| *** mlavalle has quit IRC | 00:06 | |
| johnsom | https://review.openstack.org/#/c/207671/ | 00:06 |
|---|---|---|
| *** apuimedo has joined #openstack-lbaas | 00:08 | |
| *** SumitNaiksatam has quit IRC | 00:17 | |
| *** bharath has quit IRC | 00:18 | |
| openstackgerrit | min wang proposed openstack/neutron-lbaas: WIP - Octavia driver https://review.openstack.org/174114 | 00:37 |
| openstackgerrit | min wang proposed openstack/neutron-lbaas: Set up the gate for LBaaS v2 https://review.openstack.org/207674 | 00:37 |
| *** minwang2 has quit IRC | 00:54 | |
| *** SumitNaiksatam has joined #openstack-lbaas | 01:00 | |
| *** madhu_ak has quit IRC | 01:01 | |
| *** bharath has joined #openstack-lbaas | 01:19 | |
| *** vivek-ebay has quit IRC | 01:24 | |
| *** Tiancheng has joined #openstack-lbaas | 01:27 | |
| *** Tiancheng has quit IRC | 01:31 | |
| *** Tiancheng has joined #openstack-lbaas | 01:32 | |
| *** bana_k has quit IRC | 01:52 | |
| *** bharath has quit IRC | 02:03 | |
| *** bharath has joined #openstack-lbaas | 02:05 | |
| *** cing has joined #openstack-lbaas | 02:08 | |
| *** apuimedo has quit IRC | 02:10 | |
| *** bharath has quit IRC | 02:10 | |
| *** openstackgerrit has quit IRC | 02:31 | |
| *** openstackgerrit has joined #openstack-lbaas | 02:32 | |
| *** KunalGan_ has quit IRC | 02:32 | |
| *** rm_you| has quit IRC | 02:39 | |
| *** rm_you has joined #openstack-lbaas | 02:40 | |
| *** rm_you has joined #openstack-lbaas | 02:40 | |
| *** madhu_ak has joined #openstack-lbaas | 03:01 | |
| *** bharath has joined #openstack-lbaas | 03:07 | |
| *** bharath has quit IRC | 03:11 | |
| *** Tiancheng has quit IRC | 03:23 | |
| *** Tiancheng has joined #openstack-lbaas | 03:23 | |
| *** h00327910__ has quit IRC | 03:28 | |
| openstackgerrit | Brandon Logan proposed openstack/octavia: Decouple amp driver from network using new models https://review.openstack.org/205313 | 03:43 |
| blogan | rm_work, xgerman: ^^ | 03:45 |
| *** bana_k has joined #openstack-lbaas | 03:49 | |
| *** Tiancheng_ has joined #openstack-lbaas | 03:58 | |
| *** Tiancheng has quit IRC | 03:58 | |
| *** haigang has joined #openstack-lbaas | 04:07 | |
| *** fnaval has quit IRC | 04:08 | |
| rm_work | kk blogan | 04:14 |
| *** fnaval has joined #openstack-lbaas | 04:15 | |
| *** vivek-ebay has joined #openstack-lbaas | 04:21 | |
| *** ajmiller has quit IRC | 04:25 | |
| rm_work | blogan: you really didn't make that any clearer... https://review.openstack.org/#/c/205313/7/octavia/amphorae/drivers/noop_driver/driver.py,cm | 04:27 |
| rm_work | very confusing | 04:27 |
| rm_work | and id() wtf | 04:27 |
| rm_work | what is going on there | 04:27 |
| rm_work | was that a typo? | 04:27 |
| *** bharath has joined #openstack-lbaas | 04:33 | |
| *** bharath has quit IRC | 04:38 | |
| *** bharath_ has joined #openstack-lbaas | 04:54 | |
| *** bharath_ has quit IRC | 04:56 | |
| *** minwang2 has joined #openstack-lbaas | 05:15 | |
| *** rm_work is now known as rm_work|away | 05:16 | |
| *** madhu_ak has quit IRC | 05:23 | |
| *** ig0r_ has joined #openstack-lbaas | 05:24 | |
| *** numan has joined #openstack-lbaas | 05:29 | |
| *** KunalGandhi has joined #openstack-lbaas | 05:30 | |
| *** KunalGan_ has joined #openstack-lbaas | 05:49 | |
| *** KunalGandhi has quit IRC | 05:52 | |
| *** ig0r__ has joined #openstack-lbaas | 06:11 | |
| *** ig0r_ has quit IRC | 06:12 | |
| openstackgerrit | Merged openstack/neutron-lbaas: Tempest tests for Members using testscenarios https://review.openstack.org/180436 | 06:23 |
| *** vivek-ebay has quit IRC | 06:46 | |
| *** minwang2 has quit IRC | 06:49 | |
| *** jschwarz has joined #openstack-lbaas | 06:59 | |
| *** KunalGan_ has quit IRC | 07:05 | |
| *** jschwarz has quit IRC | 07:09 | |
| *** vjay4 has joined #openstack-lbaas | 07:16 | |
| *** vjay4 has quit IRC | 07:22 | |
| *** bana_k has quit IRC | 07:53 | |
| *** bharath has joined #openstack-lbaas | 07:57 | |
| *** bharath has quit IRC | 08:01 | |
| *** Tiancheng_ has quit IRC | 09:27 | |
| *** Tiancheng has joined #openstack-lbaas | 09:28 | |
| openstackgerrit | Merged openstack/neutron-lbaas: Tempest tests for Listener using testscenarios https://review.openstack.org/179818 | 09:33 |
| openstackgerrit | Merged openstack/neutron-lbaas: Killed existing downgrade rules in migration scripts https://review.openstack.org/206986 | 09:33 |
| *** Tiancheng has quit IRC | 09:36 | |
| openstackgerrit | Merged openstack/neutron-lbaas: Install HAProxy 1.5 on Ubuntu Devstack https://review.openstack.org/204719 | 09:38 |
| openstackgerrit | yaowei proposed openstack/neutron-lbaas: modify loadbalancer to support sorting and paging. Closes-Bug: 1452529 https://review.openstack.org/207809 | 10:12 |
| openstack | bug 1452529 in neutron "Lbaas object query doesn't support sorting or paging" [Wishlist,In progress] https://launchpad.net/bugs/1452529 - Assigned to yaowei (yaowei) | 10:12 |
| openstackgerrit | yaowei proposed openstack/neutron-lbaas: modify loadbalancer to support sorting and paging. Closes-Bug: 1452529 https://review.openstack.org/207809 | 10:15 |
| openstack | bug 1452529 in neutron "Lbaas object query doesn't support sorting or paging" [Wishlist,In progress] https://launchpad.net/bugs/1452529 - Assigned to yaowei (yaowei) | 10:15 |
| *** numan has quit IRC | 10:35 | |
| *** Tiancheng has joined #openstack-lbaas | 11:02 | |
| openstackgerrit | Evgeny Fedoruk proposed openstack/neutron-lbaas: Fixing Radware LBaaS v2 driver bugs https://review.openstack.org/207148 | 11:06 |
| *** ig0r__ has quit IRC | 11:15 | |
| *** ig0r_ has joined #openstack-lbaas | 11:19 | |
| *** Tiancheng has quit IRC | 11:28 | |
| *** cing has quit IRC | 11:51 | |
| *** amotoki has quit IRC | 12:03 | |
| *** numan has joined #openstack-lbaas | 12:07 | |
| *** haigang has quit IRC | 12:25 | |
| *** h00327910__ has joined #openstack-lbaas | 14:23 | |
| *** mlavalle has joined #openstack-lbaas | 14:28 | |
| *** fnaval has quit IRC | 14:39 | |
| *** ajmiller has joined #openstack-lbaas | 14:40 | |
| *** fnaval has joined #openstack-lbaas | 14:53 | |
| *** vivek-ebay has joined #openstack-lbaas | 14:55 | |
| *** chlong has quit IRC | 14:57 | |
| *** minwang2 has joined #openstack-lbaas | 14:57 | |
| *** diogogmt has joined #openstack-lbaas | 15:00 | |
| openstackgerrit | Elena Ezhova proposed openstack/neutron-lbaas: Refactor v2 scenario tests https://review.openstack.org/207944 | 15:01 |
| openstackgerrit | Elena Ezhova proposed openstack/neutron-lbaas: Add scenario test for load balancer's session persistence https://review.openstack.org/207945 | 15:01 |
| diogogmt | does anybody know the latest state of the lbaas v2 api? | 15:01 |
| diogogmt | it is flagged as experimental on the officlal docs, would it be stable in the kilo release? | 15:02 |
| *** vivek-ebay has quit IRC | 15:03 | |
| *** chlong has joined #openstack-lbaas | 15:11 | |
| *** numan has quit IRC | 15:16 | |
| *** johnsom_ has joined #openstack-lbaas | 15:22 | |
| *** bharath has joined #openstack-lbaas | 15:27 | |
| albertom | diogogmt: i was about ot ask the same | 15:27 |
| diogogmt | albertom: even for the lbaas v1, do you know a good reference for an overview of its components? including pool, vips and members? i want to understand the rationale of setting the external lb port on the VIP and not on the pool, also why the pool and the VIP accept a protocol attribute | 15:29 |
| albertom | crap, I saved the scripts that worked for me but not the sources | 15:29 |
| * albertom did it last week | 15:29 | |
| albertom | the protocol is for the health monitor | 15:30 |
| albertom | it accepts http, and ping | 15:30 |
| albertom | and i think other one i am missing | 15:30 |
| albertom | with ping, it will mark the node as UP if it replies the ping, (but that doesnt mean the http service is up and running) | 15:31 |
| albertom | so it could potentially route request to a host that appears to be UP but it isnt | 15:31 |
| albertom | with the http protocol you ensure the host is up by having a 200 response from the web server on that node | 15:32 |
| albertom | the VIP is the entry point, any request issued to the VIP will be redirected to the nodes in the pool, | 15:33 |
| albertom | and to make it accesible from the outside, you map a floating ip to the VIP | 15:33 |
| *** vivek-ebay has joined #openstack-lbaas | 15:34 | |
| albertom | according to this guy lbaasv2 is supported in kilo | 15:40 |
| albertom | https://chapter60.wordpress.com/2015/02/20/installing-openstack-lbaas-version-2-on-kilo-using-devstack/ | 15:40 |
| * albertom tries to enable | 15:40 | |
| diogogmt | albertom: what i’m saying is that both the pool and the VIP ask for a protocol (HTTP,HTTPS,TCP) — do you know where the pool protocol is used and the VIP protocol is used? why have two different protocols and not just on in the VIP? | 15:41 |
| albertom | oh got it | 15:41 |
| albertom | no idea :) | 15:41 |
| diogogmt | albertom: looks like the lbaas v2 api drop the concept of a VIP completely | 15:42 |
| diogogmt | albertom: they introduce listeners instead, similar to the ELB aws api | 15:42 |
| *** vivek-ebay has quit IRC | 15:42 | |
| diogogmt | albertom: even the lbaas v2 documention is pretty confusing, check this out: the POST /v2.0/lbaas/listeners says: protocol: The protocol the front end listens for. Must be TCP, HTTP, or HTTPS. | 15:43 |
| diogogmt | and the POST /v2.0/lbaas/pools says: protocol: The protocol this pool and its members listen for. Must be one of TCP, HTTP, or HTTPS | 15:44 |
| diogogmt | so if i understand correctly the pool defines the protocol and port the members will be listening on and the listener defined the external protocol the lb will be listening on? | 15:44 |
| diogogmt | albertom: but the POST /v2.0/lbaas/pools/{pool_id}/members still accepts a port: protocol_port The port that the member is listening to receive traffic. | 15:45 |
| diogogmt | what is the rationale for defining port numbers in three different places? in the pool, listener and member? | 15:45 |
| albertom | and in th ehealth monitor | 15:46 |
| albertom | neutron lb-healthmonitor-create --delay 5 --type HTTP --max-retries 3 --timeout 2 | 15:46 |
| diogogmt | albertom: i’m reading through the api docs and it seems that the LISTENER defined the external properties for the loadbalancer and the POOL defined the internal properties for the LB | 15:48 |
| diogogmt | but again, why have pools and listeners, why no just one entity defined both the external and internal properties? | 15:48 |
| diogogmt | anybody knows the logic behind the desing? are there any use cases validating the flow? | 15:48 |
| albertom | because the nodes can be listening in different ports ? | 15:48 |
| diogogmt | the members of the pool? in most use cases you could have the lb external port listening on port 80 and the internal members listening on port 8080 | 15:49 |
| diogogmt | i don’t see the use case where the members would be listening on two different port, like 8080 and 9090 | 15:49 |
| albertom | crazy people :P | 15:50 |
| diogogmt | albertom: do you know what i mean? the use case usually is for application that need to be loadbalanced — in the case of webapps the external LB listens on port 80 and then dsitributes the traffic to the members that could be listening on any port **but all members are listening on the same port** | 15:51 |
| pothole | diogogmt: Where you do you see port defined on the pool itself? | 15:53 |
| albertom | neutron lb-pool-create --lb-method ROUND_ROBIN --name mypool --protocol HTTP --subnet-id $subnetid | 15:53 |
| albertom | so you dont define the port on the pool, but the protocol | 15:53 |
| pothole | yea | 15:53 |
| albertom | you define the port on the vip/listener and on each member | 15:54 |
| pothole | correct | 15:54 |
| diogogmt | pothole: in the official v2 docs: http://developer.openstack.org/api-ref-networking-v2-ext.html | 15:54 |
| pothole | I am not seeing the port | 15:54 |
| diogogmt | pothole: check the POST /v2.0/lbaas/pools request attributes | 15:55 |
| diogogmt | protocol_port: The port on which the front end listens. Must be an integer in the range from 1 to 65535. | 15:55 |
| pothole | in the response | 15:55 |
| diogogmt | pothole: no, it lists right at the top | 15:55 |
| diogogmt | The caller of this operation must specify these pool attributes, at a minimum: | 15:55 |
| diogogmt | and then protocol_port is one of the attributes | 15:55 |
| pothole | Oh, i see. So this is most likely a copy-past error then. | 15:57 |
| pothole | We have plans on updating this a a lot of other things. We have been busy actually getting code to work. If you have time and feel like contributing please feel free to help update our documentation! | 15:58 |
| diogogmt | pothole: no worries, i’m just trying to understand the flow of the v2 api, currently we don’t have the v2 enabled on our environment so i’m makinf some design decisions of our app based on the docs | 16:00 |
| diogogmt | pothole: a question for you: on the v2 api, can a pool have multiple listeners? what is the realationship between the pool and the listeners? | 16:01 |
| pothole | Right now the functionality is not there, and is something planned for the near future, but, a listener can have multiple pools | 16:02 |
| fnaval | question: will neutron-lbaas support xml? | 16:03 |
| pothole | xml is dead fnaval | 16:03 |
| fnaval | pothole: k thats what i thought. thanks | 16:04 |
| pothole | welcome | 16:04 |
| diogogmt | pothole: how would it work in the scenario where the listener have multiple pools? correct me if i’m wrong but the listener is define the external port+protocol of the LB, lets say HTTP:80 — then the pool defined the group of instances that will be loadbalanced, the internal port is defined when the instance is added to the pool , example 8080— what is the use case for having multiple pools connected to a sing | 16:06 |
| diogogmt | listener? | 16:06 |
| xgerman | L7 | 16:08 |
| pothole | was away from comp, but what xgerman said | 16:09 |
| xgerman | yeah, L7 is planned for Liberty | 16:09 |
| diogogmt | pothole: xgerman what do you mean? | 16:10 |
| *** KunalGandhi has joined #openstack-lbaas | 16:10 | |
| pothole | Layer 7 load balancing | 16:10 |
| xgerman | L7 is tye abikity to send e.g. http://example.org/test to a different pool then https://example.com/test2 | 16:10 |
| xgerman | err http for the second one | 16:10 |
| pothole | https://wiki.openstack.org/wiki/Neutron/LBaaS/l7 | 16:11 |
| pothole | eh, that might be an old one. But gives some detail | 16:12 |
| diogogmt | pothole: thanks, but how does L7 switching relates to listeners having multiple pools? would one pool of instances be serving only static content while a different pool can be hosting a rest api? | 16:15 |
| diogogmt | in the document you linked it makes mentions to the rule and policy APIs | 16:16 |
| *** KunalGandhi has quit IRC | 16:16 | |
| albertom | how do i enable the v2 api ? | 16:16 |
| albertom | I ahve put | 16:16 |
| albertom | service_plugins = router,lbaasv2 | 16:16 |
| albertom | in neutron.conf but neutron-server fails to start with it | 16:16 |
| albertom | it is fine with service_plugins = router,lbaas | 16:17 |
| *** TrevorV has joined #openstack-lbaas | 16:17 | |
| xgerman | https://chapter60.wordpress.com/2015/02/20/installing-openstack-lbaas-version-2-on-kilo-using-devstack/ | 16:17 |
| xgerman | we also have some hit and miss vagrant script in the project | 16:18 |
| xgerman | (not really on top of updating that script) | 16:18 |
| openstackgerrit | Trevor Vardeman proposed openstack/octavia: Adding amphora failover flows https://review.openstack.org/202336 | 16:21 |
| pothole | diogogmt: That is indeed one of the valid use cases for L7 | 16:22 |
| pothole | Say you want to have a group of instance that serve the images for your site, you set up the pool to serve those images and create a rule that when the image is queried the pool serving the images will be used. Otherwise all other requests go to the other pool | 16:23 |
| pothole | There's a lot more advanced and complex cases that can be used here. L7 is quite neat | 16:24 |
| *** vivek-ebay has joined #openstack-lbaas | 16:25 | |
| diogogmt | pothole: ok, thanks buddy, that makes a lot more sense now — which place would you recommend i go to find the official docs for the features already implemented and the ones that are being worked on for the lbaas v1 and v2? | 16:25 |
| xgerman | lbaasv1 is not being worked on… and won’t see any updates | 16:26 |
| pothole | Well, unfortunately we dont really have that yet, or atleast not somewhere official i dont think. v1 is going aaway and nobody is working on it | 16:26 |
| pothole | yea :P | 16:26 |
| diogogmt | kk, i’ll just hang around the irc channel and keep my ears up for the new features being worked on then | 16:27 |
| pothole | L7 is being worked on, and you can always view gerrit/launchpad for tickets/patches/bugs | 16:27 |
| xgerman | also we occasionally make specs for the things we work on... | 16:28 |
| pothole | That is probably the most accurate/up-to-date representation of what we are doing and what we plan on doing | 16:28 |
| pothole | lol | 16:28 |
| pothole | :) | 16:28 |
| diogogmt | pothole: xgerman since you guys are actively working on the lbaas v2 you could probably be able to answer this question: on AWS ELB the offer a feature where you can only have one pool of servers but multiple listeners, for example, the domain test-lb-subnet-A3801-1732967080.us-east-1.elb.amazonaws.com could have one listener mapping the external port 80 to internal port 8080 then another listener mapping external p | 16:29 |
| diogogmt | 90 to internal port 9090, however, the loadbalancer only has ONE pool of instances — do you see a use case for having ONE pool and MULTIPLE listeners? | 16:29 |
| diogogmt | the AWS ELB approach seems to differ from the lbaas v2 L7 switching where ONE listeners can have MULTIPLE pools, right? | 16:30 |
| pothole | That is a different thing and im not quite sure weve thought of this case | 16:30 |
| pothole | so its the same instances listening on multiple ports | 16:31 |
| pothole | so one pool, but the front end listens on different ports and balances them accordingly | 16:31 |
| pothole | well | 16:31 |
| pothole | you could add same member with different port and acheieve the same thing i think | 16:32 |
| xgerman | eventually we were talking many-2-many for that relationship but we wanted to go i steps since it complicates things by a lot | 16:32 |
| xgerman | so one pool multiple listeners is in the plans but that requires us to rethink some other things | 16:33 |
| pothole | Yea, then i dont remember this. Must have not talked about it in a while >< | 16:33 |
| xgerman | yep, we haven’t touched that in a while. Sam from radar is a big proponent of that | 16:33 |
| pothole | ah | 16:33 |
| xgerman | s/radar/radware | 16:33 |
| xgerman | we did some work to decouple stats as a preparation but yeah… pretty low priority at the moment | 16:34 |
| diogogmt | pothole: xgerman i’ve been trying to think of use cases for having a many to many relation ship but couldn’t come up with one | 16:34 |
| pothole | yea, what's the use of this case? | 16:34 |
| pothole | I dont quite see it | 16:34 |
| pothole | the pool would have the port then? | 16:34 |
| pothole | and it would override the members port? | 16:35 |
| xgerman | you want two listeners use the same pool and one listener have more than one pool — so many-2-many | 16:35 |
| pothole | otherwise, couldnt we just add same member, different port and achieve the same thing? | 16:35 |
| diogogmt | as far as the AWS ELB the listener defines both the internal and external port, the members are associated directly to their loadbalancer resource | 16:35 |
| pothole | interesting | 16:35 |
| diogogmt | xgerman: where would the VIP be defined, in the pool or in the listener? | 16:36 |
| xgerman | VIP is on the load balancer | 16:36 |
| xgerman | LB -> 1…N Listener -> Pool | 16:36 |
| diogogmt | xgerman: ok, got it, so in the lbaas v2 the VIP entity of v1 has been merged into a loadbalancer entity? | 16:38 |
| xgerman | sort of we still have the VIP but it’s just an IP now… port is on listener | 16:42 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/octavia: Updated from global requirements https://review.openstack.org/206828 | 16:42 |
| *** crc32 has joined #openstack-lbaas | 16:47 | |
| *** amit213 has quit IRC | 16:50 | |
| *** vjay4 has joined #openstack-lbaas | 16:50 | |
| *** amit213 has joined #openstack-lbaas | 16:50 | |
| johnsom_ | FYI, the Zuul/gerrit outages have impacted the sonar job. I have retriggered the two test runs that were impacted. | 16:50 |
| johnsom_ | There were >1300 jobs in the merge gate in zuul this morning and it is not happy... | 16:51 |
| *** vjay4 has quit IRC | 16:56 | |
| xgerman | will only get worse the closer we get to Liberty-3 | 16:57 |
| blogan | zuul gets angry | 16:58 |
| johnsom_ | Ok, Sonar is caught up. | 16:58 |
| johnsom_ | As xgerman said, merge often my friend... | 16:58 |
| blogan | zuul should die down in a few days | 16:58 |
| blogan | or just die | 16:58 |
| albertom | hm | 17:00 |
| albertom | I enabled v2 | 17:00 |
| albertom | but it doesnt show in horizon | 17:00 |
| albertom | is that work in progress too ? | 17:00 |
| xgerman | yep, horizon is slated for Liberty, too | 17:01 |
| albertom | or did i miss something ? | 17:01 |
| albertom | so v1 if you want horizon | 17:01 |
| albertom | ok | 17:01 |
| blogan | albertom: are yall using v1 in a production env somewhere? | 17:02 |
| blogan | albertom: or just testing it all out? | 17:02 |
| albertom | testing out | 17:02 |
| *** rohara has joined #openstack-lbaas | 17:03 | |
| *** TrevorV2 has joined #openstack-lbaas | 17:04 | |
| *** TrevorV has quit IRC | 17:06 | |
| *** chlong has quit IRC | 17:09 | |
| *** bana_k has joined #openstack-lbaas | 17:23 | |
| *** crc32 has quit IRC | 17:44 | |
| *** crc32 has joined #openstack-lbaas | 17:52 | |
| *** crc32 has quit IRC | 17:54 | |
| *** crc32 has joined #openstack-lbaas | 17:54 | |
| *** SumitNaiksatam has quit IRC | 17:55 | |
| *** crc32 has quit IRC | 17:56 | |
| *** crc32 has joined #openstack-lbaas | 18:02 | |
| *** madhu_ak has joined #openstack-lbaas | 18:08 | |
| openstackgerrit | Merged openstack/octavia: Updated from global requirements https://review.openstack.org/206828 | 18:19 |
| *** TrevorV2 has quit IRC | 18:39 | |
| *** madhu_ak_ has joined #openstack-lbaas | 18:41 | |
| *** madhu_ak has quit IRC | 18:42 | |
| *** madhu_ak_ is now known as madhu_ak | 18:46 | |
| *** TrevorV has joined #openstack-lbaas | 18:52 | |
| *** TrevorV has quit IRC | 18:54 | |
| *** TrevorV has joined #openstack-lbaas | 18:54 | |
| *** vivek-eb_ has joined #openstack-lbaas | 18:58 | |
| *** vivek-ebay has quit IRC | 18:59 | |
| *** ig0r_ has quit IRC | 19:26 | |
| *** mmdurrant has joined #openstack-lbaas | 19:27 | |
| openstackgerrit | Phillip Toohill proposed openstack/octavia: Updating cert_parser for cert loading https://review.openstack.org/208035 | 19:29 |
| mmdurrant | I’m confused about something in LBaaSv2. If I create my load balancer on my private network, how can I give it a floating IP so it can talk to the world as well as its pool members? | 19:29 |
| mmdurrant | the command for ip-floating-add only accepts an IP address and instance | 19:30 |
| *** openstack has joined #openstack-lbaas | 19:33 | |
| *** vivek-ebay has joined #openstack-lbaas | 19:34 | |
| *** madhu_ak has quit IRC | 19:34 | |
| albertom | and which is the private ip for the pool ? | 19:35 |
| *** vivek-eb_ has quit IRC | 19:36 | |
| mmdurrant | I don’t understand your question - pools don’t have IPs as far as I know. Are you asking what subnet I’m using when I create the load balancer versus the members? | 19:37 |
| albertom | no, i have the same question as you | 19:38 |
| albertom | in v1, a private ip is assigned to the VIP | 19:38 |
| albertom | which is assosiated with the pool | 19:38 |
| albertom | and the the floating ip is assosiated with the VIP | 19:38 |
| mmdurrant | Oh, OK. Exactly - there has to be a private IP interface on that subnet to be able to route traffic between the 2. | 19:38 |
| *** vivek-eb_ has joined #openstack-lbaas | 19:38 | |
| albertom | I have the pool now with the members but i cant see a way to comunicate with them | 19:39 |
| mmdurrant | Formerly when we created a VIP it would get an IP from the private subnet and you could assign a VIP to expose it to the world. | 19:39 |
| *** vivek-ebay has quit IRC | 19:40 | |
| mmdurrant | It seems a loadbalancer would have to create 2 SDN ports in this scenario - 1 to talk to the world, 1 to talk to the pool members, similar to routers. | 19:41 |
| mmdurrant | I can get it to work if I create the lb on the private-subnet but then I have to create routing entries that allow me to reach the private subnet and that seems less than optimal. | 19:42 |
| *** TrevorV has quit IRC | 19:43 | |
| albertom | I created 3 machines ( 10.0.10.13 - 10.0.10.15 ) | 19:45 |
| albertom | and after creating the listener and pool | 19:45 |
| albertom | I have a 10.0.10.16 ip | 19:45 |
| albertom | which i guess it is the VIP | 19:46 |
| albertom | but i found it by guess, i dont know how to tell that is the vip | 19:46 |
| *** rm_work|away is now known as rm_work | 19:49 | |
| rm_work | albertom / mmdurrant: blogan would be able to answer that one | 19:51 |
| blogan | albertom: how did you create the lb? | 19:51 |
| rm_work | aaand there he is | 19:51 |
| mmdurrant | Bam thank you | 19:51 |
| albertom | neutron lbaas-loadbalancer-create --name lb1 demo-subnet | 19:51 |
| mmdurrant | Sup Brandon | 19:51 |
| blogan | albertom: that should return a vip back to you | 19:51 |
| blogan | sup mmdurrant | 19:52 |
| mmdurrant | I’m going to sit back and listen to your conversation with albertom and see where I went wrong. | 19:52 |
| albertom | hmm | 19:52 |
| mmdurrant | Same ol’… dougwig’s gone fishing so I’m working with QA/playing with LBs/wondering why this isn’t setup intuitively like I think it should be | 19:52 |
| albertom | lol | 19:52 |
| albertom | blogan: you are right | 19:52 |
| albertom | neutron lbaas-loadbalancer-show lb1 | 19:52 |
| albertom | that shyows the vip address and port | 19:53 |
| blogan | albertom: the create call returns the vip as well | 19:53 |
| *** TrevorV has joined #openstack-lbaas | 19:53 | |
| xgerman | so you guys are dougwig’s colleagues? Anything we need to know to tease him next time we see him :-) | 19:53 |
| blogan | albertom: and if you look at the details of that port, it should show the device owner as neutron:LOADBALANCERV2 | 19:54 |
| blogan | mmdurrant: intuitive setup is very subjective :) | 19:54 |
| blogan | mmdurrant: especially if you work with dougwig | 19:54 |
| xgerman | lol | 19:54 |
| mmdurrant | Truer words have never been spoken | 19:54 |
| mmdurrant | I’m sure his phone is lighting up right now with mentions of his name | 19:55 |
| blogan | i hope he's checking it and it falls in the river | 19:55 |
| xgerman | isn’t he hunting? | 19:56 |
| blogan | fly fishing | 19:56 |
| xgerman | that’s what he told us — for all I know he might be in Africa shooting Zebras | 19:56 |
| crc32 | Or Cecils | 19:57 |
| blogan | is he a dentist as a side job? | 19:58 |
| mmdurrant | LOL | 19:58 |
| * mmdurrant reminds himself to ask why people give him such a hard time | 19:58 | |
| crc32 | I don't know. I have a hard time feeling sorry for a lion. | 19:58 |
| crc32 | or scorpians | 19:58 |
| crc32 | or system engineers | 19:59 |
| mmdurrant | ^ | 19:59 |
| dougwig | Hmm | 19:59 |
| mmdurrant | Hahahahah see I told you | 19:59 |
| mmdurrant | Get back to fishing or something’s going to end up in the river. | 19:59 |
| blogan | or a bear will have him for dinner | 19:59 |
| xgerman | or a dentist will take all his teeth | 20:00 |
| *** vivek-eb_ has quit IRC | 20:01 | |
| *** vivek-ebay has joined #openstack-lbaas | 20:01 | |
| *** madhu_ak has joined #openstack-lbaas | 20:04 | |
| rm_work | blogan: switched to +2 on your https://review.openstack.org/#/c/205313/7 but i am still not happy >_> | 20:04 |
| * rm_work will never be happy | 20:05 | |
| blogan | rm_work: ill take a +2 and your unhappiness, its almost like a +4 | 20:06 |
| blogan | rm_work: im running into an issue where the cryptography lib method is not accepting unicode, if i do a str() around the variable, will that also work with py3? or do i need to do a six method? | 20:07 |
| *** vivek-ebay has quit IRC | 20:07 | |
| rm_work | errr | 20:09 |
| rm_work | so | 20:09 |
| rm_work | yes, str() should be ok | 20:10 |
| blogan | mmdurrant: you ever get an answer to your floating ip to vip question? | 20:16 |
| mmdurrant | Nope. Digging around looking at SDN stuff, it appears all the right elements are in the right place. | 20:16 |
| mmdurrant | With the exception of my load balancer having a port to communicate to the world on and a port to communicate to the backend nodes on | 20:17 |
| blogan | mmdurrant: so the floating ip create call takes a port_id, which you would give it the vip_port id | 20:17 |
| blogan | mmdurrant: but you that port needs to be created on a network that is connected by a router tagged as external (or soemthing like that) | 20:17 |
| mmdurrant | I was using the openstack client and not the neutron client. That explains a lot. | 20:19 |
| blogan | mmdurrant: ahh | 20:19 |
| mmdurrant | “Why does this not accept the parameters he’s talking about? Surely I’m doing something wrong…” | 20:20 |
| blogan | mmdurrant: yeah i haven't used that client yet, its not "ready" last i heard, but it'd still take me a while to move over since i hate relearning | 20:20 |
| mmdurrant | Yeah I’m on the fence about whether to have a monolithic openstack client that does everything versus separate clients that each do their own thing. Perhaps creating the openstack client as a wrapper around the others | 20:21 |
| blogan | lol last i heard thats what it does | 20:21 |
| blogan | but i might be wrong now | 20:21 |
| *** TrevorV2 has joined #openstack-lbaas | 20:21 | |
| blogan | but yeah especially with the big tent, if the monolithic client really wanted to be the one to rule them all client, it'd have to support a ton of projects | 20:21 |
| *** TrevorV has quit IRC | 20:23 | |
| mmdurrant | got it | 20:24 |
| mmdurrant | Yeah I agree completely - they’d have to have some kind of shim/wrapper to ensure that they’re not reproducing functionality in the monolithic client. I like the idea of a one-stop command shop for everything I need to do, but I don’t. | 20:25 |
| mmdurrant | “got it” = got the port binding to work correctly. | 20:25 |
| mmdurrant | my LB now has a public VIP and it’s sround-robining exactly as expected. | 20:26 |
| xgerman | yeah!! | 20:27 |
| albertom | cool :D | 20:27 |
| mmdurrant | Now to get my devstack that’s uilt on a production openstack to talk to the rest of the world… :) | 20:27 |
| mmdurrant | Thanks for the help blogan, sometimes a person has to ask stupid questions to realize they’re doing something incorrectly. ;) | 20:28 |
| blogan | mmdurrant: haha np, i'm sure you get a lot of that from dougwig | 20:28 |
| blogan | mmdurrant: but what client to use and openstackclient not being on par with neutron client is a pretty shitty ux | 20:29 |
| mmdurrant | I agree completely, it’s awful. That’s been a giant PITA for me training a QA person how to do devstack as the parity isn’t there. We had to convert a bunch of commands to use the openstack client because they were no longer available in the client recommended by the docs. | 20:30 |
| blogan | rm_work: cryptography.x509.load_pem_x509_certificate doesn't accept unicode in py2, and doesn't accept str in py3 | 20:36 |
| rm_work | rofl | 20:38 |
| rm_work | bitch to reaperhulk in the crypto channel? | 20:38 |
| rm_work | here, i'll do it | 20:38 |
| blogan | im sure it has something to do with the str changes in py3, but i don't want to be forced to do if else of py version | 20:39 |
| rm_work | blogan: it's supposed to take bytes apparently | 20:40 |
| rm_work | six.b() | 20:40 |
| rm_work | which may require an encode | 20:41 |
| blogan | lol doesn't work in py2 apparently | 20:41 |
| *** TrevorV2 has quit IRC | 20:41 | |
| blogan | unless im doing it wrong which is quite possible | 20:41 |
| blogan | highly likely actually | 20:41 |
| rm_work | err nm | 20:42 |
| rm_work | cert_string.encode('ascii') | 20:42 |
| rm_work | that | 20:42 |
| rm_work | do that | 20:42 |
| rm_work | six.b() does the wrong thing according to them | 20:42 |
| blogan | ok that works for both, the encode to ascii | 20:45 |
| rm_work | yep | 20:45 |
| rm_work | cool | 20:45 |
| rm_work | there may be several places where that needs to be fixed | 20:45 |
| rm_work | where i just blindly did str() conversions | 20:45 |
| blogan | have to change the tests to not make the constants of certs to just be str literals, not bytes | 20:45 |
| rm_work | to fix py3 issues | 20:45 |
| rm_work | yeah | 20:46 |
| rm_work | k | 20:46 |
| rm_work | there were like 5 places or so i think | 20:46 |
| rm_work | where i did that | 20:46 |
| rm_work | even in the real code, you might want to check | 20:46 |
| blogan | teh str() conversion? | 20:46 |
| rm_work | yes | 20:46 |
| blogan | i dont see it in the cert_parser.py | 20:49 |
| blogan | there was some six.b's i had to remove | 20:49 |
| rm_work | k | 20:51 |
| rm_work | maybe that is what had made it in | 20:52 |
| openstackgerrit | Brandon Logan proposed openstack/octavia: Hooking TLS together https://review.openstack.org/185540 | 20:52 |
| blogan | there updated | 20:52 |
| rm_work | lol k | 20:52 |
| *** minwang2 has quit IRC | 20:52 | |
| *** minwang2 has joined #openstack-lbaas | 20:53 | |
| *** vivek-ebay has joined #openstack-lbaas | 21:04 | |
| *** vivek-ebay has quit IRC | 21:08 | |
| blogan | xgerman: what things be buggy? | 21:10 |
| blogan | crc32, minwang2, xgerman: http://paste.openstack.org/show/406637/ | 21:11 |
| blogan | structure of the heartbeat payload | 21:11 |
| blogan | feedback? | 21:11 |
| minwang2 | can you explain a bit about the status | 21:12 |
| minwang2 | tx,rs | 21:12 |
| minwang2 | rx | 21:12 |
| *** rm_you has quit IRC | 21:13 | |
| blogan | minwang2: status would just be whether its up or down for now, active or not active (UP, DOWN are probably better) | 21:13 |
| blogan | tx,rx are the outgoing and incoming bandwidth bytes | 21:13 |
| minwang2 | count? | 21:14 |
| blogan | whatever haproxy returns really | 21:14 |
| blogan | im sure they're counters though | 21:15 |
| *** divya has joined #openstack-lbaas | 21:17 | |
| xgerman | yep | 21:17 |
| blogan | xgerman: that looks fine to you? | 21:18 |
| openstackgerrit | Merged openstack/octavia: Decouple amp driver from network using new models https://review.openstack.org/205313 | 21:18 |
| divya | Hi, I am trying out HaProxy Namespace driver with Lbaasv2 plugin and had couple of questions | 21:19 |
| xgerman | so I return for members from the stats socket code I wrote: | 21:19 |
| xgerman | https://www.irccloud.com/pastebin/hP9z1RTw/ | 21:19 |
| xgerman | so we should have a pool section IHMO | 21:20 |
| divya | I am trying to associate floatingip to the loadbalancer VIP and when i try wget on VIP it doesnt seem to work ... | 21:20 |
| blogan | xgerman: does haproxy return a status for a backend? or just for the members? | 21:21 |
| divya | ping on the VIP floating IP works .. Is there any special configurations needs to be done to make the floatingip on VIP work? | 21:21 |
| blogan | divya: you're pinging the floating ip? or the vip that lbaas gave you? | 21:22 |
| divya | i tried pinging the floating ip that i assigned to the VIP | 21:22 |
| xgerman | it returns status for all of them | 21:22 |
| xgerman | I am just massaging the data to fir nicely | 21:22 |
| blogan | xgerman: so the heartbeat code will have to evaluate all of the members statuses to determine the pool status? | 21:23 |
| blogan | divya: and pinging the floatingip did not work? | 21:23 |
| xgerman | no, the pool status is extra | 21:23 |
| blogan | divya: but pinging the vip does? | 21:23 |
| xgerman | https://github.com/openstack/octavia/blob/master/octavia/amphorae/backends/utils/haproxy_query.py | 21:23 |
| xgerman | pool = backend | 21:23 |
| divya | blogan: ping works on both VIP and floating IP | 21:24 |
| xgerman | but we can add to that util area to return whatever we want | 21:25 |
| divya | blogan: and wget on VIP forwards the requests to the pool memebers | 21:25 |
| blogan | divya: ahh, okay, but not the floating ip? | 21:25 |
| divya | blogan: but wget on FIP doesn;t seem to forward the requests | 21:25 |
| divya | blogan: i don't see anything on the haproxy logs | 21:25 |
| blogan | divya: so floating ip must be hittng some kind of security group rule | 21:25 |
| divya | blogan: where do you have to change the sg settings for VIP? | 21:27 |
| blogan | divya: if you're able to wget on the VIP and that works correctly, its not a security group rule on the VIP | 21:28 |
| blogan | but the floating ip is just a nat to teh vip, so something is not working on that piece | 21:28 |
| divya | ok | 21:29 |
| blogan | xgerman: ohhh im dumb, this is already in octavia | 21:29 |
| divya | blogan: when i do a port show on VIP, it has default sg associated with it .. But port show on FIP shows sg as empty | 21:29 |
| divya | blogan: is it required to associate default sg to FIP? | 21:30 |
| *** madhu_ak has quit IRC | 21:31 | |
| *** superflyy has joined #openstack-lbaas | 21:31 | |
| bana_k | do we need to open up the ports for http/https ? | 21:31 |
| openstackgerrit | min wang proposed openstack/neutron-lbaas: Set up the gate for LBaaS v2 https://review.openstack.org/207674 | 21:32 |
| *** madhu_ak has joined #openstack-lbaas | 21:32 | |
| divya | blogan: when i do wget on FIP, is there a way to check if haproxy is receiving the request ? | 21:32 |
| crc32 | ok. Thats what I'll be doing. :/ | 21:32 |
| blogan | divya: try to add a security group rule for port 80 (or whatever port you're doing) on the floatingip | 21:32 |
| blogan | divya: what driver are you using? | 21:33 |
| blogan | divya: oh namespace driver | 21:33 |
| blogan | divya: you can do a tcpdump in the nlbaas-{loadbalancer-uuid} namespace and see if the requests get tehre | 21:34 |
| divya | blogan: i am using neutron.services.loadbalancer.drivers.haproxy.namespace_driver.HaproxyNSDriver | 21:34 |
| *** crc32 has quit IRC | 21:34 | |
| divya | blogan: Thanks will try that .. | 21:35 |
| bana_k | divya: if you are using a devstack setup | 21:36 |
| bana_k | nova secgroup-add-rule default icmp -1 -1 0.0.0.0/0; nova secgroup-add-rule default tcp 22 22 0.0.0.0/0 | 21:36 |
| bana_k | nova secgroup-add-rule default tcp 80 80 0.0.0.0/0 | 21:36 |
| bana_k | these should do | 21:36 |
| *** rm_you has joined #openstack-lbaas | 21:38 | |
| *** rm_you has joined #openstack-lbaas | 21:38 | |
| divya | bana_k: Thanks .. yes i am using devstack setup .. will try these | 21:39 |
| divya | blogan: trying running this command ip netns exec qlbaas-d22d5aa3-ec27-4f06-af70-b6789863cfae tcpdump -i any . is this the right command? | 21:39 |
| divya | blogan: i don't see any requests even when wget on VIP is returning response | 21:40 |
| bana_k | can u try this | 21:42 |
| bana_k | sudo ip netns exec qrouter-XXXXXXXX_id curl -v vip_ip* | 21:42 |
| bana_k | get the router id | 21:42 |
| bana_k | there should be only one router if u have not created any | 21:42 |
| openstackgerrit | Madhusudhan Kandadai proposed openstack/neutron-lbaas: Updation for V2 scenario tests https://review.openstack.org/207196 | 21:46 |
| blogan | divya: yeah try what bana_k suggested, that'll let you knwo if the router can hit the vip | 21:46 |
| madhu_ak | +1 as bana_k suggested. Or you can do this way: sudo ip netns exec qrouter-XX /bin/bash | 21:47 |
| madhu_ak | then ping <vip_IP> | 21:47 |
| blogan | madhu_ak: don't make it easy to run netns commands | 21:47 |
| blogan | it should always be painful | 21:47 |
| johnsom_ | +1 | 21:48 |
| xgerman | +1 | 21:48 |
| madhu_ak | oh, why is it so? | 21:48 |
| blogan | you dont like pain? | 21:49 |
| madhu_ak | heh | 21:49 |
| blogan | get out of openstack! | 21:49 |
| madhu_ak | to be honest, when working on openstack for the first time, took longer time to understand than expected | 21:50 |
| johnsom_ | You understand OpenStack?????? | 21:50 |
| johnsom_ | Grin | 21:50 |
| blogan | thats probably the nicest statement ever uttered about getting involved in openstack | 21:51 |
| madhu_ak | sorry, it should be neutron-lbaas | 21:51 |
| blogan | well it is a lot of code, and multiple stakeholders involved, it'll get complex | 21:52 |
| madhu_ak | well, should learn more now ( given that there is octavia) | 21:52 |
| madhu_ak | yep | 21:52 |
| divya | blogan: madhu_ak bana_k Thanks .. i am not using neutron l3 agent for routing .. using vendor specific OVS that does routing | 21:54 |
| divya | blogan: tried tcpdump .. on VIP seem to receive requests and on FIP it says 0 packets received | 21:55 |
| madhu_ak | divya: if you could let us know the steps how you created, so it might be useful to reproduce locally here | 21:58 |
| blogan | divya: ah that vendor specific thing mihgt be the issue | 21:59 |
| divya | blogan: madhu_ak Thanks for the help .. will debug on this more ... | 22:06 |
| blogan | divya: np, good luck | 22:12 |
| divya | blogan: bana_k Thanks for the help ... security group rule was the problem ... after adding the rule secgroup-add-rule default tcp 80 80 0.0.0.0/0 wget on FIP is working .. Thanks so much | 22:45 |
| *** superflyy has quit IRC | 22:52 | |
| bana_k | cool ! | 22:53 |
| madhu_ak | blogan, fnaval, ajmiller: need your reviews for v2 scenario tests: https://review.openstack.org/#/c/207196/ Thanks | 23:09 |
| fnaval | k, will take a look when i get a chance madhu_ak | 23:09 |
| madhu_ak | thanks | 23:10 |
| ajmiller | I will look too. | 23:11 |
| madhu_ak | thanks | 23:13 |
| *** madhu_ak has quit IRC | 23:18 | |
| *** bharath has quit IRC | 23:33 | |
| *** sballe has quit IRC | 23:36 | |
| *** mlavalle has quit IRC | 23:38 | |
| ajmiller | exit | 23:41 |
| ajmiller | LOL, I guess that shows where my mind is on this fine Friday afternoon.... Later everyone.. | 23:41 |
| *** ajmiller has quit IRC | 23:42 | |
| *** chadix has quit IRC | 23:53 | |
| *** mixos has joined #openstack-lbaas | 23:53 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!