*** yamamoto_ has joined #openstack-dragonflow | 00:00 | |
*** yamamoto_ has quit IRC | 00:35 | |
*** yamamoto_ has joined #openstack-dragonflow | 01:24 | |
*** He__ has joined #openstack-dragonflow | 01:37 | |
*** bochi-michael has joined #openstack-dragonflow | 02:43 | |
*** gongysh has joined #openstack-dragonflow | 02:47 | |
*** gongysh has quit IRC | 03:16 | |
gsagie | Hello everyone! | 05:51 |
---|---|---|
gsagie | :) | 05:51 |
*** gongysh has joined #openstack-dragonflow | 05:54 | |
*** gongysh has quit IRC | 06:13 | |
nick-ma | gsagie: thanks for this latest ppt. | 06:32 |
*** gampel has joined #openstack-dragonflow | 06:37 | |
gsagie | welcome home gampel :) | 06:37 |
gsagie | i assume? | 06:37 |
shihanzhang | hi gsagie, the public network traffic in pipeline table 0 means floatingIP or DNAT? | 06:43 |
gsagie | yes, distributed DNAT | 06:44 |
gsagie | (also for VLAN case for provider networks) | 06:44 |
*** shihanzhang has quit IRC | 06:44 | |
*** shihanzhang has joined #openstack-dragonflow | 06:44 | |
gsagie | shihanzhang: in page 4/4 you can see the "public network ingress part" | 06:45 |
gsagie | as you mentioned its for distributed DNAT | 06:45 |
gsagie | but also for VLAN/provider-networks case | 06:45 |
shihanzhang | got it, thx | 06:47 |
*** gampel1 has joined #openstack-dragonflow | 06:48 | |
*** gampel has quit IRC | 06:51 | |
gampel1 | gsagie: thanks :)Just landed few hours ago | 07:22 |
*** gampel1 has quit IRC | 07:22 | |
*** gampel has joined #openstack-dragonflow | 07:23 | |
*** gongysh has joined #openstack-dragonflow | 08:13 | |
*** gongysh has quit IRC | 08:21 | |
*** gampel has quit IRC | 08:29 | |
*** gongysh has joined #openstack-dragonflow | 08:31 | |
*** gampel has joined #openstack-dragonflow | 08:34 | |
*** gongysh has quit IRC | 08:36 | |
*** bochi-michael has quit IRC | 09:52 | |
*** gampel has quit IRC | 09:55 | |
*** gampel has joined #openstack-dragonflow | 09:56 | |
*** yuli_s has quit IRC | 10:01 | |
*** Shlomo_N has quit IRC | 10:15 | |
*** yamamoto_ has quit IRC | 10:49 | |
*** yamamoto has joined #openstack-dragonflow | 10:50 | |
*** gongysh has joined #openstack-dragonflow | 11:04 | |
*** yamamoto has quit IRC | 11:04 | |
gongysh | gampel, ping | 11:05 |
gongysh | gsagie, ping | 11:06 |
gsagie | gongysh: yes? | 11:06 |
gongysh | I am trying to install the dragon flow in distributed mode. | 11:06 |
gongysh | have some questions want to know | 11:07 |
gsagie | ok | 11:07 |
gongysh | I want to use etcd | 11:07 |
gongysh | so what service to enable on controller node? | 11:08 |
gsagie | gongysh: this example local.conf: https://github.com/openstack/dragonflow/blob/master/doc/source/single-node-conf/local_controller.conf | 11:08 |
gongysh | enable_service df-etcd | 11:08 |
gongysh | enable_service df-etcd-server | 11:08 |
gongysh | enable_service df-controller | 11:08 |
gongysh | enable_service df-ext-services | 11:08 |
gongysh | enable_service q-svc | 11:08 |
gongysh | enable_service q-l3 | 11:08 |
gongysh | what is the difference between q-l3 and q-df-l3? | 11:10 |
gsagie | in the distributed DF we dont have 1-df-l3 anymore | 11:11 |
gsagie | q-df-l3 | 11:11 |
gsagie | thats from when Dragonflow was centralized | 11:11 |
gsagie | we use Neutron l3 agent for the SNAT/DNAT right now until we have a distributed version but thats all | 11:12 |
gongysh | gsagie, in distributed mode, that is no l2 agent running on compute node, right? | 11:23 |
gsagie | right | 11:24 |
gsagie | there are no agents at all in the compute node, just df-controller | 11:24 |
gongysh | gsagie, so https://github.com/openstack/dragonflow/blob/master/devstack/plugin.sh#L433 | 11:24 |
gsagie | gongysh: this shouldnt be called, in the local.conf you need to have this: | 11:25 |
gongysh | gsagie, it is not differentiating the case, AGENT_BINARY="$DF_L2_AGENT" | 11:25 |
gsagie | Q_ENABLE_DRAGONFLOW_LOCAL_CONTROLLER=True | 11:25 |
gsagie | and remove this: | 11:25 |
gsagie | Q_ENABLE_DRAGONFLOW | 11:25 |
gsagie | We do need to fix the naming, i see now its confusing | 11:25 |
gsagie | If you look at the plugin.sh, the code to install the distributed controller is above what you just wrote | 11:26 |
gongysh | gsagie, for distributed mode, Q_ENABLE_DRAGONFLOW_LOCAL_CONTROLLER=True, | 11:26 |
gongysh | gsagie, for centralized mode, Q_ENABLE_DRAGONFLOW=True, right? | 11:26 |
gsagie | yes | 11:26 |
gsagie | we need to change that name :) | 11:27 |
gongysh | cool. | 11:27 |
gongysh | yes, and with better doc. | 11:27 |
gsagie | agreed! will add bug to fix this on me | 11:27 |
gongysh | nice | 11:28 |
gongysh | and also, I think we will not allow to enable both Q_ENABLE_DRAGONFLOW_LOCAL_CONTROLLER and Q_ENABLE_DRAGONFLOW. | 11:29 |
gongysh | at the same time | 11:30 |
gsagie | yes thats right | 11:33 |
gongysh | gsagie, I am seeing the mechnism driver, do we have plan to move from standalone plugin to ml2 mechanism driver? | 11:35 |
gsagie | yes, we are working on writing an ML2 mech driver because there are many use cases that we want to work with other drivers (for smart TORs , SR-IOV and so on..) | 11:35 |
gsagie | we had a big meetup in Hangzhou last week, thats the conclusions from this meetup and from talking with some other providers | 11:36 |
gongysh | yes, I am sorry that I failed to join u on that meetingup. | 11:37 |
gongysh | gsagie, so these codes under https://github.com/openstack/dragonflow/tree/master/dragonflow/neutron/agent are also for centralised mode, right? | 11:39 |
gsagie | no problem :) there is a big group of people that going to join us and work on Dragonflow | 11:39 |
gsagie | gongysh: yes | 11:39 |
gsagie | for the neutron part in the distributed solution we only need the plugin code | 11:39 |
*** openstackgerrit has quit IRC | 11:43 | |
*** openstackgerrit has joined #openstack-dragonflow | 11:44 | |
*** gongysh has quit IRC | 11:46 | |
*** gongysh has joined #openstack-dragonflow | 11:55 | |
gongysh | gsagie, we don't use the service_plugins=xxx to load the l3 plugin, since our core plugin is implementing neutron traditional core plugin and l3 plugin. | 11:55 |
gongysh | gsagie, do we have any other reason not to split the core and l3 plugin? | 11:56 |
gsagie | gongysh: we will do this as we move to ML2 | 11:59 |
gsagie | but no real reason, we did this now for simplicty | 11:59 |
gongysh | got it. | 12:03 |
*** gongysh has quit IRC | 12:07 | |
*** yamamoto has joined #openstack-dragonflow | 12:28 | |
*** yamamoto has quit IRC | 12:29 | |
*** yamamoto_ has joined #openstack-dragonflow | 12:29 | |
*** yamamoto has joined #openstack-dragonflow | 12:31 | |
*** yamamoto_ has quit IRC | 12:34 | |
openstackgerrit | Merged openstack/dragonflow: Not allow delete subnet gateway port https://review.openstack.org/264508 | 12:58 |
*** yamamoto has quit IRC | 13:59 | |
*** yamamoto has joined #openstack-dragonflow | 14:03 | |
openstackgerrit | Yuli proposed openstack/dragonflow: Add new ovs tests https://review.openstack.org/265642 | 14:16 |
*** yuli_s has joined #openstack-dragonflow | 14:31 | |
*** yamamoto has quit IRC | 14:33 | |
*** yamamoto has joined #openstack-dragonflow | 14:39 | |
*** yuli_s has quit IRC | 15:19 | |
*** yamamoto has quit IRC | 15:42 | |
*** yamamoto has joined #openstack-dragonflow | 15:42 | |
*** wangfeng_yellow has quit IRC | 16:11 | |
*** wangfeng_yellow has joined #openstack-dragonflow | 16:11 | |
*** yamamoto has quit IRC | 16:33 | |
*** dingboopt_ has joined #openstack-dragonflow | 18:18 | |
*** gampel1 has joined #openstack-dragonflow | 18:18 | |
*** gampel has quit IRC | 18:22 | |
*** dingboopt has quit IRC | 18:22 | |
*** dingboopt_ is now known as dingboopt | 18:22 | |
*** wangfeng_yellow has quit IRC | 18:58 | |
*** wangfeng_yellow has joined #openstack-dragonflow | 18:58 | |
*** dingboopt has quit IRC | 18:59 | |
*** dingboopt has joined #openstack-dragonflow | 18:59 | |
*** gongysh has joined #openstack-dragonflow | 20:18 | |
*** gongysh has quit IRC | 20:23 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!