openstackgerrit | ding bo proposed openstack/dragonflow: Configure security data into DF DB Sync neutron security group data into DF DB https://review.openstack.org/262634 | 00:17 |
---|---|---|
*** dingboopt has quit IRC | 04:17 | |
*** gsagie_ has joined #openstack-dragonflow | 04:18 | |
*** gal has joined #openstack-dragonflow | 04:20 | |
*** gsagie_ has quit IRC | 04:43 | |
openstackgerrit | Merged openstack/dragonflow: etcd: Conditionally start/stop etcd server https://review.openstack.org/262593 | 05:15 |
openstackgerrit | Merged openstack/dragonflow: devstack: Cleanup inconsistent spacing https://review.openstack.org/262594 | 05:21 |
*** dingboopt has joined #openstack-dragonflow | 05:39 | |
dingboopt | gsagie_: ok, I will score patch:) | 05:47 |
gsagie | nick-ma: ping | 07:01 |
*** diga has joined #openstack-dragonflow | 07:28 | |
nick-ma | gsagie: pong | 07:29 |
gsagie | nick-ma: regarding the race you noticed, i asked on the bug if you can share the exact scenario you spotted it | 07:36 |
gsagie | because its interesting to understand, as this could be something also problematic to Neutron or to our DB modeling | 07:36 |
nick-ma | https://github.com/openstack/dragonflow/blob/master/dragonflow/db/api_nb.py#L206-L234 | 07:42 |
nick-ma | add_subnet: get_key, update_subnet:get_key, update_subnet:set_key, add_subnet:set_key, then the data of update_subnet will be lost. | 07:44 |
nick-ma | sorry, two update_subnet functions cross each other. one of the data will be lost. | 07:51 |
openstackgerrit | Gal Sagie proposed openstack/dragonflow: Remove ofport=0 Exception as it handled in the applications https://review.openstack.org/262695 | 07:52 |
nick-ma | two concurrently update_subnet for a given subnet-id | 07:52 |
openstackgerrit | Eran Gampel proposed openstack/dragonflow: Remove the dhcp agent from the devstack default configuration https://review.openstack.org/262696 | 07:56 |
openstackgerrit | Eran Gampel proposed openstack/dragonflow: Run etcd as a service in fedora (systemd) https://review.openstack.org/262703 | 08:06 |
*** dingboopt has quit IRC | 08:17 | |
nick-ma | gsagie: I replied in the bug, and provide the scenario for demonstration and a screen snapshot. | 08:19 |
gsagie | nick-ma: thanks! | 08:22 |
gsagie | its very strange that it happens, because the same problem should happen for Neutron, i guess the reason it is not problematic because subnet and network are seperated in Neutron (at least thats what i believe) | 08:23 |
openstackgerrit | shihanzhang proposed openstack/dragonflow: Correct flows for router_interface in compute node https://review.openstack.org/262707 | 08:25 |
shz | hi all, who can help to review this patch | 08:26 |
shz | ping gsagie, nick-ma | 08:26 |
gsagie | shz: i will review, thanks | 08:27 |
gsagie | i think that its not needed as we add this from the L3 App, thats why i removed these lines | 08:27 |
gsagie | do you see L3 not working correctly? | 08:28 |
gsagie | i remember this issue | 08:28 |
shz | gsagie, you can check the flows in the bug description | 08:29 |
shz | in compute node, in flow table 17, it send all traffic to table 64 then to network node | 08:29 |
gsagie | but in controller its ok right? | 08:30 |
gsagie | because i see this working for me | 08:30 |
gsagie | with single node | 08:30 |
shz | just in compute node, the flow is incorrect | 08:31 |
shz | the reason is that lport.get_chassis() == self.chassis_name | 08:31 |
shz | in compute node, the the chassis_name is not equal | 08:32 |
gsagie | ohhh | 08:33 |
gsagie | i see, you added it for remote port as well | 08:33 |
shz | in compute node, the lport for router_interface will be a remote port, not local port | 08:33 |
shz | yes | 08:33 |
gsagie | yes yes i see it now, good job! :) | 08:33 |
gsagie | let me just re-verify | 08:33 |
shz | ok | 08:33 |
shz | for your patch #262695, i think if we just remove the else statement, there is other problem | 08:36 |
shz | I will comment on your patch | 08:36 |
gsagie | shz: ok thanks | 08:37 |
gsagie | shz: +2 your patch, good catch! | 08:37 |
gsagie | nick-ma: thanks for the details, i understand the problem now | 08:39 |
nick-ma | it's ok. i'll think about how to deal with it. | 08:47 |
openstackgerrit | Gal Sagie proposed openstack/dragonflow: Remove ofport=0 Exception as it handled in the applications https://review.openstack.org/262695 | 08:47 |
*** Shlomo_N has joined #openstack-dragonflow | 08:57 | |
*** dingboopt has joined #openstack-dragonflow | 09:00 | |
openstackgerrit | Merged openstack/dragonflow: Remove the dhcp agent from the devstack default configuration https://review.openstack.org/262696 | 09:02 |
openstackgerrit | Eran Gampel proposed openstack/dragonflow: Run etcd as a service in fedora (systemd) https://review.openstack.org/262703 | 09:09 |
gsagie | shz: were you meant to comment on the ofport==0 patch fix? | 09:43 |
*** gampel has joined #openstack-dragonflow | 09:50 | |
*** diga has quit IRC | 10:02 | |
*** diga has joined #openstack-dragonflow | 10:28 | |
openstackgerrit | Merged openstack/dragonflow: Correct flows for router_interface in compute node https://review.openstack.org/262707 | 11:10 |
*** dingboopt has quit IRC | 11:47 | |
gsagie | shz: ping | 11:58 |
*** raofei has quit IRC | 12:40 | |
*** diga has quit IRC | 13:31 | |
openstackgerrit | Kyle Mestery proposed openstack/dragonflow: Vagrant: Add a second compute node https://review.openstack.org/262595 | 14:18 |
openstackgerrit | Kyle Mestery proposed openstack/dragonflow: Move Vagrant configuration into a vagrant directory https://review.openstack.org/262327 | 14:18 |
openstackgerrit | Gal Sagie proposed openstack/dragonflow: Adjust unit tests for Dragonflow https://review.openstack.org/262796 | 14:32 |
openstackgerrit | Kyle Mestery proposed openstack/dragonflow: Vagrant: Add a second compute node https://review.openstack.org/262595 | 14:38 |
openstackgerrit | Kyle Mestery proposed openstack/dragonflow: Move Vagrant configuration into a vagrant directory https://review.openstack.org/262327 | 14:38 |
*** gsagie has quit IRC | 14:44 | |
*** gampel has quit IRC | 15:49 | |
openstackgerrit | Kyle Mestery proposed openstack/dragonflow: Vagrant: Add a second compute node https://review.openstack.org/262595 | 16:56 |
openstackgerrit | Kyle Mestery proposed openstack/dragonflow: Move Vagrant configuration into a vagrant directory https://review.openstack.org/262327 | 16:56 |
openstackgerrit | Kyle Mestery proposed openstack/dragonflow: devstack: Fix configure_df_plugin to work on compute nodes https://review.openstack.org/262812 | 16:56 |
*** openstackgerrit has quit IRC | 20:32 | |
*** openstackgerrit has joined #openstack-dragonflow | 20:33 | |
*** dingboopt has joined #openstack-dragonflow | 23:12 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!