| xgerman | dougwig so did we ever decide what the relationship between flavor and provider is? | 00:05 |
|---|---|---|
| xgerman | In my opinion flavor should replace provider… so we should try to phase that out and flavor in | 00:06 |
| xgerman | thoughts? | 00:06 |
| johnsom | I'm not so sure. Flavor Octavia-bronze and Flavor Octavia-silver (may tls offload) is still provider Octavia | 00:07 |
| xgerman | yeah, but I think lbaas-create should have —flavor instead of provider | 00:07 |
| xgerman | otherwise we set up bronze and the user overwrites... | 00:07 |
| openstackgerrit | Claudiu Belu proposed openstack/octavia: WIP: Removes unused posix-ipc requirement https://review.openstack.org/219065 | 00:08 |
| openstackgerrit | min wang proposed openstack/octavia: Increase connection_max_retries to be 300 https://review.openstack.org/218613 | 00:15 |
| *** diogogmt has quit IRC | 00:18 | |
| openstackgerrit | min wang proposed openstack/neutron-lbaas: Set up LBaas V2 tempest test gate against Octavia. https://review.openstack.org/209675 | 00:19 |
| *** vivek-ebay has joined #openstack-lbaas | 00:25 | |
| *** bana_k has quit IRC | 00:25 | |
| *** bana_k has joined #openstack-lbaas | 00:25 | |
| openstackgerrit | min wang proposed openstack/octavia: Increase connection_max_retries to be 300 https://review.openstack.org/218613 | 00:30 |
| openstackgerrit | min wang proposed openstack/neutron-lbaas: Set up LBaas V2 tempest test gate against Octavia. https://review.openstack.org/209675 | 00:31 |
| *** clev is now known as clev-away | 00:43 | |
| *** abdelwas has quit IRC | 00:51 | |
| *** madhu_ak has quit IRC | 00:56 | |
| *** chlong has joined #openstack-lbaas | 01:02 | |
| *** rm_you has quit IRC | 01:22 | |
| *** rm_you has joined #openstack-lbaas | 01:23 | |
| *** rm_you has joined #openstack-lbaas | 01:23 | |
| *** bana_k has quit IRC | 01:33 | |
| *** haigang has quit IRC | 01:44 | |
| *** ganeshna has joined #openstack-lbaas | 02:08 | |
| *** ganeshna has quit IRC | 02:10 | |
| *** ganeshna has joined #openstack-lbaas | 02:10 | |
| openstackgerrit | min wang proposed openstack/octavia: Increase connection_max_retries to be 300 https://review.openstack.org/218613 | 02:13 |
| openstackgerrit | min wang proposed openstack/neutron-lbaas: Set up LBaas V2 tempest test gate against Octavia. https://review.openstack.org/209675 | 02:14 |
| *** woodster_ has quit IRC | 02:19 | |
| *** I has joined #openstack-lbaas | 02:20 | |
| *** clev-away is now known as clev | 02:20 | |
| *** I is now known as Guest10903 | 02:20 | |
| *** sbalukoff has quit IRC | 02:21 | |
| *** ganeshna has quit IRC | 02:24 | |
| *** bana_k has joined #openstack-lbaas | 03:06 | |
| *** chlong has quit IRC | 03:12 | |
| *** Jian0612 has joined #openstack-lbaas | 03:13 | |
| openstackgerrit | Brandon Logan proposed openstack/neutron-lbaas: Add ability for octavia to allocate vip https://review.openstack.org/214058 | 03:20 |
| *** vivek-ebay has quit IRC | 03:34 | |
| *** chlong has joined #openstack-lbaas | 03:38 | |
| *** chlong_ has joined #openstack-lbaas | 04:04 | |
| *** chlong has quit IRC | 04:04 | |
| *** sbalukoff has joined #openstack-lbaas | 04:05 | |
| *** amotoki has joined #openstack-lbaas | 04:17 | |
| *** bana_k has quit IRC | 04:19 | |
| *** numans has joined #openstack-lbaas | 05:05 | |
| *** Alex_Stef has joined #openstack-lbaas | 05:07 | |
| openstackgerrit | min wang proposed openstack/octavia: Increase connection_max_retries to be 300 https://review.openstack.org/218613 | 05:21 |
| openstackgerrit | min wang proposed openstack/neutron-lbaas: Set up LBaas V2 tempest test gate against Octavia. https://review.openstack.org/209675 | 05:22 |
| *** Guest10903 has quit IRC | 05:42 | |
| *** I has joined #openstack-lbaas | 05:58 | |
| *** I is now known as Guest92023 | 05:58 | |
| openstackgerrit | Carlos Garza proposed openstack/octavia: Implementing EventStreamer https://review.openstack.org/218735 | 06:02 |
| *** clev is now known as clev-away | 06:15 | |
| *** ganeshna has joined #openstack-lbaas | 06:32 | |
| *** kiran-r has joined #openstack-lbaas | 06:43 | |
| *** kiran-r has quit IRC | 06:53 | |
| *** ganeshna has quit IRC | 06:56 | |
| *** nmagnezi has joined #openstack-lbaas | 07:23 | |
| *** chlong_ has quit IRC | 07:24 | |
| *** kiran-r has joined #openstack-lbaas | 07:38 | |
| *** bharathm has quit IRC | 07:41 | |
| *** nmagnezi has quit IRC | 08:33 | |
| *** nmagnezi has joined #openstack-lbaas | 08:34 | |
| *** apuimedo has joined #openstack-lbaas | 08:40 | |
| *** bharathm has joined #openstack-lbaas | 08:41 | |
| *** bharathm has quit IRC | 08:46 | |
| *** kiran-r has quit IRC | 08:57 | |
| *** kiran-r has joined #openstack-lbaas | 09:04 | |
| *** Guest92023 has quit IRC | 09:40 | |
| openstackgerrit | Stephen Balukoff proposed openstack/neutron-lbaas: Make pools independent of listeners https://review.openstack.org/218560 | 10:01 |
| *** Jian0612 has quit IRC | 10:03 | |
| *** bharathm has joined #openstack-lbaas | 10:31 | |
| *** apuimedo is now known as apuimedo|lunch | 10:34 | |
| *** bharathm has quit IRC | 10:35 | |
| *** h00327910__ has quit IRC | 10:43 | |
| *** nmagnezi_ has joined #openstack-lbaas | 10:56 | |
| *** nmagnezi has quit IRC | 10:59 | |
| *** chlong_ has joined #openstack-lbaas | 12:09 | |
| *** sballe has joined #openstack-lbaas | 12:14 | |
| *** bharathm has joined #openstack-lbaas | 12:20 | |
| openstackgerrit | venkata anil proposed openstack/neutron-lbaas: lbaasv2: validate tenant-id is non-empty in api https://review.openstack.org/219250 | 12:21 |
| *** bharathm has quit IRC | 12:24 | |
| *** woodster_ has joined #openstack-lbaas | 12:36 | |
| *** jschwarz has joined #openstack-lbaas | 12:50 | |
| *** amotoki has quit IRC | 12:57 | |
| openstackgerrit | Claudiu Belu proposed openstack/octavia: Removes unused posix-ipc requirement https://review.openstack.org/219065 | 13:04 |
| *** nmagnezi_ has quit IRC | 13:05 | |
| *** nmagnezi has joined #openstack-lbaas | 13:06 | |
| *** kiran-r has quit IRC | 13:40 | |
| *** kiran-r has joined #openstack-lbaas | 13:41 | |
| *** kiran-r has quit IRC | 13:41 | |
| *** bharathm has joined #openstack-lbaas | 14:08 | |
| *** bharathm has quit IRC | 14:13 | |
| *** amotoki has joined #openstack-lbaas | 14:20 | |
| *** alejandrito has joined #openstack-lbaas | 14:26 | |
| openstackgerrit | Armando Migliaccio proposed openstack/neutron-lbaas: Remove fall-back logic to service provider registration https://review.openstack.org/206221 | 14:54 |
| *** TrevorV has joined #openstack-lbaas | 15:00 | |
| *** TrevorV has quit IRC | 15:01 | |
| *** TrevorV has joined #openstack-lbaas | 15:03 | |
| *** diogogmt has joined #openstack-lbaas | 15:04 | |
| *** diogogmt has quit IRC | 15:07 | |
| *** diogogmt has joined #openstack-lbaas | 15:09 | |
| *** diogogmt has quit IRC | 15:11 | |
| *** pai15 has joined #openstack-lbaas | 15:11 | |
| *** yuanying has quit IRC | 15:13 | |
| pai15 | is there any way to create lbaas namespace on both controllers? It is getting created only on one controller | 15:13 |
| *** reedip has quit IRC | 15:14 | |
| *** Alex_Stef has quit IRC | 15:15 | |
| pai15 | and router namespace is getting created on another controller so as a result, ping to floating ip pointing to vip fails | 15:17 |
| *** diogogmt has joined #openstack-lbaas | 15:21 | |
| *** diogogmt has quit IRC | 15:24 | |
| *** reedip has joined #openstack-lbaas | 15:24 | |
| *** numans has quit IRC | 15:27 | |
| *** yuanying has joined #openstack-lbaas | 15:30 | |
| *** bharathm has joined #openstack-lbaas | 15:35 | |
| *** nmagnezi has quit IRC | 15:39 | |
| *** Aish has joined #openstack-lbaas | 15:45 | |
| *** jschwarz has quit IRC | 15:53 | |
| *** KunalGandhi has joined #openstack-lbaas | 15:59 | |
| *** madhu_ak has joined #openstack-lbaas | 15:59 | |
| *** ajmiller has joined #openstack-lbaas | 16:04 | |
| *** mlavalle has joined #openstack-lbaas | 16:17 | |
| *** vivek-ebay has joined #openstack-lbaas | 16:18 | |
| *** enikanorov__ has joined #openstack-lbaas | 16:18 | |
| *** enikanorov has quit IRC | 16:21 | |
| *** KunalGandhi has quit IRC | 16:22 | |
| *** pai15 has quit IRC | 16:22 | |
| *** pai15 has joined #openstack-lbaas | 16:23 | |
| *** pai15 has quit IRC | 16:26 | |
| *** pai15 has joined #openstack-lbaas | 16:26 | |
| *** madhu_ak has quit IRC | 16:30 | |
| *** ajmiller has quit IRC | 16:38 | |
| *** diogogmt has joined #openstack-lbaas | 16:46 | |
| *** vivek-ebay has quit IRC | 16:48 | |
| *** amotoki has quit IRC | 16:52 | |
| *** ajmiller has joined #openstack-lbaas | 16:52 | |
| *** ajmiller has quit IRC | 16:54 | |
| *** jschwarz has joined #openstack-lbaas | 16:57 | |
| *** mlavalle has quit IRC | 16:57 | |
| *** KunalGandhi has joined #openstack-lbaas | 16:57 | |
| *** TrevorV has quit IRC | 17:01 | |
| *** TrevorV2 has joined #openstack-lbaas | 17:01 | |
| *** ajmiller has joined #openstack-lbaas | 17:03 | |
| *** vivek-ebay has joined #openstack-lbaas | 17:09 | |
| pai15 | I I have 2 controlers, when I create a loadbalancer v2, should it created lbass namespace on 1 controller or both? | 17:11 |
| pai15 | let me correct the grammar: I have 2 controllers, when I create a loadbalancer v2, should it create lbass namespace on only 1 controller or both? | 17:12 |
| blogan | pai15: you have 2 agents? | 17:16 |
| blogan | pai15: 2 lbaas v2 agents running on two separate machines? | 17:16 |
| pai15 | blogan: yes - 2 controllers on different BMs, both running lbaasv2 agents | 17:17 |
| blogan | pai15: so the agent scheduler should choose only 1 of those to provision on | 17:18 |
| blogan | pai15: from what i remember | 17:18 |
| blogan | pai15: unless some HA mechanism got in that i dont recall | 17:18 |
| pai15 | it creates lbaas namespace on only 1 controller, but router namespace gets created on 2nd controller. Result it, floating ip associated to vip is not pingable | 17:19 |
| pai15 | is there any way to create lbaas namespace on both controllers? | 17:20 |
| *** pai15 has quit IRC | 17:20 | |
| *** pai15 has joined #openstack-lbaas | 17:21 | |
| *** bharathm has quit IRC | 17:21 | |
| *** pai15 has quit IRC | 17:22 | |
| *** pai15 has joined #openstack-lbaas | 17:22 | |
| xgerman | no | 17:25 |
| pai15 | ok | 17:25 |
| *** bana_k has joined #openstack-lbaas | 17:29 | |
| *** Aish has quit IRC | 17:29 | |
| *** TrevorV2 has quit IRC | 17:33 | |
| xgerman | but that FIP problem worries me | 17:33 |
| pai15 | Trying to thing what could be blocking it | 17:40 |
| xgerman | we have seen trouble with FIPs and DVR | 17:41 |
| *** Aish has joined #openstack-lbaas | 17:44 | |
| *** TrevorV2 has joined #openstack-lbaas | 17:45 | |
| *** Varun_Lodaya has joined #openstack-lbaas | 17:47 | |
| pai15 | xgerman: I am using not using DVR, using just the default one | 17:48 |
| xgerman | mmh | 17:48 |
| *** bharathm has joined #openstack-lbaas | 17:49 | |
| pai15 | which service provider should I be using?LOADBALANCERV2:Haproxy:neutron_lbaas.drivers.haproxy.plugin_driver.HaproxyOnHostPluginDriver:default | 17:53 |
| pai15 | or LOADBALANCERV2:Haproxy:neutron.services.loadbalancer.drivers.haproxy.synchronous_namespace_driver | 17:53 |
| xgerman | the later one in Kilo | 17:55 |
| xgerman | for L we recommend Octavia | 17:55 |
| *** madhu_ak has joined #openstack-lbaas | 17:57 | |
| pai15 | ok lbaas v2 that I setup on devstack kilo has the first one - so I used that. | 17:58 |
| pai15 | I will try to use the 2nd one. Once I make change to conf file, I just have to bound lbaas agent and do I need to bounce neutron-server ? | 17:58 |
| dougwig | octavia as ref, what's left? | 18:09 |
| xgerman | in meeting update later | 18:09 |
| johnsom | dougwig I have a couple of things. 1. the gate, we are working through upping the timeouts to allow the amphora boot(virtual in virtual without vt-x emulation is painful) | 18:11 |
| johnsom | 2. Health manager is in pretty good shape, working through bugs. The fail over flow is failing for us on REST, working on that now. | 18:12 |
| dougwig | are we down to just bugs that we think will get merged in the next 2 weeks? or are we missing any functionality or performance? | 18:12 |
| johnsom | I think the functionality is there, it is really down to testing and bug fixes. | 18:13 |
| johnsom | blogan Comments from your perspective? | 18:13 |
| *** pai15 has quit IRC | 18:14 | |
| johnsom | rm_work? | 18:14 |
| johnsom | Bueller | 18:14 |
| *** pai15 has joined #openstack-lbaas | 18:15 | |
| *** mlavalle has joined #openstack-lbaas | 18:16 | |
| *** ajmiller has quit IRC | 18:23 | |
| blogan | pai15: use that first one if you're looking to use teh namspace driver | 18:29 |
| blogan | johnsom, dougwig: i think we can get this all in decent sahpe in the next week | 18:30 |
| blogan | johnsom: not the active passive though, my opinion on that is its a lower priority and will take its own time to test | 18:31 |
| blogan | but we dont need it for ref impl | 18:31 |
| blogan | and id really like to get the active passive in too | 18:31 |
| blogan | but i just dont see it being feasible | 18:31 |
| blogan | but i could be wrong | 18:31 |
| *** jschwarz has quit IRC | 18:31 | |
| *** abdelwas has joined #openstack-lbaas | 18:32 | |
| xgerman | brogan you are wrong | 18:32 |
| xgerman | when do we need active-passive? | 18:33 |
| xgerman | I think we can get it in | 18:33 |
| johnsom | I think it is up for review now | 18:33 |
| blogan | xgerman: maybe but i dont think it should be priority before everything else we need to do | 18:33 |
| xgerman | yep, agreed | 18:34 |
| blogan | which is hwy i'm saying i dont think we will be able to get it in, but i could be wrong | 18:34 |
| abdelwas | It is ready for review | 18:34 |
| pai15 | blogan: so Haproxy:neutron_lbaas.drivers.haproxy.plugin_driver.HaproxyOnHostPluginDriver:default? | 18:35 |
| xgerman | yeah, it’s a big goal for us to get it in so we will put in the time/work necessary | 18:35 |
| *** Aish has quit IRC | 18:35 | |
| xgerman | maybe we can reprioritize things like the event queue in favor of active-passive | 18:35 |
| abdelwas | blogan In the revised patch I addressed all your prior comments | 18:36 |
| abdelwas | Also elaborated more in the commit message, to make reviewing easier | 18:37 |
| xgerman | yeah, abdelwas internship ends soon which motivates us to put our focus on it -- | 18:38 |
| pai15 | blogan: with that it creates lbaas namespace on only 1 controller, but router namespace gets created on 2nd controller. Result it, floating ip associated to vip is not pingable. | 18:38 |
| blogan | pai15: yeah but that 2nd provider will only create the haproxy process on whatever machine neutron-server is running on | 18:39 |
| blogan | xgerman, abdelwas: once the internship ends his code can't go in? | 18:39 |
| xgerman | no, it will be more challenging to reach him for bug fixes | 18:39 |
| xgerman | after all he will have classes and such | 18:39 |
| blogan | xgerman: well we shouldnt rely on him anyway to be the only person who can fix the bugs | 18:40 |
| pai15 | blogan: neutron-server is running on both controllers ( we don't have seperate neutron node) | 18:40 |
| abdelwas | technically no, I am done with classes. But I resumed working on research, even my wife struggle to preempt me | 18:40 |
| xgerman | agreed and we will have a handover — but he has 100% time dedicated to it (he doesn’t get sucked into in eternal projects like us) | 18:41 |
| blogan | pai15: oh well, still you want to use the HaProxyOnHostPluginDrievr because it is meant for that | 18:41 |
| abdelwas | s/struggle/struggles/ | 18:41 |
| blogan | ok so the way i see the priority right now is: | 18:42 |
| blogan | 1) failover flows | 18:42 |
| blogan | 2) health manager checkign for failover and intiating failover | 18:43 |
| pai15 | blogon: ok - then any pointers on the ping problem? any suggestion is appreciated for I am stumped on that. | 18:43 |
| blogan | 3) heart beats | 18:43 |
| blogan | pai15: im not sure i know how to solve that problem, but do you ahve the L3 agent running on both controllers as well? | 18:43 |
| xgerman | 2+3 work | 18:43 |
| pai15 | blogon: yes | 18:43 |
| xgerman | so as I see it we only have (1) to make work | 18:44 |
| TrevorV2 | xgerman, do you have a second? I'm not sure I agree with one of your comments in the failover review | 18:44 |
| blogan | pai15: yeah after that I've exhausted my limited knowledge after, the namespace driver hasn't really been focused on much since we came on board bc of our focus on octavia, the namespace driver is really just a stop gap until octavia | 18:44 |
| xgerman | TrevorV2 sure | 18:45 |
| blogan | xgerman: yeah but we still need to get it reviewed and tested out which will take some time, i'm in the process of doing that though | 18:45 |
| TrevorV2 | xgerman, well I don't think we should ignore failures in "nova delete" | 18:45 |
| blogan | xgerman: also testing it out for both ssh and rest, which adds more time | 18:45 |
| xgerman | cool!! | 18:45 |
| pai15 | blogon: thanks, I will put a message here when I find the root cause | 18:45 |
| TrevorV2 | Especially if it doesn't complete | 18:45 |
| TrevorV2 | If it doesn't FIND an instance, it makes sense to ignore it. | 18:45 |
| blogan | pai15: thanks, sorry i couldnt be anymore help, wish i could help ya out | 18:46 |
| TrevorV2 | However, remember, failover isn't after an amphora is deleted, it'll be when an amphora fails. | 18:46 |
| pai15 | np | 18:46 |
| TrevorV2 | So your "second" step in your test is erroneous | 18:46 |
| xgerman | well, the amphora can disappear and I would expect failover to succeed | 18:46 |
| xgerman | also a nova delete failure shouldn’t prevent a new amphora from coming up | 18:46 |
| TrevorV2 | blogan, help me out here. The port re-use requires the old amphora be deleted, right? | 18:47 |
| blogan | yes | 18:47 |
| blogan | well | 18:47 |
| xgerman | I would rather fail where we can’t reuse the port | 18:47 |
| abdelwas | blogan, xgeran I am willing to handover the whole project for continuing bug fixes. This is unquestionable. However, to the patch is good enough to get in, in my opinion. The remaining time is sufficient to address all concerns and solve major bugs. | 18:47 |
| blogan | it requires the port's device_owner/device_id to be cleared out, but that would mean 2 amphoraes would have the same IP | 18:47 |
| johnsom | TrevorV2 one of the failure modes could be the amphora got nova deleted somehow | 18:47 |
| *** pai15_ has joined #openstack-lbaas | 18:47 | |
| xgerman | or delete fails half-way but we can still use the port? | 18:48 |
| *** diogogmt has quit IRC | 18:48 | |
| xgerman | I am saying inability to delete doesn’t mean the port can’t be reused | 18:48 |
| TrevorV2 | johnsom, yeah the amphora being actually deleted is something I'm definitely okay with for "successful" failover. | 18:48 |
| TrevorV2 | But if it just "fails" to delete, but still exists, then routing to the new amphora would go crazy | 18:48 |
| blogan | TrevorV2: would require some testing, depends on how often a garp happens | 18:49 |
| johnsom | TrevorV2 Yeah, agree on that. The error German saw came from the amphora already being deleted via failure | 18:49 |
| blogan | TrevorV2: i have a feeling when the new amphora comes up with the new IP it'll garp out and the old one wouldn't ever garp out | 18:49 |
| johnsom | I think it's a check for "instance not found" exception | 18:49 |
| johnsom | (I happened to be watching him when he did it) | 18:50 |
| blogan | watching him through his window eh? | 18:50 |
| TrevorV2 | Yeah, johnsom xgerman I know, I had thought of that, but it required changes in the amphora driver that I thought I shouldn't do. Maybe I should then? | 18:50 |
| johnsom | Well, truth be told he was beating on me over config file settings being missing in devstack | 18:51 |
| blogan | xgerman sounds very abusive | 18:51 |
| *** pai15 has quit IRC | 18:51 | |
| *** pai15_ has quit IRC | 18:52 | |
| *** pai15 has joined #openstack-lbaas | 18:52 | |
| TrevorV2 | So do we think I should toss that in as a change in the failover review? Ignoring a "missing" instance when trying to delete? | 18:53 |
| TrevorV2 | Still failing for other reasons though? | 18:53 |
| johnsom | If the code is outside the changes for the failover flow I am ok with it being a bug in launchpad / separate patch | 18:54 |
| blogan | TrevorV2: is it just a matter of ignoring it? | 18:54 |
| *** diogogmt has joined #openstack-lbaas | 18:54 | |
| blogan | TrevorV2: or will there be other cascading changes? | 18:54 |
| xgerman | well, as a first step. We probably need to look at other error conditions but that can wait for M | 18:54 |
| xgerman | probably just a try except pass | 18:54 |
| TrevorV2 | blogan, basically if we try to delete a nova instance and it doesn't exist, it should be ignored, right blogan ? | 18:54 |
| johnsom | TrevorV2 yeah, I'm trying to dig into why. When I shutdown the agent (no heartbeats) the failover comes up with no vip network plugged in nova | 18:55 |
| blogan | xgerman: house keeping maanager will eventually clean all thsoe failed to delete amphorae up rgiht? theoretically at least | 18:55 |
| blogan | TrevorV2: do you have port information still on that nova instance? | 18:55 |
| xgerman | well, I want the failover to succeed and worry about the mess later | 18:55 |
| TrevorV2 | That's a good point blogan I guess not. | 18:55 |
| blogan | so how would we successfully transfer over the ports so we maintain the VIP | 18:56 |
| blogan | i have an idea but that would be best in another patch | 18:56 |
| xgerman | I think we need to understand the error conditions better | 18:56 |
| blogan | okay | 18:56 |
| xgerman | but as I said I am fine with deferring that to M | 18:57 |
| TrevorV2 | xgerman, I think what blogan just mentioned is important... If the nova instance doesn't exist, as in, was removed before the failover actually started, then we won't have appropriate networking information. | 18:57 |
| johnsom | That's a bug | 18:58 |
| xgerman | mmh, I have seen hardware crash and vms disappear | 18:58 |
| xgerman | so I am with johnsom | 18:58 |
| blogan | yeah so like i said that would fall into a nother patch to fix that | 18:58 |
| TrevorV2 | Sure, I'm on that page too, but then I'm not sure where we get that information | 18:58 |
| TrevorV2 | So blogan are you saying "failover review is okay as is" with that information? | 18:59 |
| blogan | bc that would be basically rebuilding the amphorae from scratch and requesting neutron to gie us the same IP for the vip port, which can come with its own pitfalls | 18:59 |
| TrevorV2 | For initial merge I mean? | 18:59 |
| johnsom | TrevorV2 I am not. It is not successfully failing over in scenarios where the original instance is still present | 18:59 |
| TrevorV2 | johnsom, is that with the ssh driver or with the rest driver? | 19:00 |
| blogan | johnsom: that sounds like it could merge and a follow up review could be done to fix the bug | 19:00 |
| *** Aish has joined #openstack-lbaas | 19:00 | |
| blogan | oh its not | 19:00 |
| blogan | sorry | 19:00 |
| blogan | nvm | 19:01 |
| TrevorV2 | If its with the REST driver then I was fairly certain it would fail, since I hadn't tested that. However, with the SSH driver it works just fine. blogan and I tested it pretty well | 19:01 |
| *** Aish has quit IRC | 19:01 | |
| blogan | TrevorV2: well we tested it on Friday and it worked, but i wanted to do more testing since it has been rebased | 19:01 |
| TrevorV2 | That is true. | 19:02 |
| *** Aish has joined #openstack-lbaas | 19:02 | |
| blogan | whcih is what im doing now | 19:02 |
| blogan | and wanted to test with rest driver as well | 19:02 |
| TrevorV2 | The thing I haven't done is included the port information to the REST driver. | 19:02 |
| TrevorV2 | Like I had to do with the SSH driver. | 19:03 |
| TrevorV2 | The REST driver also needs some of the failover methods I made in the SSH driver | 19:03 |
| TrevorV2 | So I can almost guarantee it'll fail. | 19:03 |
| TrevorV2 | I didn't think about that.. shit | 19:03 |
| blogan | TrevorV2: well that would be a problem then | 19:03 |
| TrevorV2 | Yeah. Still should work with SSH. | 19:03 |
| *** vivek-ebay has quit IRC | 19:04 | |
| *** vivek-ebay has joined #openstack-lbaas | 19:04 | |
| TrevorV2 | I'll get to the REST parity right now. | 19:05 |
| johnsom | Ok, so TrevorV2 are you going to work on REST? | 19:05 |
| johnsom | Ok. | 19:05 |
| TrevorV2 | Yeah, sorry about that... I was narrow minded about that... | 19:05 |
| johnsom | Note, there two open bugs on REST/HealthManager/UDP: | 19:05 |
| TrevorV2 | :( | 19:05 |
| johnsom | 1. You need to manually set the controller_ip_port_list = 172.24.4.1:5555 | 19:05 |
| johnsom | 2. You need to set the heartbeat_key | 19:06 |
| johnsom | Both are in launchpad | 19:06 |
| blogan | TrevorV2: 2 drivers is becoming a very untenable thing, but it is what it is, i tink come M we'll deprecate the ssh driver | 19:06 |
| *** KunalGandhi has quit IRC | 19:06 | |
| johnsom | Yeah, it would be nice to drop SSH. | 19:06 |
| *** KunalGandhi has joined #openstack-lbaas | 19:06 | |
| johnsom | blogan What are your thoughts about the default driver to kilo? | 19:06 |
| blogan | i do hate having to rebuild the image for rest drievr cahnges, but i assume i'd get used to it | 19:06 |
| johnsom | Sorry, Liberty | 19:07 |
| blogan | to kilo? | 19:07 |
| johnsom | Currently the health stuff only runs with REST as ssh doesn't copy over the agent | 19:07 |
| blogan | johnsom: i mean if ssh is probably going to be deprecated, doesnt make sense to have it as default | 19:07 |
| johnsom | Yeah, I agree. | 19:07 |
| TrevorV2 | johnsom, I have no idea what those 2 bugs are... Do they have to do with failover? | 19:08 |
| blogan | in fact if we do drop it i'd rather just drop it and not deprecate it | 19:08 |
| johnsom | I will fix the heartbeat_key and put up a patch to switch defaults. | 19:08 |
| blogan | johnsom: thats the issue with the call to home IP you were talking about? | 19:08 |
| johnsom | TrevorV2 No, they are UDP/health bugs | 19:08 |
| blogan | johnsom: needing the contrller to have an IP on th emgmt net | 19:09 |
| TrevorV2 | Oh okay, sorry I was thinking I had missed more steps ha ha ha | 19:09 |
| johnsom | blogan That is issue 1, and no, I though you were going to look at it | 19:09 |
| blogan | johnsom: i am, havent yet, but was going to ask more details about hte problem | 19:09 |
| blogan | johnsom: the packets do get back to the controller right? its just having a different SRC IP? | 19:10 |
| johnsom | blogan https://bugs.launchpad.net/octavia/+bug/1490033 - which you plan to work on. | 19:10 |
| openstack | Launchpad bug 1490033 in octavia "Devstack scripts need to enable lb-network controller IP" [Critical,New] | 19:10 |
| johnsom | Which leads to a fix for: https://bugs.launchpad.net/octavia/+bug/1489963 | 19:10 |
| openstack | Launchpad bug 1489963 in octavia "Amphora REST agent binds to all IPs" [High,New] - Assigned to Michael Johnson (johnsom) | 19:10 |
| TrevorV2 | Alright, I'll barrel into REST driver right now, hopefully have a patch soon. | 19:11 |
| johnsom | TrevorV2 Awesome | 19:11 |
| TrevorV2 | brb reconnecting | 19:11 |
| *** TrevorV2 has quit IRC | 19:11 | |
| blogan | johnsom: isn't the mgmt interface raised on the instance boot? so you can bind to just the mgmt ip on boot? | 19:12 |
| *** vivek-ebay has quit IRC | 19:12 | |
| *** sbalukoff has quit IRC | 19:14 | |
| *** vivek-ebay has joined #openstack-lbaas | 19:14 | |
| *** TrevorV has joined #openstack-lbaas | 19:16 | |
| johnsom | I think there is a routing issue. Maybe I am remembering that part wrong | 19:19 |
| blogan | johnsom: okay, we'll figure it out | 19:19 |
| xgerman | yeah!! | 19:19 |
| blogan | xgerman: after all of this failover stuff and heartbeat stuff is figured out | 19:20 |
| xgerman | it actually works pretty well | 19:20 |
| xgerman | just not failover + REST | 19:20 |
| blogan | well figured out for both drivers :) | 19:20 |
| xgerman | Ithought we decided to drop REST | 19:21 |
| blogan | what? | 19:21 |
| xgerman | SSH | 19:21 |
| blogan | REST? | 19:21 |
| blogan | lol | 19:21 |
| blogan | not until after L | 19:21 |
| xgerman | yeah, misspoke | 19:21 |
| xgerman | ok, but default should be REST? | 19:21 |
| xgerman | or do I need to book another vBrownbag for the REST story | 19:22 |
| blogan | i think so, assuming there aren't any issues | 19:22 |
| johnsom | I will put up a patch to make REST default | 19:25 |
| *** sbalukoff has joined #openstack-lbaas | 19:28 | |
| openstackgerrit | Michael Johnson proposed openstack/octavia: Implement UDP heartbeat sender and receiver https://review.openstack.org/201882 | 19:28 |
| openstackgerrit | min wang proposed openstack/neutron-lbaas: Set up LBaas V2 tempest test gate against Octavia. https://review.openstack.org/209675 | 19:30 |
| openstackgerrit | Michael Johnson proposed openstack/octavia: Set the Amphora REST driver as devstack default https://review.openstack.org/219421 | 19:33 |
| *** vivek-ebay has quit IRC | 19:34 | |
| *** mlavalle has quit IRC | 19:34 | |
| *** bharathm has quit IRC | 19:34 | |
| *** raginbajin has quit IRC | 19:34 | |
| *** harlowja has quit IRC | 19:34 | |
| *** whydidyoustealmy has quit IRC | 19:34 | |
| *** sc68cal has quit IRC | 19:34 | |
| *** rohara has quit IRC | 19:34 | |
| *** sc68cal has joined #openstack-lbaas | 19:34 | |
| *** rohara has joined #openstack-lbaas | 19:34 | |
| *** vivek-ebay has joined #openstack-lbaas | 19:35 | |
| *** mlavalle has joined #openstack-lbaas | 19:35 | |
| *** bharathm has joined #openstack-lbaas | 19:35 | |
| *** raginbajin has joined #openstack-lbaas | 19:35 | |
| *** whydidyoustealmy has joined #openstack-lbaas | 19:35 | |
| *** harlowja has joined #openstack-lbaas | 19:35 | |
| *** whydidyoustealmy has quit IRC | 19:35 | |
| *** harlowja has quit IRC | 19:35 | |
| *** harlowja has joined #openstack-lbaas | 19:35 | |
| johnsom | One more note for folks testing the REST driver with health manager: | 19:36 |
| *** whydidyoustealmy has joined #openstack-lbaas | 19:36 | |
| johnsom | Until it merges you need to edit /opt/stack/octavia/elements/amphora-agent/source-repository-amphora-agent and add a ref to the end of the amphora-agent line: refs/changes/82/201882/62 | 19:36 |
| johnsom | Before you stack.sh, otherwise the amphora image will have old agent code and fail | 19:37 |
| blogan | yes! | 19:37 |
| *** minwang2 has joined #openstack-lbaas | 19:38 | |
| xgerman | tripped me up yesterday as well | 19:38 |
| blogan | johnsom: does the health manager depend on that review? | 19:38 |
| blogan | johnsom: no other way around it look | 19:39 |
| blogan | s | 19:39 |
| johnsom | That review is the UDP sender | 19:39 |
| minwang2 | dougwig do you have any idea about overwirte tempest.conf content when it comes to upstream gate setting | 19:39 |
| johnsom | If you are only checking out up to health manager you can skip all three of those. | 19:39 |
| blogan | johnsom: so if testing the health manager requires some of those changes, can those changes be pulled out into the health manager review or even into its own review? | 19:39 |
| johnsom | You just need to hack the DB to trigger fail over | 19:39 |
| blogan | johnsom: yeah thats what im doing now | 19:40 |
| blogan | but since ive run into not being able to have 2 nova instances running at the same time again, its a PITA | 19:40 |
| blogan | grrr now stack issues | 19:40 |
| blogan | glance of all things | 19:41 |
| johnsom | Yeah, glance doesn't always stop after a clean.sh | 19:42 |
| johnsom | I always do a ps-ef | grep stack and kill anything that doesn't die | 19:42 |
| blogan | my devstack cleaner should have done that, but i might have not ran i | 19:43 |
| blogan | t | 19:43 |
| blogan | devstack cleaner script | 19:43 |
| *** harlowja has quit IRC | 19:47 | |
| *** vivek-ebay has quit IRC | 19:47 | |
| *** mlavalle has quit IRC | 19:47 | |
| *** bharathm has quit IRC | 19:47 | |
| *** raginbajin has quit IRC | 19:47 | |
| *** harlowja has joined #openstack-lbaas | 19:52 | |
| *** vivek-ebay has joined #openstack-lbaas | 19:52 | |
| *** mlavalle has joined #openstack-lbaas | 19:52 | |
| *** bharathm has joined #openstack-lbaas | 19:52 | |
| *** raginbajin has joined #openstack-lbaas | 19:52 | |
| *** crc32 has joined #openstack-lbaas | 20:14 | |
| *** crc32 has quit IRC | 20:16 | |
| *** Aish has quit IRC | 20:17 | |
| *** crc32 has joined #openstack-lbaas | 20:26 | |
| *** TrevorV has quit IRC | 20:27 | |
| rm_work | hmm, my IRC wasn't updating for a bit apparently, just got a slew of messages | 20:29 |
| rm_work | on shitty google-fi hotspot <_< | 20:29 |
| johnsom | At least you got them.... | 20:30 |
| rm_work | so blogan, we decided now to switch AGAIN to not using SSH driver at RAX? | 20:30 |
| rm_work | johnsom: yeah, bouncer FTW | 20:30 |
| rm_work | I am designating today to be test&review day | 20:31 |
| blogan | rm_work: no we're going to use it for initial EA, but in the middle of EA, remember we can't use it | 20:31 |
| rm_work | blogan: i do not remember that, but ok | 20:31 |
| rm_work | or rather, i remember talking about having to use like... a special version of REST that sits on the container host | 20:31 |
| rm_work | but then i remember finding out that we'd changed our minds and were going to use SSH and do it on Servicenet | 20:32 |
| blogan | rm_work: yes you do, we have to have an agent running on the hosts, which will take requests and forward them to the containers | 20:32 |
| blogan | rm_work: yes for the beginning, but thats not going to be scalable for LA or UA | 20:32 |
| rm_work | hmm k | 20:33 |
| blogan | and security reasons | 20:33 |
| rm_work | well, what reviews are top priority today? looks like trevorV still has work on the failover thing for REST | 20:33 |
| openstackgerrit | Carlos Garza proposed openstack/octavia: Implementing EventStreamer https://review.openstack.org/218735 | 20:33 |
| rm_work | i don't really understand the security issues, but that's probably irrelevant | 20:33 |
| johnsom | health manager and UDP would be good | 20:33 |
| johnsom | VRRP if you have time | 20:34 |
| rm_work | well both are stalled by Failover | 20:34 |
| johnsom | Yeah, but you can still review.... grin | 20:34 |
| *** Aish has joined #openstack-lbaas | 20:34 | |
| johnsom | Get that +2 all ready for when you can fully test.... | 20:34 |
| rm_work | lol | 20:36 |
| rm_work | erg, gotta fix my barbican devstack CR though, still racing to get that through before freeze... damnit | 20:37 |
| *** afranc202 has joined #openstack-lbaas | 20:46 | |
| rm_work | dougwig: alright, i made it a two-step | 20:47 |
| rm_work | kk will look at HM and UDP | 20:51 |
| rm_work | xgerman / minwang2: +2/+W on the HMService CR, it'll merge as soon as failovers does | 20:59 |
| johnsom | Awesome | 20:59 |
| xgerman | cool | 20:59 |
| rm_work | where is TrevorV actually?? | 20:59 |
| rm_work | weird that he's not on IRC, he was here earlier... | 20:59 |
| johnsom | He said he was going to go fix the REST api issues with fail over last I heard from him. | 21:00 |
| rm_work | yeah | 21:01 |
| crc32 | he's not here at work either. | 21:11 |
| openstackgerrit | Carlos Garza proposed openstack/octavia: Implementing EventStreamer https://review.openstack.org/218735 | 21:12 |
| *** Varun_Lodaya has quit IRC | 21:13 | |
| *** pai15 has quit IRC | 21:30 | |
| *** vivek-ebay has quit IRC | 21:30 | |
| *** vivek-ebay has joined #openstack-lbaas | 21:31 | |
| *** Varun_Lodaya has joined #openstack-lbaas | 21:33 | |
| *** Aish has quit IRC | 21:34 | |
| *** Aish has joined #openstack-lbaas | 21:35 | |
| *** Aish has quit IRC | 21:35 | |
| openstackgerrit | Michael Johnson proposed openstack/octavia: Increase connection_max_retries to be 300 https://review.openstack.org/218613 | 21:42 |
| sbalukoff | Anyone seen Evgeny here? | 21:42 |
| *** Varun_Lodaya has quit IRC | 21:43 | |
| openstackgerrit | Michael Johnson proposed openstack/neutron-lbaas: Set up LBaas V2 tempest test gate against Octavia. https://review.openstack.org/209675 | 21:47 |
| *** Varun_Lodaya has joined #openstack-lbaas | 21:50 | |
| *** Varun_Lodaya has quit IRC | 21:58 | |
| *** vivek-ebay has quit IRC | 21:59 | |
| rm_work | xgerman / johnsom / minwang2: who all is in the Seattle office? ajmiller and minwang2? | 22:02 |
| *** vivek-ebay has joined #openstack-lbaas | 22:02 | |
| minwang2 | me | 22:02 |
| minwang2 | AI is out of office today | 22:02 |
| rm_work | I am not positive on schedule yet but might be able to stop by and work from there *next* Tuesday | 22:02 |
| rm_work | will see | 22:02 |
| rm_work | :P | 22:02 |
| *** vivek-ebay has quit IRC | 22:03 | |
| *** vivek-ebay has joined #openstack-lbaas | 22:06 | |
| *** vivek-ebay has quit IRC | 22:10 | |
| *** mlavalle has quit IRC | 22:11 | |
| *** vivek-ebay has joined #openstack-lbaas | 22:17 | |
| *** vivek-ebay has quit IRC | 22:18 | |
| *** vivek-eb_ has joined #openstack-lbaas | 22:18 | |
| *** Varun_Lodaya has joined #openstack-lbaas | 22:24 | |
| *** alejandrito_ has joined #openstack-lbaas | 22:43 | |
| *** alejandrito_ has quit IRC | 22:45 | |
| *** alejandrito has quit IRC | 22:45 | |
| *** mlavalle has joined #openstack-lbaas | 22:47 | |
| *** pai15 has joined #openstack-lbaas | 22:48 | |
| *** pai15 has quit IRC | 22:52 | |
| *** apuimedo|lunch is now known as apuimedo | 22:58 | |
| *** bharathm has quit IRC | 23:01 | |
| *** KunalGandhi has quit IRC | 23:01 | |
| *** KunalGandhi has joined #openstack-lbaas | 23:01 | |
| johnsom | FYI - Canonical broke the ubuntu images for triple-o, so building the amphora image will fail (devstack stack.sh does this). | 23:03 |
| johnsom | https://bugs.launchpad.net/diskimage-builder/+bug/1490864 | 23:03 |
| openstack | Launchpad bug 1490864 in diskimage-builder "Ubuntu mirror file renaming broken "ubuntu" element" [Undecided,In progress] - Assigned to Lucas Alvares Gomes (lucasagomes) | 23:03 |
| openstackgerrit | min wang proposed openstack/octavia: Increase connection_max_retries to be 300 https://review.openstack.org/218613 | 23:04 |
| bana_k | johnsom: was breaking my head cause of it. any other way to make it work? | 23:06 |
| openstackgerrit | min wang proposed openstack/neutron-lbaas: Set up LBaas V2 tempest test gate against Octavia. https://review.openstack.org/209675 | 23:07 |
| johnsom | You can cherry pick the patch into your /opt/stack/diskimage-builder directory and restack (make sure you don't have reclone enabled in localrc) | 23:07 |
| bana_k | ok cool. I ll try that | 23:08 |
| bana_k | thanks | 23:08 |
| johnsom | Yeah, minwang2 and I have been puzzled by this today as well. Holding up our gate work | 23:08 |
| *** chlong_ is now known as chlong | 23:18 | |
| rm_work | urg that sucks | 23:20 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/neutron-lbaas: Updated from global requirements https://review.openstack.org/218910 | 23:22 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/octavia: Updated from global requirements https://review.openstack.org/219499 | 23:24 |
| openstackgerrit | min wang proposed openstack/octavia: Increase connection_max_retries to be 300 https://review.openstack.org/218613 | 23:39 |
| openstackgerrit | min wang proposed openstack/neutron-lbaas: Set up LBaas V2 tempest test gate against Octavia. https://review.openstack.org/209675 | 23:40 |
| *** KunalGandhi has quit IRC | 23:40 | |
| *** diogogmt has quit IRC | 23:40 | |
| *** KunalGandhi has joined #openstack-lbaas | 23:41 | |
| *** bharathm has joined #openstack-lbaas | 23:41 | |
| *** KunalGandhi has quit IRC | 23:41 | |
| *** diogogmt has joined #openstack-lbaas | 23:43 | |
| *** diogogmt has quit IRC | 23:45 | |
| openstackgerrit | Carlos Garza proposed openstack/octavia: Implementing EventStreamer https://review.openstack.org/218735 | 23:49 |
| *** crc32 has quit IRC | 23:53 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!