| shettyg | russellb: Had a quick look on why multi-host ping does not work with tip of ovn branch. Looks like ovn-controller does not add the correct table17's broadcast flow. It misses replicating packet for the logical port on a different host. If you are already looking at fixing this, then I won't spend more time. If you are not, I can look at it tomorrow. | 00:45 |
|---|---|---|
| russellb | shettyg: hello! | 00:46 |
| russellb | shettyg: i got pretty sidetracked today by openstack CI problems ... i was just now back to debugging it while watching tv :) | 00:46 |
| russellb | i just about 5 minutes ago figured out that arp wasn't working, so sounds like i have the same problem, but you narrowed it down further | 00:47 |
| russellb | shettyg: i can try to look at fixing it tomorrow ... i'll send you an email or something if i make any decent progress, otherwise i would assume i haven't figured it out yet | 00:48 |
| shettyg | Okay. cool | 00:48 |
| russellb | shettyg: thanks a lot for the pointer! | 00:49 |
| *** shettyg has quit IRC | 00:52 | |
| *** mestery has quit IRC | 01:45 | |
| *** smallbig has left #openstack-neutron-ovn | 06:50 | |
| *** smallbig has joined #openstack-neutron-ovn | 09:34 | |
| smallbig | Hellow | 09:34 |
| smallbig | Hello | 09:35 |
| smallbig | if I found http://paste.openstack.org/show/221660/ this message from neutron-openvswitch-agent.log, where should I dig for furthur information, thanks! | 09:36 |
| ajo | gsagie, can I ask you for some desperate help? :) | 11:05 |
| ajo | do you know what [ 7786.693741] openvswitch: netlink: VLAN tag present bit must have an exact match (tci_mask=1). | 11:05 |
| ajo | [root@ovs-compute-2 ~]# means? | 11:05 |
| ajo | russellb ^ | 11:10 |
| ajo | :) | 11:10 |
| ajo | I'm trying to debug my flow rules for security groups | 11:10 |
| ajo | having some issues with cross compute node traffic | 11:10 |
| *** mestery has joined #openstack-neutron-ovn | 12:44 | |
| mestery | russellb: How goes multi-host with OVN? Catching up on things, I've been traveling this week (not done yet) | 13:36 |
| russellb | mestery: o/ | 13:37 |
| russellb | mestery: coming along ... OVN is currently busted for that use case, and while debugging it yesterday i got majorly sidetracked unwedging the gate | 13:37 |
| mestery | Yeah, sounds like the gate had some issues. Thanks for working on that! | 13:38 |
| russellb | mestery: i think networking-ovn and its devstack plugin are in good shape, just need to fix OVN, and finish docs (and my blog post) | 13:38 |
| russellb | np | 13:38 |
| mestery | Nice! | 13:38 |
| russellb | in good shape using what's up for review at least | 13:38 |
| mestery | russellb: Is the gate ok at the moment, or are the issues from the client releases still causing headaches? | 13:38 |
| russellb | i have a patch series pending | 13:38 |
| mestery | I saw the patch series right now in fact :) | 13:38 |
| russellb | mestery: it's all sorted now. the issues i was working through were actually a neutron problem | 13:39 |
| russellb | mestery: the lbaas job that was voting against neutron was actually testing against neutron master, not the proposed patch | 13:39 |
| mestery | Oh wow! Thanks for helping there! | 13:39 |
| russellb | and that allowed neutron to merge a patch that broke the job | 13:39 |
| mestery | oh no! :( | 13:39 |
| mestery | wow | 13:39 |
| russellb | and that same problem also broke all the networking-ovn jobs (and networking-odl jobs, and some others) | 13:39 |
| mestery | I got an email from dougwig and blogan, they are looking at fixing that job I guess | 13:39 |
| russellb | basically anything that had neutron.git in requirements.txt | 13:39 |
| russellb | yeah | 13:40 |
| mestery | sucks | 13:40 |
| mestery | Man | 13:40 |
| russellb | i had to make the lbaas job non-voting to unwedge neutron | 13:40 |
| russellb | so that we could merge the fix | 13:40 |
| mestery | OK, cool | 13:40 |
| mestery | yup | 13:40 |
| russellb | anyway, all sorted now i think | 13:40 |
| russellb | except that job needs work | 13:40 |
| mestery | I assume dougwig and blogan will take care of fixing the lbaas job | 13:40 |
| mestery | Do we also need to fix the ODL and OVN jobs? | 13:40 |
| russellb | yeah, sounds like it, cool | 13:40 |
| russellb | no, they should all be fixed now by the neutron fix that merged | 13:40 |
| russellb | just rechecks | 13:41 |
| mestery | Cool | 13:41 |
| mestery | Awesome | 13:41 |
| russellb | so that was my fun yesterday :) | 13:41 |
| russellb | now back to figuring out the OVN multi-node bug | 13:41 |
| mestery | cool! | 13:41 |
| russellb | aiming to have a detailed blog post by the end of the week that goes through a 2 node setup in some detail | 13:41 |
| russellb | if i can get it working in time... | 13:41 |
| mestery | Very cool! | 13:42 |
| mestery | Heh, well, we can use that for the demo during the presentation too! :) | 13:42 |
| russellb | yeah, i'm sure we'll have everything working fine | 13:44 |
| russellb | it's more of a question of how much writing time i have | 13:44 |
| russellb | because next week i'm not writing, i'm having beers. | 13:46 |
| mestery | +1000! | 13:47 |
| ajo | gsagie -> #openstack-meeting-3 / QoS | 14:01 |
| ajo | mestery, we're going to plan the session a bit during the meeting ;) | 14:02 |
| mestery | ajo: Awesome work my friend! Thanks for doing an awesome job with QoS :) | 14:02 |
| ajo | mestery, thanks, I'm just trying ':) | 14:04 |
| russellb | ajo: cool to see you're testing security groups :) | 14:08 |
| *** shettyg has joined #openstack-neutron-ovn | 14:11 | |
| ajo | russellb, I'm testing it in the openvswitch_firewall.py I'm working on | 14:27 |
| ajo | still no OVN, is OVN starting to support security group stuff? :) | 14:27 |
| ajo | I can give a lot of input based on my experiments ;) | 14:27 |
| ajo | (for neutron parity) | 14:27 |
| russellb | ajo: not yet | 14:28 |
| russellb | ajo: we have a document in networking-ovn talking about how we'll translate neutron security groups into OVN ACL entries | 14:28 |
| russellb | but it's not implemented in OVN yet | 14:28 |
| *** mestery has quit IRC | 14:44 | |
| russellb | well, tempest runs now ... http://logs.openstack.org/80/181480/3/check/check-tempest-dsvm-networking-ovn/7ea04b7/console.html | 14:49 |
| russellb | it blows up, but it runs | 14:49 |
| openstackgerrit | Russell Bryant proposed stackforge/networking-ovn: devstack: Document adding a second compute node https://review.openstack.org/181602 | 14:54 |
| openstackgerrit | Russell Bryant proposed stackforge/networking-ovn: devstack: Ensure openvswitch modules are loaded https://review.openstack.org/182063 | 14:54 |
| ajo | russellb, one step at a time :D | 14:56 |
| ajo | russellb, yes, I remember the ACLs :) | 14:56 |
| ajo | connection tracking, and port ranges shall close a lot of gaps in that regard | 14:56 |
| russellb | http://git.openstack.org/cgit/stackforge/networking-ovn/tree/doc/source/data_model.rst#n131 | 14:56 |
| ajo | russellb, may be you should start thinking about QoS too ;D | 14:57 |
| russellb | ajo: if ovs can do it, sure :) | 14:57 |
| ajo | russellb, but better when we have that totally defined | 14:57 |
| ajo | russellb, it does :D | 14:57 |
| * russellb learning ovs as i go on this ... | 14:57 | |
| ajo | -> https://etherpad.openstack.org/p/YVR-neutron-qos | 14:58 |
| russellb | cool | 14:58 |
| ajo | -> https://review.openstack.org/#/c/182349/1/specs/liberty/ml2-ovs-qos-with-bwlimiting.rst | 14:58 |
| ajo | :-) | 14:58 |
| russellb | If anyone has things they would like to discuss in our design summit session next week, here's the page: https://etherpad.openstack.org/p/YVR-neutron-ovn | 15:05 |
| otherwiseguy | russellb: http://openvswitch.org/pipermail/dev/2015-May/055202.html a few py3 fixes, should be enough to at least get it to build/install/make py34 test not die at setup. :p | 16:47 |
| *** shettyg has quit IRC | 16:48 | |
| *** shettyg has joined #openstack-neutron-ovn | 17:03 | |
| russellb | otherwiseguy: acked | 17:25 |
| *** ajo has quit IRC | 18:10 | |
| openstackgerrit | Russell Bryant proposed stackforge/networking-ovn: devstack: Add multi-node support https://review.openstack.org/181502 | 18:11 |
| *** shettyg has quit IRC | 18:46 | |
| *** yapeng has joined #openstack-neutron-ovn | 18:56 | |
| *** yapeng has quit IRC | 19:02 | |
| *** shettyg has joined #openstack-neutron-ovn | 19:26 | |
| openstackgerrit | Russell Bryant proposed stackforge/networking-ovn: devstack: Add multi-node support https://review.openstack.org/181502 | 19:34 |
| *** ajo has joined #openstack-neutron-ovn | 20:47 | |
| *** ajo has quit IRC | 21:58 | |
| *** shettyg has quit IRC | 22:09 | |
| *** openstackgerrit has quit IRC | 22:47 | |
| *** openstackgerrit has joined #openstack-neutron-ovn | 22:49 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!