scale out (WIP) https://review.openstack.org/195366 | 01:49 |
| openstackgerrit | Adam Gandelman proposed stackforge/akanda-rug: Migrate to oslo.log and oslo.messaging https://review.openstack.org/195373 | 01:49 |
| adam_g | davidlenwell, there ya go | 01:51 |
| davidlenwell | awesome .. thanks! | 01:51 |
| openstackgerrit | Merged stackforge/akanda-rug: Detangle config option registration https://review.openstack.org/191993 | 01:51 |
| adam_g | dhellmann, if you're interested and have any spare cycles, your feedback would be most appreciated on https://review.openstack.org/195373, which rebases akanda-rug's RPC layer ontop of oslo.messaging | 01:52 |
| adam_g | okay i gotta go marinate my meats. cya | 01:52 |
| sarob | adam_g: okay, thats dirty | 01:52 |
| davidlenwell | mmm meat | 01:52 |
| davidlenwell | not as dirty as that ^ | 01:53 |
| *** puranamr has quit IRC | 02:09 | |
| *** puranamr has joined #akanda | 02:18 | |
| *** puranamr has quit IRC | 02:27 | |
| *** puranamr has joined #akanda | 02:28 | |
| openstackgerrit | Merged stackforge/akanda-rug: Migrate to oslo.log and oslo.messaging https://review.openstack.org/195373 | 02:43 |
| openstackgerrit | Merged stackforge/akanda-rug: Migrate to oslo.i18n https://review.openstack.org/192395 | 03:08 |
| *** puranamr has quit IRC | 03:26 | |
| *** puranamr has joined #akanda | 03:33 | |
| *** puranamr has quit IRC | 03:35 | |
| *** puranamr has joined #akanda | 03:35 | |
| *** puranamr_ has joined #akanda | 03:47 | |
| *** puranamr has quit IRC | 03:47 | |
| *** sarob has quit IRC | 03:57 | |
| openstackgerrit | David Lenwell proposed stackforge/akanda-rug: naming convention change vm to instance https://review.openstack.org/195404 | 04:00 |
| *** puranamr has joined #akanda | 04:21 | |
| *** puranamr_ has quit IRC | 04:21 | |
| openstackgerrit | David Lenwell proposed stackforge/akanda-rug: naming convention change vm to instance https://review.openstack.org/195404 | 04:22 |
| *** puranamr_ has joined #akanda | 04:29 | |
| *** puranamr has quit IRC | 04:29 | |
| *** puranamr_ has quit IRC | 04:34 | |
| *** puranamr has joined #akanda | 04:34 | |
| *** sarob has joined #akanda | 04:44 | |
| *** sarob has quit IRC | 04:47 | |
| *** sarob has joined #akanda | 04:47 | |
| *** puranamr has quit IRC | 04:50 | |
| *** puranamr has joined #akanda | 04:51 | |
| openstackgerrit | David Lenwell proposed stackforge/akanda-rug: naming convention change vm to instance https://review.openstack.org/195404 | 04:55 |
| *** sarob has quit IRC | 04:57 | |
| *** sarob has joined #akanda | 04:57 | |
| openstackgerrit | David Lenwell proposed stackforge/akanda-rug: naming convention change vm to instance https://review.openstack.org/195404 | 05:00 |
| *** puranamr_ has joined #akanda | 05:10 | |
| *** puranamr has quit IRC | 05:13 | |
| *** sarob has quit IRC | 05:13 | |
| *** sarob has joined #akanda | 05:14 | |
| *** sarob has quit IRC | 05:15 | |
| *** puranamr_ has quit IRC | 05:18 | |
| openstackgerrit | David Lenwell proposed stackforge/akanda-rug: naming convention change vm to instance https://review.openstack.org/195404 | 05:23 |
| *** puranamr has joined #akanda | 05:31 | |
| *** puranamr has joined #akanda | 05:33 | |
| *** puranamr has quit IRC | 05:37 | |
| openstackgerrit | David Lenwell proposed stackforge/akanda-rug: Naming convention change router to instance (wip) https://review.openstack.org/195416 | 05:44 |
| davidlenwell | markmcclain: when you are around tomorrow I'd like to talk about this ^^^ before I get too far on it.. and I want to pick your brain about expanding our neutron api wrapper.. | 05:51 |
| *** sarob has joined #akanda | 07:08 | |
| *** sarob has quit IRC | 07:12 | |
| *** rods has joined #akanda | 09:30 | |
| *** sarob has joined #akanda | 10:12 | |
| *** sarob has quit IRC | 10:17 | |
| *** rods has quit IRC | 10:35 | |
| *** rods has joined #akanda | 13:33 | |
| *** rods1 has joined #akanda | 13:37 | |
| *** rods has quit IRC | 13:39 | |
| *** puranamr has joined #akanda | 13:59 | |
| *** puranamr has quit IRC | 14:43 | |
| *** puranamr has joined #akanda | 14:44 | |
| *** puranamr has quit IRC | 14:48 | |
| *** puranamr has joined #akanda | 14:59 | |
| *** puranamr has quit IRC | 15:12 | |
| *** sarob has joined #akanda | 16:22 | |
| *** puranamr has joined #akanda | 16:48 | |
| *** puranamr has quit IRC | 17:33 | |
| *** puranamr has joined #akanda | 17:36 | |
| *** puranamr_ has joined #akanda | 17:40 | |
| *** puranamr has quit IRC | 17:43 | |
| *** puranamr_ has quit IRC | 17:51 | |
| *** puranamr has joined #akanda | 17:51 | |
| ryanpetrello | adam_g: what's preventing e.g., https://review.openstack.org/#/c/195176/ from passing? | 18:08 |
| ryanpetrello | does the devstack gating stuff not work for kilo/juno ? | 18:08 |
| adam_g | ya | 18:08 |
| adam_g | https://review.openstack.org/#/c/195274/ | 18:08 |
| adam_g | disables it on those branches | 18:08 |
| adam_g | so we can fix them up and get them working | 18:08 |
| * ryanpetrello looks | 18:08 | |
| ryanpetrello | okay, cool | 18:08 |
| ryanpetrello | thanks :D | 18:08 |
| adam_g | should be soonish | 18:08 |
| ryanpetrello | I'll recheck these when it passes | 18:09 |
| *** puranamr has quit IRC | 18:13 | |
| *** puranamr has joined #akanda | 18:16 | |
| davidlenwell | hey ryan.. So lbaas and fwaas aren't built on top of neutron routers.. this change I'm starting is just a start in a series of changes that abstract those resources better. | 18:50 |
| davidlenwell | ryanpetrello: ^^ | 18:51 |
| davidlenwell | and if you didn't like that wait till you see what I have in flight for the neutron api wrapper | 18:51 |
| ryanpetrello | heh | 18:57 |
| davidlenwell | if we are to use this code to abstract a bunch of advanced services.. they won't all be neutron routers.. does that make sense to you? | 18:58 |
| davidlenwell | also .. its a wip.. so it wasn't really ready for review yet | 18:58 |
| ryanpetrello | yea, it does | 18:58 |
| ryanpetrello | I just didn't especially love the router_id change | 18:58 |
| ryanpetrello | instance_id made me think of a Nova UUID | 18:58 |
| davidlenwell | well its not final | 18:58 |
| ryanpetrello | which seemed particularly confusing | 18:58 |
| davidlenwell | yeah.. I agree.. I posted that wip so I could have a discussion about what to call it | 18:59 |
| davidlenwell | which if you look at the patch above this one in the dependancy chain you'll see that I changed vm to instance | 19:00 |
| davidlenwell | markmcclain: thoughts on this subject? | 19:01 |
| * markmcclain reads backscroll | 19:02 | |
| davidlenwell | I am thinking nthe change we need starts here.. https://github.com/stackforge/akanda-rug/blob/master/akanda/rug/api/neutron.py | 19:02 |
| davidlenwell | what we have is very router specific.. its used by populate.py and a bunch of other things | 19:03 |
| davidlenwell | but ryanpetrello you are right .. instance doesn't make sense.. thats why I just stopped mid-way and pushed it so we could have a discussion | 19:04 |
| ryanpetrello | okay, makes sense | 19:08 |
| davidlenwell | so I want to stragitize with markmcclain today about how we can do a better job abstracting resources in the neutron wrapper | 19:08 |
| davidlenwell | markmcclain: hand wavey will do to get me started | 19:11 |
| sarob | neutron api here http://developer.openstack.org/api-ref-networking-v2.html | 19:13 |
| sarob | we need to abstract out the router from other neutron objects | 19:18 |
| markmcclain | so over time we'll just need a config hook which tells the rug | 19:27 |
| markmcclain | which logical resources we care about and the drivers that back | 19:28 |
| markmcclain | at one point in time... I had written an object mapper for the neutron api | 19:29 |
| markmcclain | we're probably better served by using the akanda sdk | 19:29 |
| markmcclain | s/akanda sdk/openstack sdk/ | 19:29 |
| markmcclain | davidlenwell: make sense? | 19:34 |
| adam_g | ryanpetrello, im not a fan of using 'instance' to describe our things internally, either | 19:51 |
| adam_g | that naming change patch would be easier swallowed if it updated more than just names, and added the ability to handle things other than routers.. which is where i think davidlenwell is heading. might make more sense with the follow up patch stacked on it? | 19:52 |
| openstackgerrit | Merged stackforge/akanda-rug: naming convention change vm to instance https://review.openstack.org/195404 | 19:58 |
| adam_g | davidlenwell, would it make sense to move neutron client usage out of the VmManager and into driver class, allowing drivers the ability to override, ie update_status() ? | 20:01 |
| *** rods1 has quit IRC | 20:02 | |
| adam_g | then https://git.openstack.org/cgit/stackforge/akanda-rug/tree/akanda/rug/instance_manager.py#n73 | 20:03 |
| adam_g | just turns into something like self.router_obj.update_status(new_status) | 20:03 |
| adam_g | (ignoring worker_context for now) | 20:03 |
| davidlenwell | adam_g: yeah.. I was thinking almost exactly that | 20:06 |
| davidlenwell | sean and I were talking about it before we left for lunch | 20:06 |
| davidlenwell | markmcclain: yes .. that makes senes... I'll sketch something rough up and push it so we can discuss firther | 20:06 |
| davidlenwell | futher | 20:06 |
| openstackgerrit | Adam Gandelman proposed stackforge/akanda-rug: Fix ability to recover from an existing appliance VM https://review.openstack.org/194841 | 20:16 |
| *** sarob_ has joined #akanda | 20:46 | |
| *** puranamr_ has joined #akanda | 20:46 | |
| *** sarob has quit IRC | 20:49 | |
| *** puranamr has quit IRC | 20:49 | |
| openstackgerrit | Adam Gandelman proposed stackforge/akanda: Add spec for RUG HA https://review.openstack.org/193680 | 21:15 |
| *** rods has joined #akanda | 21:34 | |
| *** rods has quit IRC | 21:37 | |
| openstackgerrit | Adam Gandelman proposed stackforge/akanda: Add spec for RUG HA https://review.openstack.org/193680 | 22:42 |
| *** dhellmann has quit IRC | 22:46 | |
| *** dhellmann has joined #akanda | 22:46 | |
| davidlenwell | adam_g: is that ready for review yet or are you still toying with it? | 22:51 |
| adam_g | i htink its ready, updated it based on the poc thing i pushed up yesterday | 22:51 |
| davidlenwell | k | 22:51 |
| davidlenwell | just didn't want to review while you were still patching it | 22:51 |
| adam_g | using a distributed consistent hash ring looks to be much easier than dealing with leader election, RPC, etc | 22:51 |
| adam_g | but i may be overlooking something | 22:52 |
| davidlenwell | So im running with abstracting all of the neutron calls into the driver object and out of instance_manager | 22:52 |
| davidlenwell | but populate.py is a thing too | 22:52 |
| davidlenwell | will need to modit to use drivers | 22:52 |
| davidlenwell | was thinking maybe making it loop through all the available drivers to repopulate and have all that in the drivers as well | 22:54 |
| davidlenwell | because the neutron query to get lb's is a lot different than routers.. which is a lot diff from fwass.. (at least for now until we redesign the fwaas api) | 22:55 |
| adam_g | davidlenwell, yeah | 22:55 |
| adam_g | a static list_all() method or something | 22:57 |
| *** puranamr_ has quit IRC | 23:53 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!