| *** iwamoto has joined #openstack-neutron | 00:00 | |
| *** clev has quit IRC | 00:00 | |
| *** clev has joined #openstack-neutron | 00:05 | |
| *** suresh12 has quit IRC | 00:08 | |
| *** matsuhashi has joined #openstack-neutron | 00:09 | |
| *** clev has quit IRC | 00:16 | |
| *** sweston has quit IRC | 00:23 | |
| *** matsuhashi has quit IRC | 00:33 | |
| *** suresh12 has joined #openstack-neutron | 00:39 | |
| *** matsuhashi has joined #openstack-neutron | 00:48 | |
| *** suresh12 has quit IRC | 00:48 | |
| *** clev has joined #openstack-neutron | 00:49 | |
| *** WackoRobie has quit IRC | 00:55 | |
| openstackgerrit | Arata Notsu proposed a change to openstack/neutron: Fix ValueError in ip_lib.IpRouteCommand.get_gateway() https://review.openstack.org/65917 | 01:02 |
|---|---|---|
| *** nati_ueno has joined #openstack-neutron | 01:06 | |
| *** clev has quit IRC | 01:09 | |
| openstackgerrit | Hemanth Ravi proposed a change to openstack/neutron: One Convergence Neutron Plugin Implementation. https://review.openstack.org/69246 | 01:09 |
| *** dave_tucker is now known as dave_tucker_zzz | 01:19 | |
| *** nati_ueno has quit IRC | 01:23 | |
| *** markwash has joined #openstack-neutron | 01:24 | |
| *** suresh12 has joined #openstack-neutron | 01:29 | |
| *** markwash has quit IRC | 01:31 | |
| *** suresh12 has quit IRC | 01:33 | |
| *** markwash has joined #openstack-neutron | 01:34 | |
| *** WackoRobie has joined #openstack-neutron | 01:37 | |
| *** Jianyong has joined #openstack-neutron | 01:37 | |
| *** WackoRobie has quit IRC | 01:41 | |
| *** dave_tucker_zzz is now known as dave_tucker | 01:56 | |
| *** dave_tucker is now known as dave_tucker_zzz | 02:07 | |
| *** WackoRobie has joined #openstack-neutron | 02:07 | |
| *** aymenfrikha has quit IRC | 02:10 | |
| *** WackoRobie has quit IRC | 02:12 | |
| *** xuhanp has joined #openstack-neutron | 02:15 | |
| *** xianghui has joined #openstack-neutron | 02:17 | |
| *** dave_tucker_zzz is now known as dave_tucker | 02:23 | |
| *** matsuhashi has quit IRC | 02:26 | |
| *** AlexF_ has joined #openstack-neutron | 02:26 | |
| *** markmcclain has joined #openstack-neutron | 02:31 | |
| *** matsuhashi has joined #openstack-neutron | 02:32 | |
| *** AlexF_ has quit IRC | 02:35 | |
| *** WackoRobie has joined #openstack-neutron | 02:39 | |
| *** WackoRobie has quit IRC | 02:44 | |
| *** dims has quit IRC | 02:50 | |
| *** emagana has joined #openstack-neutron | 02:59 | |
| *** thuc has joined #openstack-neutron | 02:59 | |
| *** zhhuabj has quit IRC | 03:04 | |
| *** xianghui has quit IRC | 03:08 | |
| *** thuc_ has joined #openstack-neutron | 03:11 | |
| *** thuc has quit IRC | 03:14 | |
| *** WackoRobie has joined #openstack-neutron | 03:16 | |
| *** matsuhashi has quit IRC | 03:18 | |
| *** xianghui has joined #openstack-neutron | 03:25 | |
| *** AlexF_ has joined #openstack-neutron | 03:30 | |
| *** AlexF_ has quit IRC | 03:31 | |
| openstackgerrit | ZhiQiang Fan proposed a change to openstack/python-neutronclient: Remove cliff-tablib from test-requirements https://review.openstack.org/49886 | 03:58 |
| *** markmcclain has quit IRC | 04:05 | |
| *** alex_klimov has joined #openstack-neutron | 04:06 | |
| *** WackoRobie has quit IRC | 04:09 | |
| *** alex_klimov has quit IRC | 04:09 | |
| *** dave_tucker is now known as dave_tucker_zzz | 04:09 | |
| *** chandankumar_ has joined #openstack-neutron | 04:10 | |
| *** suresh12 has joined #openstack-neutron | 04:16 | |
| *** amotoki has quit IRC | 04:24 | |
| *** chandankumar_ has quit IRC | 04:24 | |
| *** amotoki has joined #openstack-neutron | 04:25 | |
| *** ramishra has joined #openstack-neutron | 04:28 | |
| *** chandankumar_ has joined #openstack-neutron | 04:29 | |
| openstackgerrit | Mohammad Banikazemi proposed a change to openstack/neutron: Adds the new IBM SDN-VE plugin https://review.openstack.org/66453 | 04:36 |
| *** ramishra has quit IRC | 04:39 | |
| *** matsuhashi has joined #openstack-neutron | 04:40 | |
| *** thedodd has joined #openstack-neutron | 04:49 | |
| *** thuc has joined #openstack-neutron | 04:49 | |
| *** thuc_ has quit IRC | 04:53 | |
| *** chandankumar_ has quit IRC | 04:53 | |
| *** thuc has quit IRC | 04:53 | |
| *** thedodd has quit IRC | 04:54 | |
| *** aymenfrikha has joined #openstack-neutron | 05:00 | |
| *** ramishra has joined #openstack-neutron | 05:05 | |
| *** chandankumar_ has joined #openstack-neutron | 05:23 | |
| *** markwash has quit IRC | 05:26 | |
| *** markwash has joined #openstack-neutron | 05:30 | |
| *** ramishra has quit IRC | 05:51 | |
| *** chandankumar_ has quit IRC | 05:52 | |
| *** ramishra has joined #openstack-neutron | 05:57 | |
| *** ykaneko has joined #openstack-neutron | 06:00 | |
| *** suresh12 has quit IRC | 06:07 | |
| *** xianghui has quit IRC | 06:11 | |
| *** aymenfrikha has quit IRC | 06:17 | |
| *** networkstatic is now known as network_gone2Mcd | 06:20 | |
| openstackgerrit | Jenkins proposed a change to openstack/neutron: Imported Translations from Transifex https://review.openstack.org/66501 | 06:20 |
| *** markwash has quit IRC | 06:23 | |
| *** Jianyong has quit IRC | 06:24 | |
| *** xianghui has joined #openstack-neutron | 06:25 | |
| *** xianghui has quit IRC | 06:32 | |
| *** chandankumar_ has joined #openstack-neutron | 06:38 | |
| *** irenab has joined #openstack-neutron | 06:38 | |
| *** alex_klimov has joined #openstack-neutron | 06:41 | |
| *** chandankumar_ has quit IRC | 06:46 | |
| *** alex_klimov has left #openstack-neutron | 06:55 | |
| *** xianghui has joined #openstack-neutron | 06:56 | |
| *** ramishra has quit IRC | 07:00 | |
| openstackgerrit | Kevin Benton proposed a change to openstack/neutron: Base ML2 bulk support on the loaded drivers https://review.openstack.org/68996 | 07:03 |
| *** nati_ueno has joined #openstack-neutron | 07:06 | |
| *** network_gone2Mcd is now known as networkstatic | 07:08 | |
| *** nati_ueno has quit IRC | 07:13 | |
| openstackgerrit | Kevin Benton proposed a change to openstack/neutron: Enables BigSwitch/Restproxy ML2 VLAN driver https://review.openstack.org/64944 | 07:16 |
| openstackgerrit | Xiaolin Zhang proposed a change to openstack/neutron: Adds https support for metadata agent https://review.openstack.org/67181 | 07:23 |
| *** yamahata has joined #openstack-neutron | 07:24 | |
| *** itzikb has joined #openstack-neutron | 07:25 | |
| *** Jianyong has joined #openstack-neutron | 07:26 | |
| *** afazekas has joined #openstack-neutron | 07:32 | |
| *** AlexF_ has joined #openstack-neutron | 07:36 | |
| ajo | I'd like to improve the netns_cleanup_util https://bugs.launchpad.net/neutron/+bug/1273095 | 07:51 |
| ajo | https://github.com/openstack/neutron/blob/master/neutron/agent/netns_cleanup_util.py | 07:51 |
| *** BBG_Stephen has joined #openstack-neutron | 07:51 | |
| ajo | could anybody confirm this "bug" , I mean, I confirm it's like that, can actually be seen in the code: https://github.com/openstack/neutron/blob/master/neutron/agent/netns_cleanup_util.py | 07:51 |
| *** iwamoto has quit IRC | 08:00 | |
| *** rohit404 has joined #openstack-neutron | 08:05 | |
| openstackgerrit | Yohei Matsuhashi proposed a change to openstack/python-neutronclient: Enable to select specific network service type https://review.openstack.org/54534 | 08:11 |
| *** safchain has joined #openstack-neutron | 08:14 | |
| *** luqas has joined #openstack-neutron | 08:17 | |
| *** jlibosva has joined #openstack-neutron | 08:18 | |
| *** amuller has joined #openstack-neutron | 08:19 | |
| *** Mierdin has quit IRC | 08:25 | |
| openstackgerrit | Édouard Thuleau proposed a change to openstack/neutron: Implement local ARP responder onto OVS agent https://review.openstack.org/49227 | 08:27 |
| *** anand has joined #openstack-neutron | 08:28 | |
| *** lari_ has quit IRC | 08:28 | |
| *** lari_ has joined #openstack-neutron | 08:29 | |
| *** AlexF_ has quit IRC | 08:30 | |
| *** matsuhashi has quit IRC | 08:36 | |
| *** matsuhashi has joined #openstack-neutron | 08:37 | |
| openstackgerrit | Édouard Thuleau proposed a change to openstack/neutron: OVS lib defer apply doesn't handle concurrency https://review.openstack.org/63917 | 08:43 |
| *** roeyc has joined #openstack-neutron | 08:47 | |
| *** lari_ has quit IRC | 08:48 | |
| Qlawy | Hi, I have binding_failed for dhcp assigned port on neutron | 08:48 |
| *** jistr has joined #openstack-neutron | 08:48 | |
| Qlawy | debug logging show nothing :/ | 08:49 |
| Qlawy | NT-EB3E7F5 Device 9c49c453-246a-4ec3-b183-63ba90885050 requested by agent ovsb64db7a65b46 on network 17dd358c-6372-4f88-8061-bab9906e88e7 not bound, vif_type: binding_failed | 08:49 |
| *** ygbo has joined #openstack-neutron | 08:53 | |
| *** lari_ has joined #openstack-neutron | 08:55 | |
| *** jpich has joined #openstack-neutron | 08:59 | |
| *** jlibosva has quit IRC | 08:59 | |
| *** jlibosva has joined #openstack-neutron | 09:01 | |
| *** xianghui has quit IRC | 09:03 | |
| ykaneko | Hello, could anyone review this patch? https://review.openstack.org/#/c/67652/ | 09:08 |
| *** dave_tucker_zzz is now known as dave_tucker | 09:09 | |
| openstackgerrit | Sascha Peilicke proposed a change to openstack/neutron: Don't document non-existing flag '--hide-elapsed' https://review.openstack.org/67477 | 09:12 |
| openstackgerrit | Sascha Peilicke proposed a change to openstack/neutron: Support passing 'neutron_insecure' to neutronclient https://review.openstack.org/65696 | 09:13 |
| *** ihrachys has quit IRC | 09:13 | |
| *** matrohon has joined #openstack-neutron | 09:15 | |
| *** xianghui has joined #openstack-neutron | 09:17 | |
| *** rossella_s has joined #openstack-neutron | 09:19 | |
| *** sweston has joined #openstack-neutron | 09:26 | |
| *** sweston has quit IRC | 09:26 | |
| *** sweston has joined #openstack-neutron | 09:29 | |
| *** sweston has quit IRC | 09:30 | |
| *** jlibosva has quit IRC | 09:31 | |
| *** sweston has joined #openstack-neutron | 09:31 | |
| *** sweston has quit IRC | 09:32 | |
| *** sweston has joined #openstack-neutron | 09:33 | |
| *** sweston has quit IRC | 09:33 | |
| *** sweston has joined #openstack-neutron | 09:34 | |
| *** rkukura has quit IRC | 09:34 | |
| *** sweston has quit IRC | 09:36 | |
| *** sweston has joined #openstack-neutron | 09:37 | |
| *** rkukura has joined #openstack-neutron | 09:40 | |
| *** bashok has joined #openstack-neutron | 09:42 | |
| *** jlibosva has joined #openstack-neutron | 09:46 | |
| *** dave_tucker is now known as dave_tucker_zzz | 09:51 | |
| *** sweston has quit IRC | 09:51 | |
| *** afazekas has quit IRC | 09:52 | |
| *** luqas has quit IRC | 09:52 | |
| *** dave_tucker_zzz is now known as dave_tucker | 09:54 | |
| *** jp_at_hp has joined #openstack-neutron | 09:56 | |
| *** sweston has joined #openstack-neutron | 09:57 | |
| *** ihrachys has joined #openstack-neutron | 09:59 | |
| *** rohit404 has quit IRC | 10:10 | |
| ajo | amotoki, ping | 10:11 |
| ajo | https://review.openstack.org/21489 , https://bugs.launchpad.net/neutron/+bug/1115999 | 10:11 |
| amotoki | ajo: pong | 10:11 |
| ajo | I was looking into this, do you remind more or less the state of the review?, there was some stopper? could I go on with it? | 10:12 |
| *** zzelle has joined #openstack-neutron | 10:13 | |
| *** chandankumar_ has joined #openstack-neutron | 10:13 | |
| ajo | btw, I understand the changes in : https://review.openstack.org/#/c/21489/14/quantum/agent/netns_cleanup_util.py | 10:14 |
| *** afazekas has joined #openstack-neutron | 10:15 | |
| ajo | but not the other changes like this: https://review.openstack.org/#/c/21489/14/quantum/agent/linux/external_process.py (may be they come from a rebase, or they got mixed, or I'm missing something important) | 10:15 |
| ajo | amotoki, ^ (sorry, I forgot to prefix you on the lines above) | 10:16 |
| amotoki | ajo: i think the progress of the patch is stopped now. isaku | 10:17 |
| ajo | Do you mind if I restart it myself? , | 10:17 |
| amotoki | is still working on neutron. it is better to ask him | 10:17 |
| ajo | oh, sorry | 10:17 |
| amotoki | i am not the author of the patch. isaku | 10:18 |
| ajo | somehow I confused, and I thought it was you the commiter | 10:18 |
| amotoki | isaku is the author. He changed companies last year. | 10:18 |
| ajo | Ah, ok I saw you on the bug | 10:18 |
| ajo | and somehow I assumed you where the author | 10:18 |
| amotoki | :-) | 10:19 |
| ajo | ok, I will start this patch again | 10:19 |
| ajo | I think I will reuse his work, but take a simpler approach as an start | 10:19 |
| amotoki | yamahata: around? ajo and i are talking about your abandoned patch. | 10:19 |
| ajo | yamahata_ ^ | 10:20 |
| ajo | oh, it's yamahata , thanks amotoki | 10:20 |
| amotoki | I am not sure what timezone he is now. sometimes japan, sometime us. | 10:21 |
| ajo | bad time for US, and a little bit late for Japan ':D | 10:21 |
| *** luqas has joined #openstack-neutron | 10:34 | |
| *** chandankumar_ has quit IRC | 10:37 | |
| *** djoreilly has joined #openstack-neutron | 10:42 | |
| *** djoreilly_ has joined #openstack-neutron | 10:43 | |
| *** mflobo has joined #openstack-neutron | 10:51 | |
| *** mflobo has quit IRC | 10:52 | |
| *** mflobo has joined #openstack-neutron | 10:52 | |
| *** alexpilotti has joined #openstack-neutron | 10:58 | |
| *** chandankumar_ has joined #openstack-neutron | 11:02 | |
| *** alexpilotti has quit IRC | 11:03 | |
| *** chandankumar_ has quit IRC | 11:08 | |
| *** roeyc has quit IRC | 11:11 | |
| *** chandankumar_ has joined #openstack-neutron | 11:15 | |
| *** alexpilotti has joined #openstack-neutron | 11:25 | |
| *** sweston has quit IRC | 11:25 | |
| *** xianghui has quit IRC | 11:29 | |
| openstackgerrit | Xiaolin Zhang proposed a change to openstack/neutron: Adds https support for metadata agent https://review.openstack.org/67181 | 11:29 |
| *** pcm_ has joined #openstack-neutron | 11:34 | |
| *** pcm_ has quit IRC | 11:35 | |
| *** pcm_ has joined #openstack-neutron | 11:35 | |
| *** chandankumar_ has quit IRC | 11:39 | |
| *** matsuhashi has quit IRC | 11:40 | |
| *** emagana has quit IRC | 11:41 | |
| *** matsuhashi has joined #openstack-neutron | 11:45 | |
| *** djoreilly has left #openstack-neutron | 11:47 | |
| *** djoreilly_ has quit IRC | 11:48 | |
| *** itzikb has quit IRC | 11:50 | |
| yamahata | amotoki, ajo hello? | 12:03 |
| ajo | hi yamahata , I dropped you an email :) | 12:03 |
| ajo | I'm entering a meeting right now | 12:03 |
| ajo | It's just to check the status of the patch above, and ask you if I could go on with the patch | 12:04 |
| *** chandankumar_ has joined #openstack-neutron | 12:05 | |
| yamahata | ajo, feel free to do so. | 12:06 |
| yamahata | ajo, I'm willing to revive myself if you hasn't started it. | 12:07 |
| *** chandankumar_ has quit IRC | 12:08 | |
| *** amarao has joined #openstack-neutron | 12:10 | |
| *** matsuhashi has quit IRC | 12:12 | |
| *** WackoRobie has joined #openstack-neutron | 12:12 | |
| *** amotoki has quit IRC | 12:12 | |
| *** matsuhashi has joined #openstack-neutron | 12:13 | |
| amarao | As far as I can see, neutron has no idea about availability zones? Is any way to partition neworking nodes except by the different networks? | 12:13 |
| *** ykaneko has quit IRC | 12:21 | |
| *** markwash has joined #openstack-neutron | 12:21 | |
| *** dims has joined #openstack-neutron | 12:23 | |
| *** WackoRobie has quit IRC | 12:26 | |
| *** luqas has quit IRC | 12:40 | |
| ajo | yamahata, I didn't start | 12:40 |
| ajo | I was looking into the same problem and I just found your commit, so if you can revive it and rebased, it would be awesome :) | 12:41 |
| ajo | you probably understand the problem much better than me at this moment. | 12:41 |
| *** emagana has joined #openstack-neutron | 12:41 | |
| *** ewindisch has quit IRC | 12:46 | |
| *** salv-orlando has joined #openstack-neutron | 12:48 | |
| *** emagana has quit IRC | 12:50 | |
| *** heyongli has joined #openstack-neutron | 12:52 | |
| *** chandankumar_ has joined #openstack-neutron | 12:54 | |
| *** xianghui has joined #openstack-neutron | 12:57 | |
| *** ijw has joined #openstack-neutron | 12:59 | |
| *** ijw has quit IRC | 13:00 | |
| *** ijw has joined #openstack-neutron | 13:00 | |
| openstackgerrit | Salvatore Orlando proposed a change to openstack/neutron: Test do not review https://review.openstack.org/69365 | 13:12 |
| *** ewindisch has joined #openstack-neutron | 13:14 | |
| *** rossella_s has quit IRC | 13:16 | |
| *** vkozhukalov has joined #openstack-neutron | 13:20 | |
| *** ijw_ has joined #openstack-neutron | 13:21 | |
| *** chandankumar_ has quit IRC | 13:22 | |
| *** chandankumar_ has joined #openstack-neutron | 13:23 | |
| *** markwash has quit IRC | 13:24 | |
| *** ijw has quit IRC | 13:24 | |
| *** ijw_ is now known as ijw | 13:25 | |
| ajo | yamahata, can you revive https://review.openstack.org/21489 then? :-) | 13:29 |
| *** rossella_s has joined #openstack-neutron | 13:29 | |
| *** WackoRobie has joined #openstack-neutron | 13:30 | |
| *** ijw_ has joined #openstack-neutron | 13:32 | |
| *** MichielHN has quit IRC | 13:33 | |
| *** ijw has quit IRC | 13:33 | |
| *** jecarey has quit IRC | 13:34 | |
| *** ijw_ is now known as ijw | 13:35 | |
| *** luqas has joined #openstack-neutron | 13:37 | |
| *** b3nt_pin is now known as beagles | 13:40 | |
| yamahata | ajo, sure will do tomorrow. | 13:43 |
| *** chandankumar_ has quit IRC | 13:44 | |
| *** Jianyong has quit IRC | 13:45 | |
| *** markwash has joined #openstack-neutron | 13:48 | |
| ajo | great, thank you very much, just let me know if you need a hand or anythin | 13:50 |
| *** mattymo has joined #openstack-neutron | 13:57 | |
| mattymo | Does anyone have a recommendation for neutron quotas? | 13:57 |
| *** thuc has joined #openstack-neutron | 13:57 | |
| *** amuller_ has joined #openstack-neutron | 13:59 | |
| *** amuller has quit IRC | 14:00 | |
| *** heyongli has quit IRC | 14:01 | |
| *** ijw has quit IRC | 14:03 | |
| *** ijw_ has joined #openstack-neutron | 14:03 | |
| *** rohit404 has joined #openstack-neutron | 14:04 | |
| *** morganfainberg|z is now known as morganfainberg | 14:06 | |
| *** alagalah has quit IRC | 14:09 | |
| *** thuc_ has joined #openstack-neutron | 14:09 | |
| *** rustlebee is now known as russellb | 14:12 | |
| *** thuc has quit IRC | 14:12 | |
| *** dims has quit IRC | 14:13 | |
| *** ijw_ has quit IRC | 14:14 | |
| *** aymenfrikha has joined #openstack-neutron | 14:15 | |
| *** dims has joined #openstack-neutron | 14:15 | |
| *** ijw has joined #openstack-neutron | 14:16 | |
| *** ijw has quit IRC | 14:17 | |
| *** ijw has joined #openstack-neutron | 14:17 | |
| *** chandankumar_ has joined #openstack-neutron | 14:23 | |
| *** jecarey has joined #openstack-neutron | 14:24 | |
| *** djoreilly has joined #openstack-neutron | 14:24 | |
| *** jgrimm has quit IRC | 14:26 | |
| *** chandankumar_ has quit IRC | 14:27 | |
| *** rohit404 has quit IRC | 14:27 | |
| *** irenab has quit IRC | 14:28 | |
| *** chandankumar_ has joined #openstack-neutron | 14:29 | |
| openstackgerrit | Andres Buraschi proposed a change to openstack/python-neutronclient: Adding weight column to Neutron lb member list CLI https://review.openstack.org/65766 | 14:29 |
| *** thuc_ has quit IRC | 14:29 | |
| *** thuc has joined #openstack-neutron | 14:30 | |
| *** julim has joined #openstack-neutron | 14:31 | |
| *** rohit404 has joined #openstack-neutron | 14:31 | |
| *** julim has quit IRC | 14:32 | |
| *** chandankumar_ has quit IRC | 14:33 | |
| *** chandankumar_ has joined #openstack-neutron | 14:34 | |
| *** thuc has quit IRC | 14:35 | |
| *** julim has joined #openstack-neutron | 14:35 | |
| *** anand has quit IRC | 14:37 | |
| *** chandankumar_ has quit IRC | 14:44 | |
| *** peristeri has joined #openstack-neutron | 14:47 | |
| *** thuc has joined #openstack-neutron | 14:52 | |
| *** chandankumar_ has joined #openstack-neutron | 14:56 | |
| *** networkstatic has quit IRC | 15:02 | |
| *** amuller_ has quit IRC | 15:03 | |
| *** clev has joined #openstack-neutron | 15:03 | |
| *** otherwiseguy has joined #openstack-neutron | 15:03 | |
| *** chandankumar_ has quit IRC | 15:06 | |
| *** networkstatic has joined #openstack-neutron | 15:07 | |
| *** rwsu has joined #openstack-neutron | 15:16 | |
| *** sgran has quit IRC | 15:20 | |
| *** amuller_ has joined #openstack-neutron | 15:20 | |
| *** thuc has quit IRC | 15:23 | |
| *** thuc has joined #openstack-neutron | 15:23 | |
| *** amuller__ has joined #openstack-neutron | 15:24 | |
| *** xianghui has quit IRC | 15:25 | |
| *** amuller_ has quit IRC | 15:25 | |
| *** irenab has joined #openstack-neutron | 15:26 | |
| *** morganfainberg is now known as morganfainberg|z | 15:27 | |
| *** thuc has quit IRC | 15:28 | |
| *** aveiga has joined #openstack-neutron | 15:31 | |
| *** jgrimm has joined #openstack-neutron | 15:32 | |
| *** IanGovett has joined #openstack-neutron | 15:33 | |
| *** bashok has quit IRC | 15:33 | |
| *** markmcclain has joined #openstack-neutron | 15:34 | |
| *** alagalah has joined #openstack-neutron | 15:37 | |
| *** cdub has joined #openstack-neutron | 15:37 | |
| *** markmcclain has quit IRC | 15:38 | |
| openstackgerrit | Yves-Gwenael Bourhis proposed a change to openstack/neutron: Fixing lost vlan ids on interfaces https://review.openstack.org/66375 | 15:38 |
| cdub | mestery: you have an ml2/odl multi-node setup handy? | 15:40 |
| mestery | cdbu: I do, but at the moment, waiting for a fix from Madhu around DHCP not working. | 15:41 |
| mestery | I can revert to an older OpenFlow 1.0 though, which should work. | 15:41 |
| mestery | What do you need? | 15:41 |
| cdub | mestery: trying to get hsin-yi up and going w/ non-devstack based deployment | 15:41 |
| * mestery nods. | 15:42 | |
| mestery | I've got a fairly full slate of meetings, but can answer quiestions on IRC and help where needed. | 15:42 |
| mestery | Are you looking for configuration file examples for multi-node? | 15:42 |
| cdub | mestery: and she keeps hitting various problems (not to the point of actually testing, still in deploy phase) | 15:42 |
| * mestery nods. | 15:42 | |
| cdub | mestery: and i think part of her issue is that she hasn't set up devstack, so doesn't know what config should look like in the end | 15:42 |
| *** xuhanp has quit IRC | 15:43 | |
| cdub | mestery: yeah...i think config files would help | 15:43 |
| mestery | cdub: Yes, that's a problem I think. :) | 15:43 |
| mestery | Cool. | 15:43 |
| mestery | I'll create a paste with example config files now ... | 15:43 |
| *** mlavalle has joined #openstack-neutron | 15:43 | |
| cdub | mestery: thanks | 15:43 |
| *** markmcclain has joined #openstack-neutron | 15:44 | |
| cdub | mestery: i've been pointing her to snippets based on devstack (code not results of running) | 15:44 |
| mestery | cdub: OK. Do you want more than just ml2_conf.ini for both control and compute nodes? | 15:44 |
| mestery | cdub: ---> http://paste.openstack.org/show/61946/ | 15:45 |
| cdub | mestery: yeah, for example she's switched over from ml2/gre,vxlan and has leftover ovs agent (which she found confusing) | 15:45 |
| *** markwash has quit IRC | 15:45 | |
| mestery | cdub: OK, got it. | 15:46 |
| *** amarao has quit IRC | 15:46 | |
| cdub | mestery: and another issue is loading type driver w/out setup.cfg/egg-info alias for 'opendaylight' | 15:46 |
| cdub | i've hacked her around it for now | 15:46 |
| mestery | cdub: That must be a packaging issue, right? Shoot me more info on that one and I'll see what I can do. | 15:47 |
| cdub | heh, all info above ;) | 15:47 |
| mestery | ha! | 15:48 |
| mestery | :) | 15:48 |
| mestery | OK, check out the paste I sent, lets start there and see how far we can go. | 15:48 |
| cdub | type driver won't load by classname afaict | 15:48 |
| cdub | only by alias | 15:48 |
| *** ewindisch is now known as zz_ewindisch | 15:48 | |
| cdub | mestery: *nod* thanks! | 15:48 |
| mestery | Yes, ML2 uses stevedore for that functionality AFAIK. | 15:48 |
| cdub | yeah, i just thought perhaps a direct classname was still possible | 15:49 |
| * cdub thought wrong | 15:49 | |
| *** MM_at_HP has joined #openstack-neutron | 15:50 | |
| MM_at_HP | Morning | 15:50 |
| *** vkozhukalov has quit IRC | 15:50 | |
| anteaya | morning MM_at_HP | 15:51 |
| roaet | G'day all. | 15:51 |
| anteaya | hello roaet | 15:51 |
| mestery | cdub: Why do you need or want to use the direct class name? | 15:51 |
| * roaet waves to anteaya | 15:51 | |
| anteaya | roaet: going to join us for gate-blocking-bug day? | 15:51 |
| anteaya | we could use your help? | 15:51 |
| anteaya | we are in #openstack-gate | 15:51 |
| roaet | I will try. I have one bug here I gotta work today, but in theory it shouldn't be bad. After that I shall! | 15:52 |
| roaet | I will join the channel though | 15:52 |
| anteaya | thanks | 15:52 |
| cdub | mestery: because we aren't building w/ ml2, just copying it over | 15:53 |
| mestery | cdub: OK, makes sense now. | 15:53 |
| cdub | mestery: IOW, it's all still initial hack to prop it up and give it a go | 15:54 |
| anteaya | roaet: here are some instructions: http://lists.openstack.org/pipermail/openstack-dev/2014-January/025337.html | 15:54 |
| * roaet looks | 15:54 | |
| roaet | I was looking at a bug the other day and I was really stumped. I didn't know how to continue. So I will hopefully learn from this day. | 15:54 |
| roaet | anteaya: 'fingerprint' means to recheck bug #? | 15:57 |
| anteaya | fingerprint means logstash query | 15:57 |
| anteaya | the query that is unique to that failure signature | 15:57 |
| anteaya | and only collects failures, check build_status to ensure you are collecting only failures | 15:58 |
| anteaya | the idea is that it is specific enough to be discoverable and contain useful information for the person(s) who work to find the fix | 15:58 |
| roaet | Ah i see. | 15:58 |
| anteaya | thanks for asking | 15:58 |
| anteaya | good questions | 15:58 |
| *** jistr has quit IRC | 15:58 | |
| anteaya | also if a bug stumps you, share the bug url and ask for help | 15:59 |
| roaet | So is that just a matter of looking at an uncategorized bug, then finding an accurate logstash? | 15:59 |
| anteaya | if I am here I will take a look as well | 15:59 |
| openstackgerrit | Yves-Gwenael Bourhis proposed a change to openstack/neutron: Fixing lost vlan ids on interfaces https://review.openstack.org/66375 | 15:59 |
| anteaya | roaet: finding an accurate query yes | 15:59 |
| anteaya | then see if the bug has a bug report already, if not, file a bug report | 16:00 |
| *** zz_ewindisch is now known as ewindisch | 16:00 | |
| roaet | got it. | 16:00 |
| roaet | I can work on that during the busywait | 16:00 |
| anteaya | then add the logstash fingerprint and a url to the query to the bug report | 16:00 |
| anteaya | then submit a patch to the elastic-recheck repo with that bug's number as the title of the file and the contents being the logstash query | 16:01 |
| anteaya | thanks | 16:01 |
| *** markwash has joined #openstack-neutron | 16:02 | |
| anteaya | as an example: http://bit.ly/Ml2R4r is not a good fingerprint yet | 16:02 |
| anteaya | if you click on build_status on the left, it captures successes as well as failures | 16:03 |
| anteaya | I need to refine it | 16:03 |
| *** clev has quit IRC | 16:03 | |
| roaet | ok. | 16:03 |
| roaet | Now what do you mean push to the elastic recheck repo? | 16:03 |
| anteaya | I don't think I ever use the word push | 16:04 |
| anteaya | submit is the word I use | 16:04 |
| anteaya | https://review.openstack.org/#/c/69386/1 | 16:04 |
| *** clev has joined #openstack-neutron | 16:05 | |
| anteaya | submit a patch to the elastic-recheck repo like the example above | 16:05 |
| roaet | Ahhhh. | 16:05 |
| roaet | Alright that makes a lot more sense | 16:05 |
| anteaya | since in our workflow, we never push and we never merge | 16:05 |
| *** roeyc has joined #openstack-neutron | 16:05 | |
| anteaya | gerrit does that | 16:05 |
| anteaya | good | 16:05 |
| *** HenryG has quit IRC | 16:06 | |
| *** pcm__ has joined #openstack-neutron | 16:08 | |
| *** pcm_ has quit IRC | 16:11 | |
| *** pcm__ has quit IRC | 16:12 | |
| *** pcm_ has joined #openstack-neutron | 16:13 | |
| *** matsuhashi has quit IRC | 16:15 | |
| *** julim has quit IRC | 16:16 | |
| *** irenab has quit IRC | 16:19 | |
| *** julim has joined #openstack-neutron | 16:19 | |
| *** mfink has joined #openstack-neutron | 16:22 | |
| jlibosva | markmcclain: hi, can I ask something wrt changing down_revisions in db migration? | 16:22 |
| markmcclain | jlibosva: sure… what's your question? | 16:23 |
| jlibosva | markmcclain: there is this issue: https://bugs.launchpad.net/neutron/+bug/1271231 , basically creation of securitygroups table didn't get into havana release | 16:23 |
| *** clev has quit IRC | 16:23 | |
| jlibosva | markmcclain: that means this table is not created as a part of neutron-db-manage upgrade head during devstack deployment and is attempted to be created later, when doing neutron upgrade via grenade | 16:24 |
| jlibosva | at this time, table is already created by neutron itself | 16:24 |
| *** thuc has joined #openstack-neutron | 16:24 | |
| markmcclain | yeah.. create_all() is still executes in Havana | 16:25 |
| jlibosva | markmcclain: I had an idea of fix, that I'll send patch to stable/havana and master branch, where I'll change the dependency in chain in order to have creation of this table on correct place. I was told you will not like it :) | 16:25 |
| *** thuc_ has joined #openstack-neutron | 16:25 | |
| jlibosva | markmcclain: so I'd like to hear (read) your opinion on such fixes | 16:25 |
| *** irenab has joined #openstack-neutron | 16:26 | |
| *** AlexF_ has joined #openstack-neutron | 16:27 | |
| *** amuller__ has quit IRC | 16:27 | |
| markmcclain | depends on the fix | 16:27 |
| markmcclain | if create_all() is always ensuring this table exists than add the migrations makes sense | 16:27 |
| jlibosva | markmcclain: I wanted to change down_revision variable in several versions scripts | 16:28 |
| jlibosva | specifically 49f5e553f61f_ml2_security_groups.py to be executed before "havana" stamp | 16:29 |
| *** thedodd has joined #openstack-neutron | 16:29 | |
| *** thuc has quit IRC | 16:29 | |
| *** ewindisch is now known as zz_ewindisch | 16:30 | |
| *** dguitarbite has joined #openstack-neutron | 16:32 | |
| *** jistr has joined #openstack-neutron | 16:34 | |
| *** morganfainberg|z is now known as morganfainberg | 16:36 | |
| markmcclain | I think that we could move this one to before havana | 16:39 |
| markmcclain | jlibosva: and avoid breaking migrations for folks tracking either the master branch or the stable branches | 16:40 |
| *** aveiga has quit IRC | 16:42 | |
| jlibosva | markmcclain: ok, so I'll make a patch that changes the order in chain for both branches, right? | 16:44 |
| markmcclain | move it first in master | 16:45 |
| markmcclain | and then backport it to stable | 16:45 |
| jlibosva | ok | 16:46 |
| jlibosva | markmcclain: thank you | 16:46 |
| markmcclain | you're welcome | 16:46 |
| *** emagana has joined #openstack-neutron | 16:47 | |
| *** Mierdin has joined #openstack-neutron | 16:55 | |
| *** AlexF_ has quit IRC | 16:58 | |
| *** thedodd has quit IRC | 17:00 | |
| *** thedodd has joined #openstack-neutron | 17:01 | |
| *** morganfainberg is now known as morganfainberg|z | 17:07 | |
| *** ijw has quit IRC | 17:08 | |
| *** roeyc has quit IRC | 17:08 | |
| *** ijw has joined #openstack-neutron | 17:08 | |
| *** dave_tucker is now known as dave_tucker_zzz | 17:10 | |
| *** dave_tucker_zzz is now known as dave_tucker | 17:10 | |
| *** rohit404 has quit IRC | 17:15 | |
| openstackgerrit | Jakub Libosvar proposed a change to openstack/neutron: Move db migration of ml2 security groups to havana https://review.openstack.org/69416 | 17:17 |
| *** dave_tucker is now known as dave_tucker_zzz | 17:19 | |
| *** MM_at_HP has quit IRC | 17:23 | |
| *** HenryG has joined #openstack-neutron | 17:24 | |
| *** tongli has joined #openstack-neutron | 17:30 | |
| *** markmcclain has quit IRC | 17:31 | |
| *** markmcclain has joined #openstack-neutron | 17:32 | |
| *** markmcclain has quit IRC | 17:33 | |
| *** markmcclain has joined #openstack-neutron | 17:33 | |
| *** jpich has quit IRC | 17:36 | |
| *** thedodd has quit IRC | 17:38 | |
| *** afazekas has quit IRC | 17:38 | |
| openstackgerrit | Armando Migliaccio proposed a change to openstack/neutron: improve UT coverage for nicira_db operations https://review.openstack.org/66693 | 17:39 |
| *** thedodd has joined #openstack-neutron | 17:39 | |
| *** marun has joined #openstack-neutron | 17:39 | |
| *** dave_tucker_zzz is now known as dave_tucker | 17:39 | |
| *** vkozhukalov has joined #openstack-neutron | 17:40 | |
| *** marun has quit IRC | 17:41 | |
| *** ivar-lazzaro has joined #openstack-neutron | 17:42 | |
| *** ygbo has quit IRC | 17:50 | |
| *** luqas has quit IRC | 17:53 | |
| *** ijw has quit IRC | 17:59 | |
| *** dguitarbite has quit IRC | 18:00 | |
| *** ijw has joined #openstack-neutron | 18:00 | |
| *** dave_tucker is now known as dave_tucker_zzz | 18:00 | |
| *** dave_tucker_zzz is now known as dave_tucker | 18:03 | |
| *** ijw has quit IRC | 18:04 | |
| *** nati_ueno has joined #openstack-neutron | 18:05 | |
| *** julim has quit IRC | 18:05 | |
| *** harlowja_away is now known as harlowja | 18:05 | |
| *** safchain has quit IRC | 18:06 | |
| *** ijw has joined #openstack-neutron | 18:13 | |
| *** zzelle_ has joined #openstack-neutron | 18:15 | |
| *** ijw has quit IRC | 18:16 | |
| *** ijw has joined #openstack-neutron | 18:17 | |
| openstackgerrit | Jakub Libosvar proposed a change to openstack/neutron: Check during db migration if service* tables exist https://review.openstack.org/69433 | 18:25 |
| *** nati_ueno has quit IRC | 18:25 | |
| *** thedodd has quit IRC | 18:26 | |
| *** nati_ueno has joined #openstack-neutron | 18:26 | |
| *** nati_ueno has quit IRC | 18:26 | |
| *** nati_ueno has joined #openstack-neutron | 18:27 | |
| *** jistr has quit IRC | 18:32 | |
| *** MM_at_HP has joined #openstack-neutron | 18:40 | |
| openstackgerrit | Shiv Haris proposed a change to openstack/neutron: This migration changes the id field for the brocadenetworks table from int to string (uuid). This brocade specific table indexes using uuid https://review.openstack.org/68199 | 18:41 |
| *** jprovazn has joined #openstack-neutron | 18:44 | |
| *** jlibosva has quit IRC | 18:48 | |
| *** marun has joined #openstack-neutron | 18:56 | |
| *** thedodd has joined #openstack-neutron | 19:00 | |
| *** afazekas has joined #openstack-neutron | 19:04 | |
| yjiang5_1 | irenab: hi | 19:05 |
| *** itzikb has joined #openstack-neutron | 19:06 | |
| yjiang5_1 | ijw: hi | 19:08 |
| *** vkozhukalov has quit IRC | 19:08 | |
| *** afazekas is now known as afazekas_pub | 19:21 | |
| *** dguitarbite has joined #openstack-neutron | 19:21 | |
| *** hshen__ has joined #openstack-neutron | 19:25 | |
| *** armax has joined #openstack-neutron | 19:28 | |
| *** dguitarbite has quit IRC | 19:28 | |
| *** dguitarbite has joined #openstack-neutron | 19:29 | |
| *** jprovazn has quit IRC | 19:34 | |
| *** clev has joined #openstack-neutron | 19:36 | |
| *** djoreilly has quit IRC | 19:39 | |
| openstackgerrit | Shiv Haris proposed a change to openstack/neutron: Change index of brocade networks table from int to string https://review.openstack.org/68199 | 19:49 |
| pcm_ | nati_ueno: ping | 19:56 |
| *** gdubreui has joined #openstack-neutron | 19:59 | |
| *** thuc_ has quit IRC | 20:00 | |
| *** thuc has joined #openstack-neutron | 20:01 | |
| *** thuc_ has joined #openstack-neutron | 20:01 | |
| sc68cal | markmcclain: updated my part of the agenda for the meeting | 20:05 |
| *** thuc has quit IRC | 20:05 | |
| *** rossella_s has quit IRC | 20:05 | |
| *** openstackgerrit has quit IRC | 20:06 | |
| *** openstackgerrit has joined #openstack-neutron | 20:06 | |
| *** julim has joined #openstack-neutron | 20:06 | |
| markmcclain | sc68cal: thank you | 20:06 |
| *** nati_uen_ has joined #openstack-neutron | 20:08 | |
| *** nati_ueno has quit IRC | 20:09 | |
| pcm_ | Anyone: Had some questions on vendor drivers for VPN service... | 20:11 |
| pcm_ | Vendor device I have will require different identification for some items (e.g. IKE policy). | 20:12 |
| pcm_ | Was thinking of generating and persisting a mapping of OpenStack VPN IDs to device's IDs. | 20:12 |
| pcm_ | Should I do that persisting in service driver or device driver side? | 20:13 |
| openstackgerrit | Abhishek Raut proposed a change to openstack/neutron: Support Port Binding Extension in Cisco N1kv plugin https://review.openstack.org/68833 | 20:15 |
| *** IanGovett has left #openstack-neutron | 20:15 | |
| *** clev has quit IRC | 20:16 | |
| anteaya | based on the conversation in -gate, I have to ask, how did neutron code become dependant on a kernel other than the one we gate on? | 20:19 |
| anteaya | everyone tests using precise, correct? | 20:19 |
| enikanorov_ | i think ppl are testing with what they have at hand | 20:20 |
| *** clev has joined #openstack-neutron | 20:21 | |
| anteaya | so is there an awareness of the enviroment we gate on? | 20:22 |
| enikanorov_ | why does it matter? | 20:22 |
| anteaya | ha ha ha ha | 20:23 |
| anteaya | only because neutron regurarly breaks tests | 20:24 |
| anteaya | and people constantly try to shorehorn broken code into master | 20:24 |
| anteaya | because it works for them | 20:24 |
| enikanorov_ | what kind of neutron's broken code cause kernel crash? | 20:24 |
| enikanorov_ | you do understand that it is unreladed, don't you? | 20:25 |
| anteaya | how did neutron get to be using a kernel we don't gate on? | 20:25 |
| enikanorov_ | easily. just because we can't gate on the whole matrix | 20:25 |
| anteaya | I am asking a fundamental workflow problem | 20:26 |
| enikanorov_ | fundamental problem is that kernel should not crash | 20:26 |
| anteaya | good habits breed good code | 20:26 |
| enikanorov_ | we can't fix it by fixing our code | 20:26 |
| *** snowblind2 has joined #openstack-neutron | 20:26 | |
| anteaya | and being oblivious to what is going on around them is a constant issue | 20:27 |
| *** dsockwell has quit IRC | 20:27 | |
| enikanorov_ | yeah, it's better to use interface without relying on implementation | 20:27 |
| anteaya | they why bother haing neutron in the gate at all? | 20:28 |
| anteaya | it seems to be really hard to manage | 20:28 |
| enikanorov_ | it is hard when the bottom of software stack is not as stable as we expect | 20:29 |
| enikanorov_ | there's not much the top of the stack can do about it | 20:29 |
| *** dsockwell has joined #openstack-neutron | 20:29 | |
| marun | anteaya: people test on os's different from ubuntu, btw | 20:29 |
| marun | anteaya: the general rh policy is to validate on fedora/rhel before sending for review and falling back to ubuntu testing only when we experience gate failure. | 20:30 |
| enikanorov_ | i think if gating on rhel/fedora was introduced, we'd have # of bugs doubled | 20:31 |
| anteaya | marun: which version of fedora/rhel? | 20:31 |
| marun | anteaya: depends on the person. | 20:32 |
| anteaya | rhel/fedora is officially supported | 20:32 |
| anteaya | so no kernel version is specified? | 20:32 |
| marun | anteaya: usually the latest stable release | 20:32 |
| marun | anteaya: We don't depend on developers to discover gate problems. | 20:32 |
| marun | anteaya: that's what we have the gate for! | 20:33 |
| anteaya | great | 20:33 |
| marun | anteaya: it's not that we don't test | 20:33 |
| anteaya | who fixes it when bad coded gets in? | 20:33 |
| marun | anteaya: but we need a consistent test to validate merges | 20:33 |
| anteaya | I thought that was the gate | 20:33 |
| marun | anteaya: I'm afraid I'm not clear on what you're driving at. | 20:34 |
| anteaya | do you not feel the gate is a consistent test to validate merges | 20:34 |
| openstackgerrit | Mohammad Banikazemi proposed a change to openstack/neutron: Adds the new IBM SDN-VE plugin https://review.openstack.org/66453 | 20:34 |
| enikanorov_ | anteaya: you do realize that it was kernel that failed gate tests, not neutron or nova? | 20:34 |
| anteaya | I am driving at workflow | 20:34 |
| marun | anteaya: at present? The evidence would appear to point to 'no'. | 20:34 |
| marun | anteaya: But it's getting better. | 20:34 |
| marun | anteaya: workflow? | 20:35 |
| anteaya | well I still haven't seen any patches up to address the isolated tests | 20:35 |
| marun | anteaya: can you be more specific? | 20:35 |
| anteaya | yes | 20:35 |
| anteaya | sure | 20:35 |
| anteaya | start from a common env | 20:35 |
| marun | anteaya: how common? | 20:35 |
| anteaya | iterated code locally, tests pass | 20:35 |
| anteaya | submit code | 20:36 |
| *** jbrende has joined #openstack-neutron | 20:36 | |
| anteaya | will I have been using ubuntu precise myself | 20:36 |
| anteaya | anyone who asks me about how to set up testing is suggested to use ubuntu precise | 20:36 |
| marun | anteaya: and rh developers will be using fedora or rhel | 20:36 |
| anteaya | sure | 20:37 |
| anteaya | which is officially supported | 20:37 |
| mestery | The software in Ubuntu Precies is ancient. | 20:37 |
| mestery | Really old. Kernel and OVS are both super old there. | 20:37 |
| anteaya | mestery: they are lts | 20:37 |
| *** nati_uen_ has quit IRC | 20:37 | |
| mestery | lts with lots of known performance issues, yes. :) | 20:37 |
| marun | mestery: is kernel still old if you install the latest release? | 20:37 |
| anteaya | and what we gate on | 20:37 |
| enikanorov_ | "old" means 1 year? | 20:37 |
| marun | 12.04-3 has 3.8 | 20:37 |
| enikanorov_ | that kind of funny | 20:37 |
| mestery | OVS 1.4 is 2+ years old enikanorov_ | 20:37 |
| anteaya | we gate on lts | 20:38 |
| *** nati_ueno has joined #openstack-neutron | 20:38 | |
| mestery | No one in their right mind would run that anywhere near production | 20:38 |
| marun | mestery: if only that was the root of our problem it would be an easy fix! | 20:38 |
| enikanorov_ | even 2... | 20:38 |
| anteaya | when the next lts support comes out we will use that | 20:38 |
| mestery | marun: Not the root, but certianly not helping either | 20:38 |
| anteaya | mestery: take that up with jeblair | 20:38 |
| marun | mestery: not helping, but we have things in our control that have more influence on reliability | 20:38 |
| mestery | marun: Agreed. :) | 20:38 |
| marun | :) | 20:39 |
| enikanorov_ | i wonder how operators deploy into prod at all: new software are not proven, old one is... too old | 20:39 |
| marun | isn't that what mirantis is for? ;) | 20:39 |
| mestery | marun: Ha! :) | 20:39 |
| enikanorov_ | yes, that's the part of our tricky plan :) | 20:40 |
| * mestery loves it when a plan comes together. | 20:40 | |
| nati_ueno | pcm_: pong | 20:41 |
| marun | mestery: I'll tell you what I want to see happen. | 20:41 |
| marun | mestery: no more vm nonsense when we're debugging things. lxc/docker support for neutron so we can iterate faster. | 20:42 |
| mestery | Roger! | 20:42 |
| pcm_ | nati_ueno: Had some Qs on VPN. Emailed to openstack-dev. Can you take a peek at some point? | 20:42 |
| *** frankbutt has joined #openstack-neutron | 20:42 | |
| *** frankbutt has left #openstack-neutron | 20:43 | |
| ijw | yjiang5_1: hey, you called? | 20:43 |
| marun | anteaya: huh | 20:43 |
| marun | anteaya: I've been testing with 12.04-3 which defaults to the backported raring kernel. | 20:44 |
| marun | anteaya: no wonder reproduction has been so hard :( | 20:44 |
| nati_ueno | pcm_: sure | 20:45 |
| pcm_ | nati_ueno: Thanks! | 20:45 |
| *** carl_baldwin has joined #openstack-neutron | 20:47 | |
| openstackgerrit | Armando Migliaccio proposed a change to openstack/neutron: Remove extra network scheduling from vmware nsx plugin https://review.openstack.org/69465 | 20:48 |
| *** jlibosva has joined #openstack-neutron | 20:48 | |
| *** ajo has quit IRC | 20:54 | |
| *** terrylhowe has joined #openstack-neutron | 20:54 | |
| anteaya | marun: might be a part of it, yeah | 20:55 |
| sc68cal | We use 12.04 LTS in production, with the ubuntu cloud archive for more updated kernels and OVS | 20:59 |
| sc68cal | https://wiki.ubuntu.com/ServerTeam/CloudArchive | 21:00 |
| *** sweston has joined #openstack-neutron | 21:00 | |
| *** zz_ewindisch is now known as ewindisch | 21:01 | |
| *** banix has joined #openstack-neutron | 21:01 | |
| *** ivar-lazzaro has quit IRC | 21:05 | |
| *** sn6i23a has joined #openstack-neutron | 21:05 | |
| *** Sukhdev has joined #openstack-neutron | 21:05 | |
| mestery | sc68cal: Thanks for the data point helping to prove no one uses OVS 1.4 in production at any sort of resonable scale :) | 21:06 |
| *** irenab has quit IRC | 21:06 | |
| sc68cal | I have a bug that I need to file for Neutron, where some calls to the ovs_lib blows up when you use it with OVS 1.4 | 21:06 |
| *** irenab has joined #openstack-neutron | 21:07 | |
| sc68cal | I think it's the timeout parameter that it sends on the CLI, it makes 1.4 barf | 21:07 |
| *** jgrimm has quit IRC | 21:07 | |
| sc68cal | since it's not a valid option I think | 21:07 |
| sc68cal | in 1.4 | 21:07 |
| sc68cal | I think calls to mod_flow in the ovs_lib, when using neutron and 1.4, it blows up. Sort of recalling it in bits and pieces now | 21:07 |
| yjiang5_1 | ijw: Just wondering if Neutron support sR-IOV compute node now. | 21:08 |
| *** ajo has joined #openstack-neutron | 21:10 | |
| hshen__ | mestery: Hi kyle, I have neutron running with odl driver now. But it fails when creating network | 21:11 |
| mestery | hshen__: OK, it must be because of Neutron to ODL communication? | 21:11 |
| hshen__ | 2014-01-27 21:06:26.490 2329 ERROR neutron.plugins.ml2.managers [-] Mechanism driver 'opendaylight' failed in create_network_postcommit | 21:12 |
| *** jgrimm has joined #openstack-neutron | 21:12 | |
| hshen__ | mestery: this is from neutron server log | 21:12 |
| mestery | Yes: What version of ODL are you running? | 21:12 |
| mestery | You should grab the very latest build | 21:12 |
| *** irenab has quit IRC | 21:13 | |
| hshen__ | mestery: Mine is Janurary 16. | 21:15 |
| mestery | hshen__: That is really old :) | 21:16 |
| ijw | yjiang5_1: OK, apparently I'm writing a document that describes the pci_stats and the pci_flavor_attrs interactions. Might not get to it before tomorrow evening GMT, though - I just wanted to make you aware so that you look at it when I've done | 21:16 |
| mestery | I would recommend getting the latest and run with OpenFlow 1.3 support. | 21:16 |
| ijw | yjiang5_1: and Neutron doesn't support it yet, irenab is writing the code at the moment | 21:16 |
| yjiang5_1 | ijw: thanks. | 21:16 |
| mestery | Pop over to #opendaylight-ovsdb and look here: https://gist.github.com/anonymous/7bd7076ff74774cacd9c | 21:16 |
| yjiang5_1 | ijw: Sorry I didn't ask clearly. I'm talking about Robert's mail. I think currently Neutron (w/o network nova scheduler ) requires all host has same network environment, right? that means, if we have SR-IOV only mode, it will cause trouble. | 21:18 |
| ijw | Haven't read it yet, hang on | 21:18 |
| hshen__ | mestery: : OK. Let me grab a new odl. | 21:18 |
| ijw | Dunno what he's written, but the idea was that Neutron-capable ports would have an extra-info attr that indicated their network attachment, Neutron would check it and would fail a VM start if it wasn't both there and suitable | 21:19 |
| yjiang5_1 | ijw: for example, if we have a system w/o OVS support, it will be wrong if a VM is scheduled on it and the nic type is not specified as SR_IOV. So to that reason, we can't support SR-IOV only compute node. | 21:19 |
| ijw | (Decent scheduling to be sorted later, but that's no worse than provider networks now) | 21:19 |
| ijw | yjiang5_1: I don't think there's such a thing as an 'SRIOV only' compute node, I think he wants a compute node that is used only by VMs that have at least some SRIOV requirement | 21:20 |
| ijw | I see the logic, in some respects, but it's confusing the issue a bit because this isn't something we're working on before Juno | 21:20 |
| yjiang5_1 | ijw: For "a compute node that is used only by VMs that have at least some SRIOV requirement" , why do we have such requirement? IIUC, in Irena's doc, only if a port is requires explicitly with SR-IOV requirement, it will not use SR-IOV resource, right? | 21:22 |
| yjiang5_1 | ijw: sorr, only if a port is requires explicitly with SR-IOV requirement, *otherwise* it will not use SR-IOV resource, right | 21:22 |
| ijw | yjiang5_1: I think the problem he's trying to solve is (in a mixed machine cloud) we make sure that SRIOV supporting machines are not overloaded with VMs that don't need SRIOV resources | 21:22 |
| ijw | So a VM with *no* SRIOV requirements will not only not map in SRIOV ports but it will avoid that machine altogether | 21:23 |
| yjiang5_1 | ijw: got it. So I think that should be implemented as a weight filter. After all, it's just one factor of host selection. | 21:23 |
| *** alagalah has quit IRC | 21:25 | |
| yjiang5_1 | ijw: One extreme example is, a host is the only one left in the cloud with GPU assignment support, and is one of multiple host with SR-IOV support. When a user requires a GPU with vNIC network, we will assign this host to that VM still. | 21:25 |
| *** networkstatic has quit IRC | 21:28 | |
| *** networkstatic has joined #openstack-neutron | 21:28 | |
| ijw | yjiang5_1: OK, I've read his email now. I'm with Irena - I would expect there to always be a software switch in addition to the SRIOV devices. I think we should eventually add something to avoid using resources on machines with scarce limited resources, absolutely - though doing that for PCI in isolation is not a general solution, it could be useful. | 21:34 |
| *** afazekas_pub has quit IRC | 21:35 | |
| yjiang5_1 | ijw: I agree with you that we need avoid using host w/ SRIOV NIC, I just think it should be achieved through weight to keep it as just one factor. Of course, admin can always use host aggregate to achieve it. | 21:37 |
| ijw | yjiang5_1: I'm still not sure I understand what sriov_group actually is. | 21:37 |
| ijw | Robert keeps using it but not really defining it | 21:37 |
| ijw | yjiang5_1: Yeah, weight is preferable but I think the point was that because it's an enhancement then we can leave it till last rather than do it first | 21:38 |
| *** tongli has quit IRC | 21:38 | |
| yjiang5_1 | ijw: totally agree. | 21:38 |
| hshen__ | mestery: OK. got odl controller lastest version running and I am able to create network. | 21:41 |
| mestery | hshen__: Nice! | 21:41 |
| *** thuc_ has quit IRC | 21:41 | |
| *** thuc has joined #openstack-neutron | 21:42 | |
| *** thuc has quit IRC | 21:42 | |
| *** thuc has joined #openstack-neutron | 21:43 | |
| hshen__ | mestery: But I got VM binding failed. | 21:47 |
| *** alexpilotti has quit IRC | 21:47 | |
| mestery | hshen__: Very weird. | 21:47 |
| *** alexpilotti_ has joined #openstack-neutron | 21:47 | |
| hshen__ | he reason is in mechanism_odl.py, we cehck the constants.TYPE_LOCAL. The "constants" is imported from neutron.plugins.common. When I check /usr/lib/python2.6/site-packages/neutron/plugins/common/constants.py, I don't see any TYPE_LOCAL defined | 21:48 |
| hshen__ | mestery: so the neutron server log complains this: | 21:48 |
| hshen__ | mestery: | 21:48 |
| hshen__ | 2014-01-27 21:42:00.950 2329 TRACE neutron.plugins.ml2.managers if network_type in [constants.TYPE_LOCAL, constants.TYPE_FLAT, | 21:48 |
| hshen__ | 2014-01-27 21:42:00.950 2329 TRACE neutron.plugins.ml2.managers AttributeError: 'module' object has no attribute 'TYPE_LOCAL' | 21:48 |
| hshen__ | 2014-01-27 21:42:00.950 2329 TRACE neutron.plugins.ml2.managers | 21:48 |
| mestery | hshen__: Yes, there were changes there, what versoin of underlying RDO are you using? | 21:49 |
| mestery | You may need to adjust some imports if it's Icehouse, but a bit older. | 21:49 |
| hshen__ | This is RDO havana | 21:49 |
| mestery | Ah | 21:49 |
| mestery | OK, so you will definitely need to adjust those then :) | 21:49 |
| hshen__ | OK. where could I adjust? | 21:49 |
| mestery | I'm not sure where that is in Havana, you may have to search a bit. | 21:50 |
| *** ajo has quit IRC | 21:51 | |
| *** ajo has joined #openstack-neutron | 21:52 | |
| *** bjornar has joined #openstack-neutron | 21:59 | |
| hshen__ | mestery: mm.. let me find out. | 22:00 |
| *** dims has quit IRC | 22:00 | |
| *** dims has joined #openstack-neutron | 22:02 | |
| *** markwash has quit IRC | 22:03 | |
| *** WackoRob_ has joined #openstack-neutron | 22:04 | |
| *** WackoRob_ has quit IRC | 22:04 | |
| *** samuelbercovici has joined #openstack-neutron | 22:04 | |
| *** markwash has joined #openstack-neutron | 22:04 | |
| *** WackoRob_ has joined #openstack-neutron | 22:04 | |
| *** markwash has quit IRC | 22:05 | |
| enikanorov_ | salv-orlando: markmcclain: samuelbercovici: so going back to the patch, we want to set additional parameters to providers | 22:05 |
| enikanorov_ | the patch does that in 'configuration way' | 22:05 |
| markmcclain | what are tenant's configuring? | 22:06 |
| enikanorov_ | the alternative would be to create 'flavors' in the db, dynamically, which is not easy to do, because providers are conf objects, not persistent | 22:06 |
| *** pcm_ has quit IRC | 22:06 | |
| salv-orlando | hi folks - I will be in & out in this discussion as I'm multitasking on a few things | 22:06 |
| openstackgerrit | Armando Migliaccio proposed a change to openstack/neutron: Fix error while connecting to busy NSX L2 Gateway https://review.openstack.org/69482 | 22:07 |
| enikanorov_ | markmcclain: setting up flavors is admin's task, tenant only uses it | 22:07 |
| samuelbercovici | the specifc use case we have is that we have the the same driver able to provide lbaas in an HA manner and in a non-ha manner, I am not sure how this would be done using "flavors" | 22:07 |
| markmcclain | markmcclain: right | 22:07 |
| markmcclain | that's what we'd need to figure out | 22:08 |
| *** WackoRobie has quit IRC | 22:08 | |
| enikanorov_ | so my concern with dynamic flavors that they will be based on non-persistent object, so we would need a kind of configuration validation on neutron-service restart | 22:08 |
| markmcclain | much of this discussion about HA load balancers reminds me of the HA router discussion | 22:09 |
| *** WackoRob_ has quit IRC | 22:09 | |
| enikanorov_ | HA loadbalancer here however is a very specific use case | 22:09 |
| samuelbercovici | markmcclain: do you say that a "sub" property of the provider could be "flavors"? | 22:09 |
| *** SumitNaiksatam has quit IRC | 22:09 | |
| samuelbercovici | if so, that is still a configuration option, right? | 22:09 |
| markmcclain | no.. I think that providers should be removed from the API | 22:09 |
| *** armax has left #openstack-neutron | 22:09 | |
| enikanorov_ | markmcclain: in favor of flavors? | 22:10 |
| markmcclain | yes | 22:10 |
| enikanorov_ | that makes sense. | 22:10 |
| enikanorov_ | but it seems that providers (in fact, drivers) should still be somewhere as we need mapping from driver to flavor | 22:10 |
| markmcclain | flavors provide a couple of benefits: tenants don't care about the implementation, can enable a multiple vendor strategy | 22:11 |
| markmcclain | the mapping is scheduling | 22:11 |
| markmcclain | our current API requires the tenant to know too much about the backend | 22:11 |
| enikanorov_ | i think that tenants may or may not want about the backend | 22:12 |
| samuelbercovici | so instead of specifying providers, you soecify flavors? | 22:12 |
| enikanorov_ | yes | 22:12 |
| enikanorov_ | that is more flexible | 22:13 |
| enikanorov_ | but i'm not sure i get how the mapping is done between drivers and flavors | 22:13 |
| markmcclain | enikanorov: if a deployer wants to give the tenant more control | 22:13 |
| markmcclain | enikanorov_: that's what needs to be determined | 22:13 |
| salv-orlando | we can unless the tenant must provide some specific detail to configure an HA load balancer, for instance. Sambercovici, would that be possible? | 22:13 |
| samuelbercovici | and then when selecting a flavor, it would select one of the providers that can address this flavor? | 22:14 |
| *** jlibosva has quit IRC | 22:14 | |
| markmcclain | right | 22:14 |
| samuelbercovici | currently our driver is configure slightly different if it provide ha vs. non-ha | 22:14 |
| salv-orlando | i.e., having the tenant just saying "I want a 'yellow-diamond-flavor' loadbalancer, and then map it to a HA lb driver in the backend? | 22:14 |
| salv-orlando | or a white-diamond-flavor and then it map to the non-HA version? | 22:15 |
| enikanorov_ | honestly it doesn't seem more consistent than what is on the review :) | 22:15 |
| samuelbercovici | so if we add a "property" to the provider, saying it can support 'yellow-diamond-flavor' , this requires (at least in radware case) some specific paramater that will instantiate another driver with different parameters | 22:15 |
| enikanorov_ | because we would still need to do pretty complex matching | 22:15 |
| enikanorov_ | samuelbercovici: it would be capabilities matching, i think | 22:16 |
| enikanorov_ | otherwise it would not make sense | 22:16 |
| markmcclain | enikanorov_: I don't see how this any more complex than what the other projects are doing | 22:16 |
| salv-orlando | I have a feeling that we're making it a bit more complex than what it is, but I'm sure it's just me not being informed. so I'll shut down now and learn in the background | 22:17 |
| *** jecarey has quit IRC | 22:17 | |
| enikanorov_ | i think our flavors are a bit more complex than, say, nova's | 22:17 |
| enikanorov_ | because it is not a fixed set of parameters, and in general that set is vendor-specific | 22:17 |
| samuelbercovici | enikanorov_: correct. | 22:18 |
| EmilienM | markmcclain: hey Mark, if you want to have a look at blockers to have Grenade / Neutron working: https://review.openstack.org/#/c/69416/ | 22:19 |
| markmcclain | enikanorov_: I think we need about making that set of parameters more fixed and evolving over time | 22:19 |
| enikanorov_ | markmcclain: that moves us back to the question if vendors are able to expose their specific features | 22:20 |
| markmcclain | enikanorov_: given that we aren't exposing base features now | 22:20 |
| markmcclain | I think we're considering items out of order | 22:20 |
| markmcclain | EmilienM: I'll take a look | 22:20 |
| enikanorov_ | what particular base features do you mean? | 22:21 |
| markmcclain | let's say l4, ssl, or l7 | 22:22 |
| openstackgerrit | Mohammad Banikazemi proposed a change to openstack/neutron: Adds the new IBM SDN-VE plugin https://review.openstack.org/66453 | 22:22 |
| markmcclain | right now we're requiring users to know too much about the deployment | 22:22 |
| samuelbercovici | markmcclain: not sure how this is related to the providors/flavors discussion | 22:24 |
| enikanorov_ | my initial question on particular features is realted to flavors, however | 22:24 |
| enikanorov_ | i'm wondering how can particular vendor configure different flavors for its solution | 22:25 |
| enikanorov_ | if flavor parameters are fixed and not specific to that vendor | 22:25 |
| markmcclain | enikanorov_: vendors don't configure flavors | 22:25 |
| markmcclain | it is the deployers who configure | 22:25 |
| enikanorov_ | right | 22:25 |
| enikanorov_ | so deployers want to configure fast and cheap LBs | 22:25 |
| samuelbercovici | there are specifics that define how the vendor's driver needs to be set-up to deliver the APIs and the SLA required | 22:26 |
| enikanorov_ | and for vendor A and B it could be different set of parameters | 22:26 |
| salv-orlando | I think that allowing tenants to specify required SLA in terms of capability and/or performance criteria is a rabbit hole where we'd better not go. Just my personal opinion here though | 22:27 |
| markmcclain | salv-orlando: I agree | 22:28 |
| enikanorov_ | the idea is that deployers define that | 22:28 |
| enikanorov_ | tenants choose presets, e.g. flavors | 22:28 |
| samuelbercovici | so if a deployer does dev&test and needs no ha deployment and then later on for integ and prod ha is required, there need to be something that some one selects | 22:29 |
| samuelbercovici | who is this someone and what does he needs to select and when | 22:29 |
| samuelbercovici | ? | 22:29 |
| *** thuc has quit IRC | 22:29 | |
| samuelbercovici | opps deployer == tenant | 22:30 |
| enikanorov_ | i think 'flavor' is the right term here | 22:30 |
| openstackgerrit | Nachi Ueno proposed a change to openstack/neutron: Add enable_security_group option https://review.openstack.org/67281 | 22:30 |
| *** thuc has joined #openstack-neutron | 22:30 | |
| enikanorov_ | flavor is selected by tenant, configured by deployer | 22:30 |
| samuelbercovici | guys, do you agree with enikanorov_? | 22:31 |
| enikanorov_ | we kind of moved out of the line of initial question which is in my opinion is 'how that flavor looks like?' | 22:32 |
| *** ewindisch is now known as zz_ewindisch | 22:32 | |
| markmcclain | enikanorov_: yes flavors are selected by tenant | 22:32 |
| samuelbercovici | gr8 | 22:32 |
| markmcclain | and configured by deployers based on their deployment | 22:32 |
| enikanorov_ | of course. we need to agree on how flavor is constructed | 22:33 |
| samuelbercovici | so the deployer define flavors, lets say plain and ha, and tenant can select the flavor when creating the pool | 22:33 |
| *** sweston has quit IRC | 22:33 | |
| samuelbercovici | the information that radware can support plan and ha and that for plain a specific radweare driver configuration is needed neeeds to somehow happen | 22:34 |
| salv-orlando | this workflow looks good to me. I think next up is how flavour should be defined to make this happen | 22:34 |
| *** thuc has quit IRC | 22:34 | |
| *** mfink has quit IRC | 22:34 | |
| markmcclain | yeah.. we need to really think about the definition process | 22:35 |
| salv-orlando | for instance let's take the basic case, 1:1 to mapping between flavour and drivers. Won't work with the radware case unless they make two wrappers drivers which differently configure the same drivers according to the particular flavor | 22:35 |
| markmcclain | because this applies to more than just LBaaS | 22:35 |
| enikanorov_ | 1:1 mapping is what providers are now | 22:35 |
| salv-orlando | Or one could have driver and a list of params which are passed to a driver init function. So you could use the same driver in different configurations | 22:36 |
| enikanorov_ | salv-orlando: that is very close to what patch implements | 22:36 |
| enikanorov_ | the only issue with it is that it doesn't move us towards flavor framework | 22:37 |
| enikanorov_ | which is, however, for deployers mostly | 22:37 |
| salv-orlando | I am just thinking aloud here. | 22:37 |
| enikanorov_ | and tenant still chooses something, is it provider or flavor | 22:37 |
| samuelbercovici | I agree, that | 22:37 |
| samuelbercovici | opps..disregard | 22:38 |
| salv-orlando | I think both enikanorov_ are samuelbercovici are then suggesting we should go with icehouse with the current proposed approach and then look at flavours for the next release? | 22:38 |
| samuelbercovici | salv-orlando: even if we had flavors, we would still need somthing like the proposed patch and then map the flvors to the appropriate providers | 22:39 |
| enikanorov_ | salv-orlando: my opinion is that proposed configuration-approach is consistent with what we have right now (providers) | 22:39 |
| salv-orlando | ok, so we come full full circle back to the starting point, as usual | 22:39 |
| markmcclain | I'm still concerned about the technical debt | 22:40 |
| enikanorov_ | basically yes, I'd say it is fine for Icehouse | 22:40 |
| enikanorov_ | i don't fully understand what is techical debt here | 22:40 |
| openstackgerrit | Carl Baldwin proposed a change to openstack/neutron: Reduce severity of log messages in validation methods https://review.openstack.org/69037 | 22:41 |
| enikanorov_ | the intent is to add flavors, which a tenant-facing description of the service capability | 22:41 |
| enikanorov_ | we may implement it this or that way, from tenant perspective it should not change | 22:41 |
| enikanorov_ | for deployer it may change, however | 22:41 |
| pcarver | enikanorov_: Do you have use cases other than HA vs non-HA? I've worked with Alteon and F5 load balancers and I'm trying to think of any other feature beyond whether it's a single or pair that might fall under "flavor" | 22:42 |
| pcarver | I can think of lots of complex features, but for most of what I can think of "flavor" isn't enough. | 22:42 |
| enikanorov_ | pcarver: bandwidth limits? | 22:42 |
| enikanorov_ | or connection limits | 22:43 |
| pcarver | Hmm, when I think of bandwidth I think more of metering and billing, but I suppose you could put in limits. | 22:43 |
| enikanorov_ | billing is related, but orthogonal | 22:43 |
| samuelbercovici | salv-orlando: I would understand that a "flavor" would enable higher selection logic to scheule the implementation based on it vs. specificaly selecting a provider. none-the-less, the capability to convey the meaning via the specific needed configuration could be done in way of the proposed proposal or in a very similar alternative. I would not like to see solving one be tied with providing the other if possible. | 22:44 |
| enikanorov_ | for istance, you may want to schedule high-loaded lbs to certain appliances | 22:44 |
| samuelbercovici | pcarver: there might be sepcifc configuration related to how to connect to the overlay network | 22:44 |
| *** silvrax has quit IRC | 22:45 | |
| *** ajo has quit IRC | 22:45 | |
| samuelbercovici | also, specifying whether using HW or SW base lbs | 22:46 |
| pcarver | Ok, if you want to support bandwidth or connection limits at all rather than letting the dollars rule, then I see why flavors would be a simpler approach than fully extending the API to allow the tenant to tailor those settings. | 22:47 |
| *** gdubreui has quit IRC | 22:47 | |
| enikanorov_ | i'm not against extending the API, absolutely | 22:47 |
| pcarver | If you had a low cost, low end model of LB and a high cost high end model would those be the same provider in the current approach? | 22:47 |
| enikanorov_ | pcarver: if those models are operated by the same driver, then yes | 22:48 |
| pcarver | ok, in that case I definitely see why you need another means to distinguish. | 22:49 |
| samuelbercovici | pcarver: in radware case, its the same driver with different configuration, hence the discussed patch to allow registering different providers using the same driver and pssing differetn configuration oprions to the driver | 22:49 |
| samuelbercovici | markmcclain: salv-orlando: can we agree to go forward on this? | 22:51 |
| enikanorov_ | is it better to not add feature, or to add feature that most probably will change? | 22:52 |
| *** sweston has joined #openstack-neutron | 22:52 | |
| samuelbercovici | enikanorov_: I think that it will not change. the "falvors" will be mapped on top so can't see a reason to change. | 22:53 |
| salv-orlando | I think you both agree that this feature goes in the right direction and you both want to move to flavours. And I say you both, but I assume I am saying the whole lbaas community. | 22:53 |
| enikanorov_ | i don't see why one needs flavors with what is done in that patch | 22:53 |
| markmcclain | enikanorov_: we need flavors because providers do not provide a multiple vendor solution | 22:54 |
| samuelbercovici | to address the following use case: I have both radware and lets say citrix, both can be used as "gold" standard and as a tenant I don't care which will be used | 22:54 |
| enikanorov_ | markmcclain: how is that? | 22:54 |
| markmcclain | right now the provider dictates the driver | 22:55 |
| salv-orlando | I think it's been 18 months we have been discussing this now. | 22:55 |
| enikanorov_ | markmcclain: yep | 22:55 |
| samuelbercovici | so as a tenant, I want to just say vip on "gold" and let the system choose the actual implementation for me | 22:55 |
| enikanorov_ | markmcclain: do you mean tenant choses provider, and neutron schedules to *some* driver that is ok with provider? | 22:55 |
| samuelbercovici | enikanorov_: did you mean provider == flavor? | 22:56 |
| enikanorov_ | samuelbercovici: right now i'm trying to understand why providers don't give multivendor solution | 22:57 |
| salv-orlando | I just think that when I say m1.micro when booting a VM, this actually translates into CPU, RAM, and disk specs. I don't have anything like that in neutron. | 22:57 |
| markmcclain | salv-orlando: you're also not choosing the underlying virt tech either | 22:57 |
| enikanorov_ | salv-orlando: exactly. and possibly that can be diffrent for each vendor | 22:58 |
| markmcclain | it could be kvm, xen, hyperv, esx, etc | 22:58 |
| markmcclain | that's the way LB flavors should work | 22:58 |
| markmcclain | based on the attributes set agree on for flavors the system should schedule the pool to a LB instance | 22:58 |
| enikanorov_ | hmm | 22:58 |
| enikanorov_ | i think that is solved by filter scheduler | 22:59 |
| enikanorov_ | the ability to amend the choice | 22:59 |
| enikanorov_ | (in nova) | 22:59 |
| markmcclain | the are lots of different ways to impact scheduling | 22:59 |
| salv-orlando | ok, let's not digress into the capacity scheduling which is something specific to the hypervisor | 22:59 |
| markmcclain | we first have to disconnect the logical configuration from the backend that implements | 23:00 |
| salv-orlando | like RAM available. This might apply to LB as well in the future, but I don't see yet that possibly happening | 23:00 |
| enikanorov_ | ability to disconnecting logical config from the backend is a strict requirement for vendors | 23:01 |
| *** thuc has joined #openstack-neutron | 23:01 | |
| enikanorov_ | but i see your point | 23:01 |
| enikanorov_ | funny thing is that such kind of scheduling was proposed in grizzly, but apparently it was a bit forward-looking at that point | 23:02 |
| markmcclain | enikanorov_: the proposal in Grizzly tried to do too many things at once | 23:02 |
| salv-orlando | I don't think anybody here was talking about capability based scheduling as it was proposed for grizzly | 23:02 |
| carl_baldwin | Hi, I've got a couple of approved changes that need reverifying. I added "reverify bug 1253896" to the comments on Friday but nothing has happened. Am I missing something? | 23:03 |
| enikanorov_ | not capability, but it actually was about chosing the driver | 23:03 |
| enikanorov_ | and scheduling to available device | 23:03 |
| salv-orlando | I think we have digressed enough from the point where we started. | 23:03 |
| enikanorov_ | yes, and trying to move back | 23:04 |
| enikanorov_ | flavors framework with the requirement of making logical configuration fully independent from backend implementation has a number of prerequisities | 23:04 |
| salv-orlando | Summarizing, is there value in having flavours which might translate into a service offering concept similar to what nova flavour do for VM? If the answer is no, and it's agreed, let's forget about it. | 23:05 |
| enikanorov_ | that service (lbaas) need to comply first | 23:05 |
| *** sweston has quit IRC | 23:05 | |
| *** Sukhdev has quit IRC | 23:05 | |
| enikanorov_ | btw, isn't flavor and service offering a synonyms? | 23:05 |
| salv-orlando | they would be if the neutron team reaches consensus on a model common to all advanced services. | 23:08 |
| enikanorov_ | salv-orlando: my take is that we need flavors. that however doesn't mean to me that we should not provide simpler and less flexible solution while better solution is on the way | 23:08 |
| salv-orlando | reaching a consensus has always been impossible on this regard. | 23:08 |
| markmcclain | carl_baldwin: link? | 23:08 |
| carl_baldwin | Right, https://review.openstack.org/#/c/67475/ and https://review.openstack.org/#/c/67558/ | 23:09 |
| salv-orlando | reverify has been turned off for good I think | 23:09 |
| *** thuc has quit IRC | 23:09 | |
| salv-orlando | they need again a +1 | 23:09 |
| salv-orlando | +A sorry, but the neutron core team is not approving anything until the gate is unwedged, as this would probably just cause a reset of gate queue | 23:10 |
| markmcclain | carl_baldwin: let's run a recheck first | 23:10 |
| carl_baldwin | Fair enough. | 23:10 |
| markmcclain | salv-orlando: agree we need to build consensus across all of the advance services | 23:10 |
| *** jgrimm has quit IRC | 23:12 | |
| carl_baldwin | markmcclain: I'll watch for the rechecks to complete, assuming they are successful, what will trigger the gate to run? | 23:12 |
| salv-orlando | anyone can share the links to montreal ether pads? | 23:12 |
| markmcclain | if the recheck passes | 23:12 |
| *** clev has quit IRC | 23:12 | |
| carl_baldwin | markmcclain: That makes sense. Thanks. | 23:12 |
| markmcclain | than a core can add another +A to it | 23:12 |
| enikanorov_ | ok, i'll initiate a discussion on flavors on the ml soon | 23:13 |
| enikanorov_ | thanks for your time | 23:13 |
| salv-orlando | yeah, a +A will put it back on the gate queue. carl_baldwin… however, with the current kernel issue we have we need to be *very* lucky to pass check and then gate ;) | 23:13 |
| enikanorov_ | its 3:12 damn | 23:13 |
| samuelbercovici | markmcclain: the proposal is not trying to replace flavors and it will be an underlying building block fo it. | 23:13 |
| samuelbercovici | anyway, lets see the ML response | 23:14 |
| carl_baldwin | salv-orlando: understood, thanks. | 23:15 |
| samuelbercovici | bye | 23:15 |
| salv-orlando | adieu | 23:15 |
| *** oda-g has joined #openstack-neutron | 23:16 | |
| *** oda-g has left #openstack-neutron | 23:16 | |
| *** jecarey has joined #openstack-neutron | 23:17 | |
| *** samuelbercovici has quit IRC | 23:17 | |
| *** jbrende has quit IRC | 23:18 | |
| hshen__ | mestery: I fix the issue. Now VMs are up. But it looks like openstack cannot spawn VM in other compute node. | 23:18 |
| hshen__ | mestery: I have 4 VMs on the host of control node. And the network looks OK. They can ping each other. But I got erro when spawning VM5 with "no valid host" error. | 23:18 |
| hshen__ | mestery: Besides, I don't see any change in ODL GUI window. | 23:19 |
| hshen__ | mestery: Could you give some suggestion? | 23:19 |
| *** jbrende has joined #openstack-neutron | 23:21 | |
| *** aymenfrikha has left #openstack-neutron | 23:25 | |
| *** sdague has quit IRC | 23:27 | |
| *** sdague has joined #openstack-neutron | 23:28 | |
| *** bjornar has quit IRC | 23:29 | |
| *** ijw has quit IRC | 23:29 | |
| *** zzelle_ has quit IRC | 23:30 | |
| *** ijw has joined #openstack-neutron | 23:31 | |
| *** banix has quit IRC | 23:33 | |
| *** dims has quit IRC | 23:33 | |
| *** ijw has quit IRC | 23:36 | |
| *** morganfainberg|z is now known as morganfainberg | 23:41 | |
| *** doude_ has joined #openstack-neutron | 23:44 | |
| *** thedodd has quit IRC | 23:45 | |
| *** yamahata__ has joined #openstack-neutron | 23:46 | |
| openstackgerrit | Youcef Laribi proposed a change to openstack/neutron: Implements an LBaaS driver for NetScaler devices https://review.openstack.org/57524 | 23:46 |
| *** mlavalle has quit IRC | 23:48 | |
| *** mflobo_ has joined #openstack-neutron | 23:48 | |
| *** doude has quit IRC | 23:48 | |
| *** yamahata has quit IRC | 23:48 | |
| *** mflobo has quit IRC | 23:48 | |
| *** dims has joined #openstack-neutron | 23:48 | |
| *** thuc has joined #openstack-neutron | 23:49 | |
| *** puck has quit IRC | 23:50 | |
| *** gdubreui has joined #openstack-neutron | 23:52 | |
| *** thuc_ has joined #openstack-neutron | 23:53 | |
| *** peristeri has quit IRC | 23:53 | |
| *** hshen__ has quit IRC | 23:54 | |
| *** thuc has quit IRC | 23:55 | |
| *** networkstatic has quit IRC | 23:56 | |
| *** zzelle_ has joined #openstack-neutron | 23:56 | |
| *** BBG_Stephen has quit IRC | 23:59 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!