Thursday, 2018-03-15

« Wednesday, 2018-03-14 Index Friday, 2018-03-16 »
*** yamamoto has joined #openstack-lbaas00:07
openstackgerritOpenStack Proposal Bot proposed openstack/neutron-lbaas master: Updated from global requirements  https://review.openstack.org/55277200:10
openstackgerritOpenStack Proposal Bot proposed openstack/octavia-dashboard master: Updated from global requirements  https://review.openstack.org/55154000:13
openstackgerritMichael Johnson proposed openstack/octavia master: Log health manager exceptions  https://review.openstack.org/55311000:34
*** fnaval has joined #openstack-lbaas00:49
*** jniesz has quit IRC00:54
imacdonnhttps://review.openstack.org/#/c/548729/ passed recheck .... \o/01:08
*** harlowja has quit IRC01:59
openstackgerritZhaoBo proposed openstack/octavia master: UDP jinja template  https://review.openstack.org/52542001:59
openstackgerritZhaoBo proposed openstack/octavia master: UDP for [2]  https://review.openstack.org/52965101:59
openstackgerritZhaoBo proposed openstack/octavia master: UDP for [3][5][6]  https://review.openstack.org/53939102:00
*** jaff_cheng has joined #openstack-lbaas02:26
openstackgerritMerged openstack/octavia master: Updated from global requirements  https://review.openstack.org/55219302:27
*** jaff_cheng has quit IRC02:28
openstackgerritHengqing Hu proposed openstack/octavia-dashboard master: Use pool name as hint for selecting pool id  https://review.openstack.org/55312402:53
openstackgerritHengqing Hu proposed openstack/octavia-dashboard master: Replace noop tests with registration test  https://review.openstack.org/55072102:59
johnsomcgoncalves You are hopefully sleeping, but I have a few more "issues" for you....03:12
johnsomhttps://www.irccloud.com/pastebin/gdVhwAQW/03:12
johnsomFound by the new gate I am working on03:12
openstackgerritHengqing Hu proposed openstack/octavia-dashboard master: List children pools on LB details page  https://review.openstack.org/55130503:13
openstackgerritHengqing Hu proposed openstack/octavia-dashboard master: Being able to edit default pool of listener  https://review.openstack.org/55143603:14
openstackgerritHengqing Hu proposed openstack/octavia-dashboard master: Add l7 support  https://review.openstack.org/55194703:14
openstackgerritHengqing Hu proposed openstack/octavia-dashboard master: Add rbac support for octavia service apis  https://review.openstack.org/55031903:14
openstackgerritHengqing Hu proposed openstack/octavia-dashboard master: Use pool name as hint for selecting pool id  https://review.openstack.org/55312403:15
openstackgerritHengqing Hu proposed openstack/octavia-dashboard master: Being able to change insert headers of listener  https://review.openstack.org/54999903:25
*** annp has joined #openstack-lbaas03:26
openstackgerritHengqing Hu proposed openstack/octavia master: Log output of subprocess check_output() errors  https://review.openstack.org/54509003:40
*** links has joined #openstack-lbaas03:41
openstackgerritMichael Johnson proposed openstack/octavia master: Add a gate using lower-constraints  https://review.openstack.org/55313403:41
openstackgerritMichael Johnson proposed openstack/octavia master: Add a gate using lower-constraints  https://review.openstack.org/55313403:45
kongjohnsom: hi, any chance you are around?03:47
johnsomkong Hi, yeah, around sadly.03:48
kong:-)03:48
kongjohnsom: i am wondering if the octavia l7 functionality is totally back-compatible with that in neutron-lbaas?03:48
kongs/back/backward03:48
johnsomYes, should be a superset, meaning there may be more functionality in Octavia, but fully backward compatible with neutron-lbaas.03:49
johnsomThat said, pretty sure the devs put more effort into the octavia L7 than neutron-lbaas.03:50
johnsomThey were fans...03:50
johnsomIs there something not lining up?03:50
kongjohnsom: cool, that's also what i understand03:51
kongbecause i'm working with gophercloud, but i found the l7 functionality is not supported03:51
kongbefore i put effort into that, i'd be good to hear some history story03:52
johnsomAh, yeah. Update this page when you have gophercloud up and running: https://docs.openstack.org/octavia/latest/user/sdks.html03:53
johnsomTrying to track the sdk support for users03:53
kongjohnsom: sure thing03:54
johnsomyeah, the devs for L7 did it first in octavia then added neutron-lbaas. They were primarily octavia users, but new we needed to support neutron-lbaas too03:54
johnsoms/new/knew/g03:54
kongjohnsom: i'm surprised that gophercloud is not in that page03:54
kongit's widely used for k8s03:55
kongfor lb type service implementation03:55
johnsomI recently created it and did my best. If gophercloud has octavia native support, please feel free to add it03:55
kongahh, ok03:55
kongyeah, i can confirm octavia is supported03:55
johnsomExcellent!03:56
johnsomkong I don't see load-balancer/v2 there03:56
johnsomhttps://github.com/gophercloud/gophercloud/tree/master/openstack03:57
kongwait03:58
kongjohnsom: https://github.com/gophercloud/gophercloud/blob/master/openstack/client.go#L39204:01
kongwe deployed octavia and magnum in our preproduction, that lb type service just works well04:02
johnsomOk, yeah, it should just be switching the endpoint from neutron to octavia/load-balancer04:02
johnsomI see, so I was looking in the wrong place04:03
openstackgerritMichael Johnson proposed openstack/octavia master: Add a gate using lower-constraints  https://review.openstack.org/55313404:09
*** harlowja has joined #openstack-lbaas04:11
*** Swami has quit IRC04:14
*** gans has joined #openstack-lbaas04:23
*** harlowja has quit IRC04:50
johnsomOk, created patches for G-R to fix the rest of the minimum version issues found by the gate.  Calling it a night.05:26
*** gans has quit IRC06:01
openstackgerritSanthosh Fernandes proposed openstack/octavia master: Adding exabgp-speaker element to amphora image  https://review.openstack.org/49016406:23
rm_worknice, yeah +1'd those06:53
*** kobis has joined #openstack-lbaas07:11
*** kobis has quit IRC07:12
openstackgerritOpenStack Proposal Bot proposed openstack/neutron-lbaas master: Updated from global requirements  https://review.openstack.org/55277207:23
openstackgerritOpenStack Proposal Bot proposed openstack/neutron-lbaas-dashboard master: Updated from global requirements  https://review.openstack.org/55153407:23
openstackgerritOpenStack Proposal Bot proposed openstack/octavia master: Updated from global requirements  https://review.openstack.org/55321307:27
openstackgerritOpenStack Proposal Bot proposed openstack/octavia-dashboard master: Updated from global requirements  https://review.openstack.org/55154007:27
openstackgerritOpenStack Proposal Bot proposed openstack/octavia-tempest-plugin master: Updated from global requirements  https://review.openstack.org/55154107:28
*** pcaruana has joined #openstack-lbaas07:48
*** pcaruana has quit IRC07:50
*** pcaruana has joined #openstack-lbaas07:50
*** pcaruana is now known as pcaruana|wfh|07:53
*** kobis has joined #openstack-lbaas08:03
*** AlexeyAbashkin has joined #openstack-lbaas08:13
*** KeithMnemonic has joined #openstack-lbaas08:22
*** tesseract has joined #openstack-lbaas08:24
*** rcernin has quit IRC09:00
*** ispp is now known as burned09:04
cgoncalvesjohnsom: new min requirement versions for master? if so, no problem. if for queens, uh-oh :/09:14
cgoncalvesnooo! cryptography>=2.1 also for queens? :(((09:44
*** salmankhan has joined #openstack-lbaas10:17
*** salmankhan has quit IRC10:20
*** salmankhan has joined #openstack-lbaas10:21
cgoncalvesjohnsom: werkzeug is missing in our requirements.txt :O10:29
*** devfaz has quit IRC10:34
*** devfaz has joined #openstack-lbaas10:35
*** yamamoto has quit IRC11:04
openstackgerritAntoni Segura Puimedon proposed openstack/neutron-lbaas master: Support OVN driver for lbaas v2  https://review.openstack.org/51092111:06
*** ianychoi_ has joined #openstack-lbaas11:19
*** fnaval_ has joined #openstack-lbaas11:20
*** mnaser_ has joined #openstack-lbaas11:20
*** Alexey_Abashkin has joined #openstack-lbaas11:20
*** dosaboy_ has joined #openstack-lbaas11:22
andreykurilinHi folks! Do we have any psychic here? Can someone explain me the possible reason of such error - http://logs.openstack.org/42/551942/15/check/rally-task-neutron-with-extensions/5ddd727/results/report.html.gz#/NeutronLoadbalancerV2.create_and_list_loadbalancers/failures ?11:24
andreykurilinthere are no more logs due to issue of infra...11:25
*** pcaruana|wfh| has quit IRC11:27
*** fnaval has quit IRC11:27
*** dosaboy has quit IRC11:27
*** mnaser has quit IRC11:27
*** mnaser_ is now known as mnaser11:27
*** pcaruana|wfh| has joined #openstack-lbaas11:28
*** AlexeyAbashkin has quit IRC11:28
*** ianychoi has quit IRC11:28
*** mnaser has quit IRC11:28
*** mnaser has joined #openstack-lbaas11:28
*** Alexey_Abashkin is now known as AlexeyAbashkin11:28
*** salmankhan has quit IRC11:45
*** salmankhan has joined #openstack-lbaas11:46
*** annp has quit IRC11:48
*** dmellado has quit IRC11:50
*** dmellado has joined #openstack-lbaas11:58
*** yamamoto has joined #openstack-lbaas12:04
*** dmellado has quit IRC12:05
*** yamamoto has quit IRC12:09
*** pcaruana|wfh| is now known as pcaruana12:14
*** salmankhan has quit IRC12:15
*** yamamoto has joined #openstack-lbaas12:16
*** dmellado has joined #openstack-lbaas12:23
*** salmankhan has joined #openstack-lbaas12:26
*** dmellado has quit IRC12:30
rm_workcgoncalves: i mean yeah basically that's what we need for queens :/12:48
rm_worksince we're basically still on the queens SHA12:48
rm_workbesides gatefixes (all 10 of them?) lol12:48
cgoncalvesrm_work: manually created lower-constraints.txt out of requirements.txt from stable/queens. I can confirm that we also need newer barbicanclient et al.12:49
*** dayou has quit IRC12:50
rm_work:/12:52
openstackgerritHengqing Hu proposed openstack/octavia-dashboard master: Allow cascade deletion of load balancer  https://review.openstack.org/55338112:52
cgoncalvesrm_work: werkzeus is also a dependency but is not set in requirements.txt12:53
cgoncalveswe never faced any issue because it gets installed as a dependency of something we depend on12:53
*** salmankhan has quit IRC12:55
cgoncalvesrm_work, johnsom: http://paste.openstack.org/show/701708/12:58
rm_workhmmmmmm12:58
rm_workit is amphora agent only IIRC12:59
rm_workbut yeah i think we use the same reqs file still12:59
rm_workbleh :/12:59
rm_workI thought it was in there12:59
rm_workwhelp12:59
rm_workcgoncalves: review all my patches plox :P13:00
rm_workwe really use werkzeug directly? :/13:01
rm_workaugh, just exceptions13:01
cgoncalvesyet, we need to add to requirements.txt13:02
rm_worki wonder if there's a way around that13:02
rm_workhold on13:02
rm_worki wonder13:02
rm_workaugh nah not going to be easy to replace13:02
rm_workalright T_T13:02
cgoncalvesrm_work: let me first calm down the storm in RDO and OSP :)13:02
*** dayou has joined #openstack-lbaas13:02
*** atoth has joined #openstack-lbaas13:06
*** fnaval_ has quit IRC13:06
*** salmankhan has joined #openstack-lbaas13:12
*** yamamoto has quit IRC13:18
*** yamamoto has joined #openstack-lbaas13:19
*** dmellado has joined #openstack-lbaas13:19
*** salmankhan has quit IRC13:19
*** dmellado has quit IRC13:20
*** ianychoi_ is now known as ianychoi13:23
*** yamamoto has quit IRC13:23
*** links has quit IRC13:42
*** salmankhan has joined #openstack-lbaas13:43
*** salmankhan has quit IRC14:02
-openstackstatus- NOTICE: POST_FAILURE results on Tempest-based jobs since the most recent Zuul security fixes are being investigated; rechecking those won't help for now but we'll keep you posted once a solution is identified14:09
*** ChanServ changes topic to "POST_FAILURE results on Tempest-based jobs since the most recent Zuul security fixes are being investigated; rechecking those won't help for now but we'll keep you posted once a solution is identified"14:09
rm_workwut14:14
rm_workthat's our topic now O_o14:14
cgoncalvestemporarily, yes14:14
*** yamamoto has joined #openstack-lbaas14:20
*** yamamoto has quit IRC14:25
*** salmankhan has joined #openstack-lbaas14:26
johnsomcgoncalves: http://lists.openstack.org/pipermail/openstack-dev/2018-March/128352.html14:30
rm_workohai johnsom14:31
rm_workgood morning14:31
johnsomYeah, haven’t made my coffee yet.14:32
cgoncalvesjohnsom: yeah, I've read it (tl;dr part) :)14:33
*** salmankhan has quit IRC14:33
johnsomDid you catch the part where he ALSO says you should have used upper-constraint versions?14:35
rm_workugh we need to backport the neutron job fix i think14:37
rm_workassuming that's why https://review.openstack.org/#/c/553399/ is broken14:38
cgoncalvesI'd love to lock you in a room with packagers/release delivery people. last man standing gets the key out14:38
rm_workdid it to see what happens14:38
rm_workcgoncalves: that sounds like *fun*14:39
* rm_work rubs hands together mischievously14:40
cgoncalvesrm_work: I could also throw you in ;)14:40
rm_worki have some choice words for some packagers / release teams, especially the ones from RHEL/CentOS14:41
johnsomOffer accepted, only if we get permanent invites to the RH parties14:41
rm_workI know way too many redhat people to have not been able to get to that party14:41
cgoncalveshahahaha! sure, assuming you're the one walking out ;)14:42
rm_worki'm going to claim the only reason i didn't go is because I didn't feel like it :P14:42
*** salmankhan has joined #openstack-lbaas14:45
cgoncalves:))14:45
cgoncalvesrm_work: refering to backporting https://review.openstack.org/#/c/552978/ ?14:45
*** dmellado_ has joined #openstack-lbaas14:46
rm_workyes14:46
rm_workit's here:14:46
rm_workhttps://review.openstack.org/55341414:46
rm_workgonna see if it "just works"14:46
cgoncalvescool, thanks!14:46
rm_worktoday i am thinking of trying to tackle cascade-delete on the remaining object types14:46
rm_workand if i have time, look at single-create from the remaining object types as well14:47
rm_workI think it should "just work"14:47
rm_workor be hopefully very minimal effort14:47
cgoncalvesrm_work: I recall we talked about possibility of https://review.openstack.org/#/c/548989/ being backported but I don't remember the outcome14:51
cgoncalvesthe patch requires timestamps which were introduced after queens branch14:52
rm_workyeah...14:52
rm_worki'm trying to remember exactly all of the things i did in this patch and why14:52
rm_workit was very ... complex14:52
rm_workthe interworkings of that stuff makes me a little loopy14:53
*** dmellado_ is now known as dmellado14:54
cgoncalvesrm_work: limiting to reading your commit message, seems like you're addressing two bugs in a single patch14:58
rm_workmy shower thought of the day: https://i.imgflip.com/26g0mf.jpg14:58
rm_workprobably not but who knows :P14:59
rm_workcgoncalves: hmm yeah, that might be true14:59
cgoncalves-1'ing from now on14:59
rm_worklolol14:59
*** beagles is now known as beagles_food15:02
rm_workso... they're kinda related15:06
rm_worki think one kinda falls out of the other15:06
*** beagles_food is now known as beagles15:09
*** yamamoto has joined #openstack-lbaas15:21
*** yamamoto has quit IRC15:27
rm_workummm15:34
rm_workwhy does our `get_delete_load_balancer_flow` include listener delete? O_o15:34
rm_workseems the main difference between that and the cascade flow is the inclusion of pools?!15:34
rm_workwhy would it be doing `delete_LB_flow.add(listeners_delete)`15:35
rm_workthis is handled really strangely anyway15:36
openstackgerritMichael Johnson proposed openstack/octavia master: Log health manager exceptions  https://review.openstack.org/55311015:37
*** kobis has quit IRC15:54
*** kobis has joined #openstack-lbaas15:55
*** kobis has quit IRC15:55
xgerman_em_work did you have to put indexes on your dbs? My API servers are awfully slow…15:57
xgerman_rm_work:15:57
rm_workumm15:58
rm_workdon't we define some15:58
rm_workI thought we did15:58
rm_workI haven't done any custom ones15:58
rm_workah I did have a ton of API slowness but it was due to DNS issues looking up keystone for auth constantly15:59
johnsomWe have indexes on the key columns by default15:59
rm_workand also some interesting opts in the wsgi runner15:59
rm_workwhich are you using?15:59
rm_worki ended up setting some custom stuff in my uwsgi config:16:00
rm_workhttp://paste.openstack.org/show/701888/16:00
rm_workoh ... lol16:00
rm_workuhh hey so, my uwsgi does hot-reload of the API code on config change :P16:01
rm_worki forgot about that16:01
*** harlowja has joined #openstack-lbaas16:03
*** KeithMnemonic has quit IRC16:15
*** yamamoto has joined #openstack-lbaas16:23
*** yamamoto has quit IRC16:29
*** AlexeyAbashkin has quit IRC16:32
xgerman_sorry, was distracted with anither meeting16:42
*** harlowja has quit IRC17:09
rm_workugh so long devstack17:11
rm_worki wish i had access to a cloud with ubuntu hosts that had vxt17:12
*** kobis has joined #openstack-lbaas17:17
xgerman_hot take: uwsgitop is a nice tool17:22
*** yamamoto has joined #openstack-lbaas17:25
*** SumitNaiksatam has joined #openstack-lbaas17:29
*** yamamoto has quit IRC17:30
*** kobis has quit IRC17:40
*** imacdonn has quit IRC17:44
*** imacdonn has joined #openstack-lbaas17:44
cgoncalvesjohnsom: https://review.openstack.org/#/c/553447/1/deliverables/pike/neutron-lbaas.yaml@4017:59
johnsomcgoncalves Yeah, they should not be releasing that18:00
*** fnaval has joined #openstack-lbaas18:13
*** fnaval_ has joined #openstack-lbaas18:14
*** fnaval has quit IRC18:14
*** fnaval_ has quit IRC18:18
*** fnaval has joined #openstack-lbaas18:18
*** harlowja has joined #openstack-lbaas18:20
*** yamamoto has joined #openstack-lbaas18:27
*** yamamoto has quit IRC18:33
*** SumitNaiksatam has quit IRC18:38
*** tesseract has quit IRC18:39
rm_workok so, object types that can possibly have children:18:45
rm_workPools: have members; are referred to by l7policies18:45
rm_workListeners: have l7policies18:46
rm_workL7Policies: have L7Rules18:46
rm_workLoadbalancers: have listeners, have pools18:46
rm_workis that it?18:46
rm_workanyway....18:48
rm_workLoadbalancer: has a cascade delete option, if false it won't delete on any children present18:48
johnsompools have health monitors18:48
rm_workListeners: has no cascade option, automatically takes L7Policies with it18:48
rm_workPools: has no cascade option, won't delete if L7Policies refer to it18:49
rm_work^ but will delete members18:49
rm_workand I think also HMs18:49
johnsomyes18:49
rm_workyep18:50
rm_work?!?!18:50
rm_workso why18:50
rm_workwhy are L7 rules somehow special for pools18:50
rm_worknothing else even requires a cascade18:50
rm_workjust LB18:51
rm_workyeah, L7Policy: has no cascade option, deletes rules automatically18:52
rm_workso I think the real answer isn't to add cascade options to stuff18:52
rm_workit's to make pools auto-cascade the l7policies18:52
rm_workbecause that matches everything else18:52
rm_workthoughts?18:53
*** AlexeyAbashkin has joined #openstack-lbaas18:54
*** beagles is now known as beagles|biab18:57
*** AlexeyAbashkin has quit IRC18:58
*** salmankhan has quit IRC19:00
*** kobis has joined #openstack-lbaas19:18
*** ChanServ changes topic to "Discussion of OpenStack Load Balancing (Octavia) | Rocky is open for development!"19:24
-openstackstatus- NOTICE: The regression stemming from one of yesterday's Zuul security fixes has been rectified, and Devstack/Tempest jobs which encountered POST_FAILURE results over the past 24 hours can safely be rechecked now19:25
*** yamamoto has joined #openstack-lbaas19:29
*** yamamoto has quit IRC19:33
*** beagles|biab is now known as beagles19:42
xgerman_rm_work no20:13
rm_workso you'd rather still pass a cascade param20:13
xgerman_no, L7 and pools are sort of independent20:14
rm_workeven though *every other object with children, including pool itself for all other types* just does a cascade automatically20:14
xgerman_you might want to delete all pools but keep L7 and vice versa20:14
rm_workerrr20:14
rm_workso you can't really keep an L7 that refers to a pool20:14
rm_workit gets orphaned and it's an L7 policy referring to a pool-id that doesn't exist20:14
rm_workwhich generates a broken haproxy config20:14
xgerman_yes, made a mistake20:14
rm_workL7s are actually *on* Listeners20:15
xgerman_but deleteing L7 and keeping pools is valid20:15
rm_workright20:15
rm_workthose wouldn't cascade20:15
xgerman_yeah, I was referring to:20:15
xgerman_still fell funny to cascade L7 with pools20:16
xgerman_but I am ok with listener20:16
rm_workerr20:16
rm_workSo20:16
rm_workif you have a Listener, and a Pool (separate)20:17
rm_workand you create an L7Policy on the Listener, pointing to the pool20:17
rm_workright now, if you delete the Listener, the L7Policy would go away20:17
rm_workbut if you try to delete the pool, it'll complain20:17
rm_worki'd like to change it so if you delete the pool, it deletes that L7Policy20:17
rm_workso that it matches20:18
xgerman_ok, but you could update L7 poicy to point to a different pool20:18
rm_workright now it's weirdly inconsistent20:18
xgerman_yes20:18
rm_workhmmm20:18
rm_workso you'd rather just20:18
rm_workun-point the policy20:18
rm_workand maybe disable it?20:18
rm_workI can do that too20:19
xgerman_I think so - but I am no UX expert20:19
rm_workwould like to hear from johnsom20:19
rm_workand maybe sbalukoff :P20:19
xgerman_ha20:19
rm_worki think your proposal does make sense though20:19
rm_workyou might have a lot of complex rules20:20
xgerman_let’s see what cgoncalves thinks, too20:20
rm_workand really just want to re-point20:20
xgerman_yep20:20
rm_workand it'd suck to delete a pool and just WHOOPS there it all went20:20
rm_workok i'll start re-tooling for that while i wait20:20
xgerman_yeah, I think if you want to blow stuff away you do lb, listener and then it gets very selective20:20
rm_workyeah20:21
rm_workthings that blow away children:20:21
rm_workLB (with cascade option)20:21
rm_workListener (automatically as of now will delete L7Policies on it)20:22
rm_workPool (automatically as of now will delete all members and HMs, adding auto-unsetting L7Policies)20:22
rm_workL7Policy (will delete all rules automatically)20:22
xgerman_can a gelath monitor be used by more than one pool?20:23
rm_workno20:23
rm_workwe never did that20:23
xgerman_ok20:23
rm_worki think we decided to never allow it too20:23
xgerman_k20:24
*** kobis has quit IRC20:26
*** yamamoto has joined #openstack-lbaas20:29
*** yamamoto has quit IRC20:35
rm_workdid johnsom and cgoncalves both disappear?21:07
johnsomIn a meeting21:08
rm_workah k21:08
xgerman_is that meeting still going on?21:11
rm_workxgerman_: easiest merge: https://review.openstack.org/#/c/548991/21:11
rm_workand i gave you a +2 on https://review.openstack.org/55311021:12
rm_workxgerman_: ah this would be good too: https://review.openstack.org/55264121:13
xgerman_done21:15
johnsomxgerman_ different meeting21:17
xgerman_k21:17
johnsomrm_work Sorry, between lunch and an internal meeting I have been ignoring the chatter.  What is up?21:20
*** mburrows has quit IRC21:21
rm_workdeciding how to deal with pools and cascading l7policies21:21
rm_workdecision i think is to "unset" the pool from relevant l7policies, and just do it automatically (no new arg)21:22
rm_workabout to push up a change21:22
johnsomSo, the pool would just still be owned by the load balancer. That makes sense to me21:25
rm_workerrr21:29
openstackgerritAdam Harwell proposed openstack/octavia master: Make Pools with L7Policies deletable  https://review.openstack.org/55361521:29
rm_workother way21:29
rm_workyou can see it there ^^, but ... when deleting a pool, disassociate the l7policies21:29
rm_workinstead of just blocking the delete21:29
johnsomUgh, that seems like a bad idea21:30
rm_workI think I agree with xgerman_ that it is better than deleting them21:30
rm_worksince they aren't actually children of the pool21:30
rm_workthey're on the Listener... they just refer to a pool by id21:30
johnsomThat would change the behavior of the L7policy21:31
rm_workit basically just snips the connection21:31
rm_workerr21:31
rm_workso functionally, it'd behave as if it were deleted21:31
rm_workbecause it's being disassociated and disabled21:31
*** yamamoto has joined #openstack-lbaas21:31
rm_workthe user can go back and re-enable it and set a new pool-id, without losing all the rules that were on it21:32
rm_workwhile it's disabled, it's excluded from rendering to haproxy21:32
rm_workso it's essentially the same as if we deleted it21:32
johnsomI mean, you are basically deleting the l7policy when deleting the pool, so traffic would switch to flowing down the listener default pool, if there is one.21:33
rm_workyes21:33
rm_workmy point was that this is identical to deleting it21:33
rm_workfunctionally for the LB21:33
johnsomThis feels like an un-intended consequence of an action21:33
rm_workerr well21:34
rm_workwhat would you like it to do21:34
rm_workthis is the only object left we can't fully "cascade delete"21:34
rm_workwould you rather switch the L7Policy to REJECT? :P21:34
rm_workseems worse to me21:34
rm_workbut21:34
*** cgoncalves has quit IRC21:34
rm_workmaybe functionally more sane21:34
rm_worksince if you delete the pool, you're essentially black-holing the traffic that would have gone there21:35
johnsomI think a delete on a pool in use by an L7 policy should error and say "I'M in use!", if they use the cascade flag, then yeah, delete the pool, l7policy, and L7rule21:35
rm_workerg you really want to fully delete the policy and rules? :/21:35
rm_workthey aren't *on* the pool though21:35
rm_workthey're on the listener21:35
*** cgoncalves has joined #openstack-lbaas21:36
xgerman_yep, might get confusing that way21:36
rm_worki could be convinced to update the policy to REJECT and leave it enabled21:36
*** yamamoto has quit IRC21:36
rm_workthen it wouldn't shuffle any traffic flows21:36
johnsomYeah, but as you noted, they become useless and behave essentially as if they were deleted. To me "cascade" is a big hammer.21:36
rm_workwell, note this is also implicit21:36
rm_workthere's no new flag21:37
rm_workbecause *no other object* requires one21:37
rm_workeverything we have will cascade automatically21:37
rm_workthis is the only exception at the moment21:37
johnsomI think a delete on a pool in use by an L7 policy should error and say "I'M in use!"21:37
johnsomNo, listeners don't21:37
rm_workyes they do21:37
rm_workI just tested it :/21:37
johnsomuhhhh21:37
rm_workthey'll wipe out L7Policies under them21:37
rm_workno cascade arg21:37
rm_workjohnsom: http://paste.openstack.org/show/702250/21:39
rm_workah and a policy show in case you are curious:21:39
rm_workUnable to locate f8d6197e-1197-46dc-bcdd-27b717e20cfa in l7policies21:39
rm_workit's *gone*21:40
johnsomI am noodling on if that is what we intended or if that is just how it ended up21:40
xgerman_I can see a use case for cascade LB and listener… pools clearly should delete members; l7policies the rules but other than that…21:42
xgerman_who is going to use/care21:42
rm_workmaybe no one? but21:49
xgerman_yeah, dunno21:49
rm_worktrying to get things more ... consistent21:49
rm_workand i had "get cascade delete and single-create on all objects"21:49
xgerman_I think whatever we decide is better than what we had before ;-)21:49
rm_workso this was one of my stops :P21:49
xgerman_yep21:49
rm_worksingle-create entrypoints on all objects is my next stop21:50
xgerman_k21:50
rm_workbut, i'm about to call it for the day21:50
xgerman_kk21:53
rm_worki've got a number of patches waiting on reviews still -- the barbican API fix one I think would be the clearest one to get in soon, as it fixes an actual problem21:55
rm_workhttps://review.openstack.org/55030321:56
rm_worki fixed that in response to an issue imacdonn ran into21:56
rm_workanywho, bbl21:56
johnsomo/21:56
*** rcernin has joined #openstack-lbaas22:02
openstackgerritGerman Eichberger proposed openstack/octavia master: Periodic job to build + publish diskimage  https://review.openstack.org/54925922:19
*** yamamoto has joined #openstack-lbaas22:32
*** yamamoto has quit IRC22:38
johnsomOk, one more patch and G-R will be up to date with our minimum requirements22:56
johnsom(to merge)22:56
openstackgerritMerged openstack/octavia master: Reduce runtime for l7rule too_many_rules test  https://review.openstack.org/54899122:58
openstackgerritMerged openstack/octavia master: Log health manager exceptions  https://review.openstack.org/55311022:58
openstackgerritMerged openstack/octavia master: Devstack plugin image build option quoting fix  https://review.openstack.org/55264122:58
openstackgerritGerman Eichberger proposed openstack/octavia master: Periodic job to build + publish diskimage  https://review.openstack.org/54925923:17
*** sticker has joined #openstack-lbaas23:17
*** yamamoto has joined #openstack-lbaas23:27
*** rcernin has quit IRC23:39
« Wednesday, 2018-03-14 Index Friday, 2018-03-16 »

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!