| *** yuanying has quit IRC | 00:00 | |
| *** yuanying has joined #openstack-lbaas | 00:01 | |
| *** cpuga has joined #openstack-lbaas | 00:18 | |
| *** sanfern has quit IRC | 00:47 | |
| *** yamamoto_ has joined #openstack-lbaas | 01:19 | |
| *** sanfern has joined #openstack-lbaas | 01:41 | |
| *** leitan has joined #openstack-lbaas | 01:55 | |
| *** leitan_ has joined #openstack-lbaas | 01:58 | |
| *** leitan has quit IRC | 02:01 | |
| *** ipsecguy_ has joined #openstack-lbaas | 02:04 | |
| *** ipsecguy has quit IRC | 02:06 | |
| *** leitan has joined #openstack-lbaas | 02:30 | |
| *** leitan__ has joined #openstack-lbaas | 02:31 | |
| *** leitan_ has quit IRC | 02:34 | |
| *** leitan has quit IRC | 02:35 | |
| *** yamamoto_ has quit IRC | 03:06 | |
| *** cpuga has quit IRC | 03:06 | |
| *** cpuga has joined #openstack-lbaas | 03:07 | |
| *** gans has joined #openstack-lbaas | 03:07 | |
| *** links has joined #openstack-lbaas | 03:09 | |
| *** cpuga has quit IRC | 03:11 | |
| *** sanfern has quit IRC | 03:13 | |
| *** sanfern has joined #openstack-lbaas | 03:14 | |
| *** sanfern has quit IRC | 03:14 | |
| *** cpuga has joined #openstack-lbaas | 03:19 | |
| *** armax has joined #openstack-lbaas | 03:23 | |
| gans | nmagnezi, your code is working good (y) | 03:26 |
|---|---|---|
| *** links has quit IRC | 03:32 | |
| *** cpuga has quit IRC | 03:48 | |
| *** cpuga has joined #openstack-lbaas | 03:48 | |
| *** links has joined #openstack-lbaas | 03:49 | |
| *** cpuga has quit IRC | 03:55 | |
| *** redrobot has quit IRC | 03:56 | |
| *** leitan__ has quit IRC | 03:56 | |
| *** leitan has joined #openstack-lbaas | 03:57 | |
| *** leitan has quit IRC | 03:57 | |
| *** redrobot has joined #openstack-lbaas | 03:57 | |
| *** redrobot is now known as Guest72345 | 03:57 | |
| *** yamamoto_ has joined #openstack-lbaas | 03:58 | |
| *** aojea has joined #openstack-lbaas | 03:59 | |
| *** Dinesh_Bhor has quit IRC | 04:02 | |
| *** aojea has quit IRC | 04:04 | |
| *** dayou has quit IRC | 04:08 | |
| *** krypto has joined #openstack-lbaas | 04:14 | |
| *** amotoki is now known as amotoki_away | 04:15 | |
| *** dayou has joined #openstack-lbaas | 04:16 | |
| *** amotoki_away is now known as amotoki | 04:17 | |
| *** gcheresh has joined #openstack-lbaas | 04:20 | |
| *** amotoki is now known as amotoki_away | 04:30 | |
| *** dineshbhor has joined #openstack-lbaas | 04:33 | |
| *** amotoki_away is now known as amotoki | 04:33 | |
| *** dineshbhor has quit IRC | 04:37 | |
| *** Dinesh_Bhor has joined #openstack-lbaas | 04:37 | |
| *** JudeC has joined #openstack-lbaas | 04:38 | |
| JudeC | Anyone around? | 04:38 |
| *** sanfern has joined #openstack-lbaas | 04:40 | |
| *** cpuga has joined #openstack-lbaas | 04:51 | |
| *** cpuga has quit IRC | 05:00 | |
| *** amotoki is now known as amotoki_away | 05:02 | |
| *** amotoki_away is now known as amotoki | 05:10 | |
| *** armax has quit IRC | 05:22 | |
| *** cpuga has joined #openstack-lbaas | 05:23 | |
| *** aojea has joined #openstack-lbaas | 05:39 | |
| *** cpuga has quit IRC | 05:43 | |
| *** krypto has quit IRC | 05:44 | |
| *** cpuga_ has joined #openstack-lbaas | 05:45 | |
| *** gcheresh has quit IRC | 05:52 | |
| *** aojea has quit IRC | 05:54 | |
| *** aojea has joined #openstack-lbaas | 05:54 | |
| *** aojea has quit IRC | 05:58 | |
| *** Guest72345 has quit IRC | 06:04 | |
| *** JudeC has quit IRC | 06:10 | |
| *** rcernin has joined #openstack-lbaas | 06:14 | |
| nmagnezi | gans, good to know :-) | 06:27 |
| nmagnezi | xgerman, rm_work, johnsom ^ | 06:27 |
| *** cpuga_ has quit IRC | 06:28 | |
| *** cpuga has joined #openstack-lbaas | 06:28 | |
| *** cpuga has quit IRC | 06:33 | |
| *** gcheresh has joined #openstack-lbaas | 06:51 | |
| *** krypto has joined #openstack-lbaas | 07:04 | |
| *** tesseract has joined #openstack-lbaas | 07:08 | |
| rm_work | nmagnezi: yeah I saw | 07:10 |
| rm_work | I'll look deeper in the morning | 07:10 |
| *** aojea has joined #openstack-lbaas | 07:31 | |
| *** pcaruana has joined #openstack-lbaas | 07:33 | |
| *** krypto has quit IRC | 07:34 | |
| rm_work | nmagnezi: i know the interface naming is a bit wonky but it SHOULD work with the old-style interface names set, which we do | 07:36 |
| rm_work | i'm unclear as to why it would have broken now (though I guess it obviously DID?) | 07:37 |
| *** pcaruana has quit IRC | 07:39 | |
| *** pcaruana has joined #openstack-lbaas | 07:39 | |
| openstackgerrit | Adam Harwell proposed openstack/octavia master: Pool name/desc needs to be "" when empty, not null https://review.openstack.org/467780 | 08:11 |
| *** openstackgerrit has quit IRC | 08:18 | |
| *** gans819 has joined #openstack-lbaas | 08:31 | |
| *** gans has quit IRC | 08:34 | |
| *** gans819 has quit IRC | 08:49 | |
| *** gans819 has joined #openstack-lbaas | 08:55 | |
| *** gans819 has quit IRC | 09:23 | |
| *** todin has quit IRC | 09:27 | |
| *** jmccrory has quit IRC | 09:27 | |
| *** afranc has quit IRC | 09:27 | |
| *** dayou has quit IRC | 09:27 | |
| *** links has quit IRC | 09:27 | |
| *** nmagnezi has quit IRC | 09:27 | |
| *** jidar has quit IRC | 09:27 | |
| *** raginbajin has quit IRC | 09:27 | |
| *** todin has joined #openstack-lbaas | 09:49 | |
| *** jmccrory has joined #openstack-lbaas | 09:49 | |
| *** afranc has joined #openstack-lbaas | 09:49 | |
| *** dayou has joined #openstack-lbaas | 10:13 | |
| *** links has joined #openstack-lbaas | 10:13 | |
| *** nmagnezi has joined #openstack-lbaas | 10:13 | |
| *** jidar has joined #openstack-lbaas | 10:13 | |
| *** raginbajin has joined #openstack-lbaas | 10:13 | |
| *** kbyrne has quit IRC | 10:17 | |
| *** kbyrne has joined #openstack-lbaas | 10:18 | |
| *** yamamoto_ has quit IRC | 10:31 | |
| *** sanfern has quit IRC | 10:39 | |
| *** gans has joined #openstack-lbaas | 10:52 | |
| *** csomerville has joined #openstack-lbaas | 11:05 | |
| *** cody-somerville has quit IRC | 11:06 | |
| *** yamamoto has joined #openstack-lbaas | 11:12 | |
| *** gans has quit IRC | 11:16 | |
| *** aojea has quit IRC | 11:49 | |
| *** chlong has joined #openstack-lbaas | 11:58 | |
| *** chlong has quit IRC | 12:01 | |
| *** links has quit IRC | 12:10 | |
| *** atoth has quit IRC | 12:23 | |
| *** mestery has quit IRC | 12:24 | |
| *** mestery has joined #openstack-lbaas | 12:27 | |
| *** atoth has joined #openstack-lbaas | 12:38 | |
| *** atoth has quit IRC | 12:39 | |
| *** atoth has joined #openstack-lbaas | 12:40 | |
| *** leitan has joined #openstack-lbaas | 12:51 | |
| *** chlong has joined #openstack-lbaas | 13:08 | |
| *** gans has joined #openstack-lbaas | 13:17 | |
| *** gcheresh has quit IRC | 13:35 | |
| *** aojea has joined #openstack-lbaas | 13:43 | |
| *** armax has joined #openstack-lbaas | 13:47 | |
| *** cpuga has joined #openstack-lbaas | 13:48 | |
| *** cpuga_ has joined #openstack-lbaas | 13:49 | |
| *** cpuga has quit IRC | 13:53 | |
| leitan | Hi guys, im working with octavia now, i have created a vxlan network on a tenant where i created my loadbalancer, i tested the LB from inside the namespace and everything is working ok, but when i go to associate a floating ip , to publish de loadbalancer into my network, the LB port does not appear, cause the ports are created with neutron/service user i guess, how its supposed this to be handled in a normal scenario ? | 14:00 |
| leitan | should i suppose to create the LB directly into the public network ??? (i dont find this friendly due to ip usage) | 14:03 |
| leitan | let me known when you can jump in, rm_work nmagnezi johnsom xgerman :) thanks | 14:04 |
| *** reedip_ has joined #openstack-lbaas | 14:18 | |
| *** gans has quit IRC | 14:24 | |
| leitan | ignore this, i was missing the router to the external network from the loadbalancer network, still not working but im taking a look at it | 14:30 |
| *** sanfern has joined #openstack-lbaas | 14:34 | |
| *** gcheresh has joined #openstack-lbaas | 14:51 | |
| *** rcernin has quit IRC | 15:09 | |
| *** armax has quit IRC | 15:12 | |
| *** armax has joined #openstack-lbaas | 15:14 | |
| *** gcheresh has quit IRC | 15:22 | |
| *** catintheroof has joined #openstack-lbaas | 15:25 | |
| *** catintheroof has quit IRC | 15:25 | |
| *** catintheroof has joined #openstack-lbaas | 15:26 | |
| *** yamamoto has quit IRC | 15:28 | |
| *** yamamoto has joined #openstack-lbaas | 15:28 | |
| *** yamamoto has quit IRC | 15:33 | |
| *** gans has joined #openstack-lbaas | 15:50 | |
| *** cody-somerville has joined #openstack-lbaas | 15:50 | |
| *** cody-somerville has quit IRC | 15:50 | |
| *** cody-somerville has joined #openstack-lbaas | 15:50 | |
| *** csomerville has quit IRC | 15:53 | |
| *** sanfern has quit IRC | 15:58 | |
| *** aojea has quit IRC | 15:58 | |
| *** sanfern has joined #openstack-lbaas | 15:59 | |
| *** aojea has joined #openstack-lbaas | 15:59 | |
| *** gans has quit IRC | 16:01 | |
| *** tesseract has quit IRC | 16:03 | |
| *** aojea has quit IRC | 16:03 | |
| *** gans has joined #openstack-lbaas | 16:07 | |
| *** gans has quit IRC | 16:07 | |
| *** catintheroof has quit IRC | 16:08 | |
| *** catintheroof has joined #openstack-lbaas | 16:09 | |
| *** armax has quit IRC | 16:32 | |
| *** yamamoto has joined #openstack-lbaas | 16:35 | |
| *** armax has joined #openstack-lbaas | 16:36 | |
| *** JudeC has joined #openstack-lbaas | 16:39 | |
| *** pcaruana has quit IRC | 16:43 | |
| *** reedip_ has quit IRC | 16:46 | |
| *** yamamoto has quit IRC | 16:47 | |
| *** cpuga_ has quit IRC | 17:08 | |
| *** cpuga has joined #openstack-lbaas | 17:09 | |
| *** harlowja has joined #openstack-lbaas | 17:14 | |
| *** sshank has joined #openstack-lbaas | 17:19 | |
| rm_work | leitan: are you in a real cloud or just devstack? | 17:19 |
| leitan | rm_work: real cloud | 17:19 |
| rm_work | i assume a real cloud | 17:19 |
| rm_work | k | 17:19 |
| rm_work | yeah, so i did run into one issue where for some reason associating a FLIP wasn't actually bringing the interface up | 17:19 |
| rm_work | but i think it was because our internal networking is a little odd | 17:20 |
| rm_work | it's supposed to automatically create an interface for it in the netns | 17:20 |
| leitan | rm_work: i debugged this a little | 17:20 |
| leitan | when i list the neutron port | 17:20 |
| leitan | and list the floating ips | 17:20 |
| rm_work | yeah, debug mode for the amp agent is a good idea, and looking at those logs | 17:20 |
| leitan | i see that its actually asociated with the vip port | 17:20 |
| rm_work | johnsom: BTW, o/ welcome back | 17:20 |
| rm_work | leitan: that's correct | 17:21 |
| leitan | but, theres no nat rule on any of the router namespace | 17:21 |
| leitan | where the amphorae instances are running | 17:21 |
| johnsom | o/ Still digging out from being gone | 17:21 |
| rm_work | ah, yeah we don't touch the routers | 17:21 |
| rm_work | we assume the networking layer is all routable correctly | 17:21 |
| rm_work | johnsom: ok when you get done with that, i have a list of prioritized reviews for you | 17:22 |
| rm_work | ;) | 17:22 |
| leitan | rm_work: ok, so then the LB_NETWORK where i create the loadbalancers, CANT be a isolated vxlan and then associate a floating ip to the vip | 17:22 |
| leitan | that is not going to work | 17:22 |
| rm_work | hmmm | 17:23 |
| leitan | i assume that is going to work just like the agent mode, where you can do that | 17:23 |
| leitan | buy that doesnt use allowed address pairs | 17:23 |
| leitan | so maybe theres a thing there | 17:23 |
| leitan | i dont know | 17:23 |
| johnsom | Why would it not work??? (other than DVR mode bugs in neutron) | 17:23 |
| leitan | dont know, its not working for me | 17:24 |
| leitan | just debugging | 17:24 |
| leitan | everything worked perfectly, but associating a floating ip to the private isolated vip is not working | 17:24 |
| johnsom | Make sure your router is not set for "distributed" (DVR), if it is, there are known bugs in the DVR code for floating IPs. | 17:24 |
| leitan | all my routers are DVR on the setup | 17:24 |
| leitan | dvr_snat and dvr | 17:25 |
| rm_work | :( | 17:25 |
| johnsom | Yeah, that is the problem then. With DVR routers the floating IPs don't pass traffic | 17:25 |
| johnsom | It's a known bug in DVR | 17:25 |
| leitan | a bug with DVR + allowed address pairs ? | 17:25 |
| johnsom | Yes | 17:25 |
| leitan | crap | 17:26 |
| rm_work | yes. >_> | 17:26 |
| rm_work | crap is approximately correct | 17:26 |
| johnsom | I think Kevin was putting some effort into getting that finally fixed in this release, but I'm not up to date on their efforts | 17:26 |
| johnsom | That bug has been open since like mitaka | 17:26 |
| rm_work | kevinbenton is looking into it for us? :P maybe we are saved | 17:26 |
| leitan | ok, ill do this, ill create the LB on the public network, to se if it works, allocating the ips on a flat space | 17:26 |
| rm_work | trust in the kevinbenton | 17:26 |
| johnsom | Yeah, Kevin is the man. | 17:27 |
| leitan | just to test everything, that should work, because it worked correctyle on the vlxan network | 17:27 |
| johnsom | Yeah, other thing to test, is set it up on a non dvr router | 17:27 |
| leitan | johnsom: yes but i dont have available agents to schdule a non dvr | 17:27 |
| leitan | i can set then a business rule to enforce that public LBs should be created on public net and consume 3 ips | 17:28 |
| * leitan cries while the PM slap him in the face | 17:28 | |
| *** catintheroof has quit IRC | 17:29 | |
| leitan | in kevinbenton we trust then | 17:30 |
| johnsom | https://bugs.launchpad.net/neutron/+bug/1583694 | 17:30 |
| openstack | Launchpad bug 1583694 in neutron "[RFE] DVR support for Allowed_address_pair port that are bound to multiple ACTIVE VM ports" [Wishlist,In progress] - Assigned to Swaminathan Vasudevan (swaminathan-vasudevan) | 17:30 |
| leitan | in the meantime, ill try it that way | 17:30 |
| leitan | summing up some "it affects me" | 17:31 |
| johnsom | This is the last bug I know was tracking the exact issue. | 17:31 |
| johnsom | There may be others | 17:31 |
| leitan | thanks guys you saved me from a lot of unnecesary debugging johnsom rm_work | 17:34 |
| johnsom | Sorry the answer is "it's broke" | 17:34 |
| leitan | johnsom: hey, no worries, i know how it goes been deploying Openstack since Cactus release | 17:35 |
| johnsom | Ouch | 17:35 |
| rm_work | yeah that sounds spiky | 17:36 |
| rm_work | johnsom: so you think you'll be done catching up by sometime in July? :P | 17:37 |
| johnsom | Ha, well, I still need to catch up on my OpenStack-dev e-mail, so.... | 17:38 |
| johnsom | Itching to look into this interface thing. We SET that name, so puzzled what the issue is | 17:38 |
| johnsom | rm_work do you have an etherpad list for me? | 17:40 |
| rm_work | johnsom: well that's the #1 priority anyway | 17:42 |
| johnsom | I'm not real excited about the 3GB thing either. It's a penalty for the other OS's. | 17:43 |
| rm_work | only in devstack | 17:43 |
| rm_work | anyone making their own image should be able to do whatever | 17:43 |
| johnsom | Yeah | 17:43 |
| rm_work | i dunno | 17:44 |
| johnsom | I wonder what the difference is in gate test time... | 17:44 |
| rm_work | i'm kinda on the fence but | 17:44 |
| rm_work | i feel like in general we tend towards LCD | 17:44 |
| johnsom | I lean toward encouraging the cloud images to be smaller... Leaving it default 2gb, but printing warnings it RHEL is selected, making the gates use 3gb if it is RHEL. (though our centos gate isn't failing...) | 17:46 |
| johnsom | I mean, in reality we can't use RHEL in the gates, so... | 17:46 |
| *** leitan has quit IRC | 17:51 | |
| *** yamamoto has joined #openstack-lbaas | 17:53 | |
| *** aojea has joined #openstack-lbaas | 17:53 | |
| sanfern | Do we have tempest scenario test cases and Rally test cases to test Octavia ? Any document around it | 17:54 |
| johnsom | Rally, no. | 17:54 |
| johnsom | Tempest yes, but they are old and poor. I have created a repository to start a "reboot" of the tempest tests for Octavia | 17:54 |
| johnsom | https://github.com/openstack/octavia-tempest-plugin | 17:55 |
| johnsom | What we currently have for tempest test for Octavia are: | 17:55 |
| johnsom | 1. The neutron-lbaas tests, since they use the Octavia driver: https://github.com/openstack/neutron-lbaas/tree/master/neutron_lbaas/tests/tempest | 17:56 |
| johnsom | 2. The Octavia tempest tests, v1 API: https://github.com/openstack/octavia/tree/master/octavia/tests/tempest | 17:56 |
| sanfern | ok thanks johnsom, i will look into that and run those in our environment | 17:57 |
| johnsom | 3. some patches for V2 API, but still the old bad code: https://review.openstack.org/#/c/413697 and https://review.openstack.org/#/c/429825 | 17:57 |
| sanfern | ok | 17:57 |
| johnsom | The patches need some work I think. | 17:58 |
| sanfern | johnsom, Are we planning to add Rally test cases? | 17:58 |
| *** JudeC has quit IRC | 17:59 | |
| johnsom | There has been debate on the value. What are you looking for out of Rally tests? | 17:59 |
| *** JudeC has joined #openstack-lbaas | 17:59 | |
| sanfern | creating 100 LBs and collect the VIP response time | 18:00 |
| johnsom | Ok | 18:01 |
| *** leitan has joined #openstack-lbaas | 18:01 | |
| johnsom | I'm open to adding Rally tests in addition to tempest. We just need to understand the value and what we can run on the OpenStack gate hosts. | 18:02 |
| sanfern | I want to benchmark results whenever we bringup the new cloud | 18:03 |
| *** blogan has joined #openstack-lbaas | 18:12 | |
| *** openstackgerrit has joined #openstack-lbaas | 18:17 | |
| openstackgerrit | Merged openstack/octavia master: Update example config to have more correct keystone_authtoken example https://review.openstack.org/465772 | 18:17 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/octavia master: Updated from global requirements https://review.openstack.org/467473 | 18:20 |
| *** gcheresh has joined #openstack-lbaas | 18:23 | |
| rm_work | johnsom: sooo with https://review.openstack.org/#/c/468963/ i don't understand WHY it fixes the gate but it does? | 18:27 |
| rm_work | johnsom: i can say, we are broken ATM for some reason | 18:27 |
| johnsom | Yeah, I'm stacking now to take a look at it. | 18:28 |
| johnsom | Something just is odd since we assign those names as we add interfaces to the namespace. | 18:28 |
| *** armax has quit IRC | 18:37 | |
| leitan | Guys is there any quick api call to know wich amphorae is the master without checking the vrrp config in the instance ? | 18:43 |
| johnsom | No, not currently. As it turns out keepalived makes this a bit hard to identify who is currently the master and it's logging is sometimes not accurate. | 18:44 |
| johnsom | If it is a fresh pair you can query the database | 18:45 |
| leitan | its a fresh pair indeed | 18:45 |
| johnsom | But if there has been a failover we currently don't have a way to tell. It | 18:45 |
| johnsom | It is in the amp info API, but we found we didn't have a good way to get reliable info. | 18:45 |
| leitan | since from the controllers you have trusted ssh access to the amphorae what about some kind of polling from the healthmanager that actually parse the ip command to get wich one is advertising the vip ? | 18:48 |
| leitan | and refreshing that in a field on the loadbalancer details | 18:49 |
| johnsom | Well, the agent would just do that, no need for remote query | 18:49 |
| *** sanfern has quit IRC | 18:49 | |
| leitan | yes youre right, but the "ip method" sounds too awful ? | 18:50 |
| johnsom | The issue is we also have another little bug, where at some points, the IP is listed on the interface even though it's not the active master. Since it doesn't impact functionality it has been super low on the fix list | 18:50 |
| leitan | johnsom: i see | 18:50 |
| leitan | maybe checking the access_log from haproxy, if its growing in a timelapse of X minutes, its the active | 18:51 |
| *** sshank has quit IRC | 18:58 | |
| leitan | or on the keepalived template that is built, an script that actualy leaves a true or false in a path when for example is seeing logs on the access of the balancer | 18:59 |
| leitan | sorry, just thinking about options :) | 19:00 |
| *** yamamoto has quit IRC | 19:00 | |
| *** yamamoto has joined #openstack-lbaas | 19:05 | |
| *** leitan has quit IRC | 19:13 | |
| *** leitan has joined #openstack-lbaas | 19:15 | |
| *** yamamoto has quit IRC | 19:25 | |
| *** JudeC has quit IRC | 19:32 | |
| *** ianychoi_ has joined #openstack-lbaas | 19:36 | |
| *** ianychoi has quit IRC | 19:38 | |
| *** armax has joined #openstack-lbaas | 20:02 | |
| *** JudeC has joined #openstack-lbaas | 20:24 | |
| *** yamamoto has joined #openstack-lbaas | 20:26 | |
| *** yamamoto has quit IRC | 20:34 | |
| *** PagliaccisCloud has joined #openstack-lbaas | 20:39 | |
| *** cody-somerville has quit IRC | 20:42 | |
| *** cody-somerville has joined #openstack-lbaas | 20:45 | |
| *** cody-somerville has quit IRC | 20:45 | |
| *** cody-somerville has joined #openstack-lbaas | 20:45 | |
| *** gcheresh has quit IRC | 20:46 | |
| *** atoth has quit IRC | 20:50 | |
| *** atoth has joined #openstack-lbaas | 21:02 | |
| *** leitan has quit IRC | 21:04 | |
| *** KeithMnemonic has joined #openstack-lbaas | 21:13 | |
| johnsom | Ugh, I keep getting an image from the 10th, which runs fine. too many cached versions. Cleaned up now, trying again | 21:13 |
| *** cpuga has quit IRC | 21:14 | |
| *** aojea has quit IRC | 21:17 | |
| *** sshank has joined #openstack-lbaas | 21:28 | |
| *** cpuga has joined #openstack-lbaas | 21:49 | |
| *** sshank has quit IRC | 21:50 | |
| *** leitan has joined #openstack-lbaas | 22:07 | |
| openstackgerrit | Jude Cross proposed openstack/octavia master: Add filtering and field selection to API https://review.openstack.org/469275 | 22:11 |
| *** JudeC has quit IRC | 22:26 | |
| *** JudeC has joined #openstack-lbaas | 22:45 | |
| rm_work | johnsom: how goes that... | 22:52 |
| johnsom | Well, if I build an image with the May 10th 875cb2db03f4a5551f823e82de3afd18566bdf53 commit the agent is fine. | 22:53 |
| rm_work | yeah k | 22:53 |
| rm_work | hmm | 22:53 |
| rm_work | so it's not something upstream | 22:53 |
| rm_work | but something internally? | 22:54 |
| rm_work | except | 22:54 |
| johnsom | I see that the interface moved into the namespace has the wrong name | 22:54 |
| rm_work | how would that have merged | 22:54 |
| johnsom | I am thinking it's maybe a pyroute2 issue? | 22:54 |
| johnsom | Still digging | 22:54 |
| johnsom | It's just odd | 22:55 |
| rm_work | and why would nmagnezi's fix work | 22:56 |
| johnsom | Well, I think he is preserving the interface name from outside the namespace to inside, but I think it will have problems | 22:57 |
| rm_work | i'm actually unclear as to why it had to be done the way you did it | 22:57 |
| rm_work | and not the way he's doing it | 22:57 |
| johnsom | This is why I forced the names inside the namespace | 22:57 |
| rm_work | k well, once we get this fixed, it'd be nice to merge ... err... basically the whole client | 22:58 |
| rm_work | and then a bunch of my bugfixes | 22:58 |
| johnsom | Well, for one thing think about this scenario: | 22:58 |
| johnsom | amp has eth1 (mgmt) and eth2 (VIP) We move eth2 into netns as eth2. World happy. | 22:59 |
| johnsom | Then we add member network, which when hotplugged comes in as eth2 since in the default namespace there is no longer an eth2. Now migrate that into the netns and you have a conflict. | 23:00 |
| johnsom | Basically the numbering changes based on what is plugged and unplugged in what order | 23:00 |
| johnsom | So, I chose to "manage" the naming in the netns | 23:01 |
| rm_work | ah, k | 23:01 |
| rm_work | does your code actually solve for that either though? | 23:02 |
| rm_work | i only saw one constant for naming that | 23:02 |
| rm_work | otherwise it seems the same? | 23:02 |
| rm_work | anywho, if you make a patch that makes the gate work, i'll +2 it :) | 23:02 |
| johnsom | Yes, the VIP is always first and doesn't come/go. So VIP is always eth1 in my naming/numbering scheme. I.e. the OS could call that port em2000 and I will always name it eth1 in the netns | 23:03 |
| rm_work | but then if you add another member network | 23:03 |
| rm_work | couldn't it also get em2000, and then it'd be moved into the netns as ??? | 23:04 |
| rm_work | and even if it gets moved as em2000 | 23:04 |
| rm_work | doesn't the second member network suffer from the same issue? | 23:04 |
| johnsom | They are moved in as eth2, 3, 4, 5 etc. All tracked by MAC | 23:04 |
| rm_work | ah | 23:04 |
| johnsom | /var/lib/octavia/plugged_interfaces | 23:04 |
| rm_work | i need to reread the code i guess, didn't catch that | 23:04 |
| rm_work | ah i guess he didn't touch that in his patch | 23:04 |
| *** sshank has joined #openstack-lbaas | 23:23 | |
| openstackgerrit | Jude Cross proposed openstack/octavia master: Add filtering and field selection to API https://review.openstack.org/469275 | 23:24 |
| *** leitan has quit IRC | 23:31 | |
| *** leitan has joined #openstack-lbaas | 23:36 | |
| johnsom | Yeah, it is definitely an issue related to pyroute2. The command line still works, but I can reproduce the bug via pyroute2 | 23:41 |
| cpuga | johnsom: is tomorrow's meeting still schedule to occur? | 23:43 |
| johnsom | Yes, we will have a meeting tomorrw | 23:43 |
| cpuga | would it be possible to add "flavors" as a topic for tomorrow | 23:43 |
| johnsom | Sure, NP | 23:44 |
| cpuga | k, thx | 23:44 |
| rm_work | oh i need to add a topic... | 23:47 |
| rm_work | or two... | 23:48 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!