Thursday, 2017-04-06

« Wednesday, 2017-04-05 Index Friday, 2017-04-07 »
*** armax has quit IRC00:21
*** gongysh has joined #openstack-lbaas00:31
*** ducttape_ has joined #openstack-lbaas00:34
*** harlowja has quit IRC00:39
*** ducttape_ has quit IRC00:46
eanderssonI managed to get the list loadbalancer down to 2s :D00:54
*** ducttape_ has joined #openstack-lbaas00:55
*** ducttape_ has quit IRC01:00
*** ducttape_ has joined #openstack-lbaas01:09
*** ducttape_ has quit IRC01:25
*** harlowja has joined #openstack-lbaas01:26
*** harlowja has quit IRC01:26
eanderssonxgerman, johnsom so the issue is that when performing something as simple as the lbaas-loadbalancer-list it gets everything01:31
eanderssonIf you reduce the data parsed to only the relevant data the list operation takes 1-2s01:31
eanderssonWith all data (including unused) it takes 20-30s01:32
*** ducttape_ has joined #openstack-lbaas01:34
*** ducttape_ has quit IRC01:37
*** ducttape_ has joined #openstack-lbaas02:02
*** ducttape_ has quit IRC02:06
*** ducttape_ has joined #openstack-lbaas02:08
eanderssonUnwanted data would include things like a list of pools and listeners02:17
eanderssonWhile wanted data would include https://github.com/openstack/python-neutronclient/blob/33b8a0cf0bfc66dc68d84dd0592480e5e1ae3288/neutronclient/neutron/v2_0/lb/v2/loadbalancer.py#L4202:18
*** reedip has quit IRC02:25
*** ducttape_ has quit IRC02:25
*** aojea has joined #openstack-lbaas02:54
*** fnaval has joined #openstack-lbaas02:58
*** aojea has quit IRC02:59
openstackgerritZhaoBo proposed openstack/octavia master: Support QoS policy apply on vip port  https://review.openstack.org/44191203:09
*** reedip has joined #openstack-lbaas03:09
*** cody-somerville has joined #openstack-lbaas03:10
*** ducttape_ has joined #openstack-lbaas03:11
*** csomerville has quit IRC03:11
*** cody-somerville has quit IRC03:25
*** cody-somerville has joined #openstack-lbaas03:25
*** cody-somerville has quit IRC03:25
*** cody-somerville has joined #openstack-lbaas03:25
*** ducttape_ has quit IRC03:30
*** reedip has quit IRC03:44
*** reedip has joined #openstack-lbaas03:45
*** links has joined #openstack-lbaas03:45
*** bbzhao has quit IRC03:47
*** bbzhao has joined #openstack-lbaas03:48
*** fnaval_ has joined #openstack-lbaas04:02
*** fnaval has quit IRC04:10
*** gongysh has quit IRC04:17
*** ducttape_ has joined #openstack-lbaas04:28
*** ducttape_ has quit IRC04:33
*** amotoki has joined #openstack-lbaas04:37
openstackgerritShashank Kumar Shankar proposed openstack/octavia master: Run Octavia API in a WSGI server  https://review.openstack.org/44093405:10
*** ducttape_ has joined #openstack-lbaas05:58
*** ducttape_ has quit IRC06:02
*** reedip has quit IRC06:17
*** aojea has joined #openstack-lbaas06:30
*** gcheresh_ has joined #openstack-lbaas06:36
*** voelzmo has joined #openstack-lbaas06:41
*** voelzmo has quit IRC06:47
*** voelzmo has joined #openstack-lbaas06:47
*** pcaruana has joined #openstack-lbaas07:02
*** tesseract has joined #openstack-lbaas07:04
*** rcernin has joined #openstack-lbaas07:09
*** rcernin has quit IRC07:10
*** rcernin has joined #openstack-lbaas07:10
*** reedip has joined #openstack-lbaas07:10
*** kobis has joined #openstack-lbaas07:20
*** aojea has quit IRC07:29
*** yamamoto has quit IRC07:33
*** yamamoto has joined #openstack-lbaas07:34
*** gongysh has joined #openstack-lbaas07:56
*** fnaval_ has quit IRC08:00
*** aojea has joined #openstack-lbaas08:46
*** aojea_ has joined #openstack-lbaas08:47
*** yamamoto has quit IRC08:48
*** yamamoto has joined #openstack-lbaas08:48
*** aojea has quit IRC08:50
*** reedip has quit IRC09:22
*** foutatoro has joined #openstack-lbaas09:24
*** gongysh has quit IRC09:52
*** yamamoto has quit IRC09:57
*** yamamoto has joined #openstack-lbaas09:57
*** fnaval has joined #openstack-lbaas10:01
*** fnaval has quit IRC10:06
*** ndahiwade has joined #openstack-lbaas10:13
*** yamamoto has quit IRC10:13
*** yamamoto has joined #openstack-lbaas10:15
*** links has quit IRC10:39
*** links has joined #openstack-lbaas10:54
*** reedip has joined #openstack-lbaas10:58
*** voelzmo has quit IRC11:24
*** voelzmo has joined #openstack-lbaas11:25
reedipo/11:42
*** ducttape_ has joined #openstack-lbaas12:00
*** ducttape_ has quit IRC12:01
*** ducttape_ has joined #openstack-lbaas12:02
*** fnaval has joined #openstack-lbaas12:02
*** fnaval has quit IRC12:06
openstackgerritNir Magnezi proposed openstack/octavia master: Auto detect haproxy user_group  https://review.openstack.org/42939812:07
*** ducttape_ has quit IRC12:32
xgermano/12:37
*** voelzmo has quit IRC12:38
*** voelzmo has joined #openstack-lbaas12:38
*** ducttape_ has joined #openstack-lbaas12:39
nmagnezixgerman, o/12:39
*** voelzmo has quit IRC12:40
*** voelzmo has joined #openstack-lbaas12:41
nmagnezixgerman, do you know if v2 api pools are ready?12:49
nmagnezixgerman, i see we still import from v1 here: https://github.com/openstack/octavia/blob/master/octavia/api/v2/types/listener.py#L1912:50
xgermanyes, it’s ready and merged12:50
nmagnezixgerman, so I'll send a patch to change that12:50
xgermanmmh? I tested it and it worked…12:50
xgermanmaybe they are the same thpes?12:50
xgermank, sounds good12:50
nmagnezixgerman, it caught my eye when i looked at https://review.openstack.org/#/c/453721/12:51
nmagnezixgerman, but since i was not involved, i imagined is is worth asking12:51
xgermangood eye!12:51
nmagnezi:D12:52
nmagnezixgerman, same case for v2 l7policy?12:55
nmagneziis it ready?12:56
nmagnezihttps://github.com/openstack/octavia/blob/master/octavia/api/v2/types/listener.py#L1812:56
xgermanyes, I think they merged as well12:56
xgermanyep, probably needs updating, too12:56
xgermanbit surprised they are different models since this was a late addition and I thought we made carbon copies between v1 and lbaasv212:57
nmagnezixgerman, i'll update both of them in the same patch :)13:00
xgermansound sgood13:02
openstackgerritBrent Eagles proposed openstack/neutron-lbaas master: Create header with authtoken even if no args  https://review.openstack.org/45417213:05
*** ducttape_ has quit IRC13:05
*** catintheroof has joined #openstack-lbaas13:05
*** gongysh has joined #openstack-lbaas13:08
*** links has quit IRC13:09
openstackgerritNir Magnezi proposed openstack/octavia master: Fix imports in v2 listener types  https://review.openstack.org/45417713:09
nmagnezixgerman, ^13:09
xgermanok13:09
*** bzhao has quit IRC13:11
*** chlong has joined #openstack-lbaas13:31
*** ducttape_ has joined #openstack-lbaas13:36
beaglesxgerman: I've been working on adding support for octavia in tripleo and we've come across an issue with re-creating the  loadbalancer listen port (o-hm0) on reboot. Is there a standard solution or set of recommendations for creating a config that survives reboot?13:37
xgermanwhat are tour rebooting? The amphora?13:38
beaglesxgerman: the node hosting the octavia services (api, health manager, worker controller, etc)13:39
beaglesxgerman: we are creating the neutron port and then manually instniating an o-hm0 port on the br-int bridge - but on reboot all of this disappears (or apppears to)13:40
beaglesxgerman: so we're wondering what others do to make the o-hm0 interface survive reboot13:42
xgermanwell, I haven’t tried rebooting but the bridges I made would come up without manual intervention13:42
xgermanI have bolted veth pairs on them13:43
xgermanthe veth pair connects to neutron as a provide rnet13:43
beaglesxgerman: ah right13:43
xgermanthen my control plane runs in a container which is again connected to the bridge13:43
xgermanall my ips are fix (only neutron dhcp for amphora vms)13:44
xgermanI also recommend to exlude the fix ips from the pool13:45
xgerman(dhcp pool)13:45
beaglesxgerman: right13:45
beaglesxgerman: makes sense.. I'll try something similar for tripleo13:46
xgermancool13:46
beaglesxgerman: thanks!13:46
*** ducttape_ has quit IRC13:48
*** fnaval has joined #openstack-lbaas14:03
*** ducttape_ has joined #openstack-lbaas14:03
*** ducttape_ has quit IRC14:05
*** ducttape_ has joined #openstack-lbaas14:05
*** ducttape_ has quit IRC14:05
*** ducttape_ has joined #openstack-lbaas14:06
*** ducttape_ has quit IRC14:06
*** ducttape_ has joined #openstack-lbaas14:07
*** fnaval has quit IRC14:07
*** ducttape_ has quit IRC14:07
*** amitry_ has joined #openstack-lbaas14:09
*** gongysh has quit IRC14:10
*** amitry has quit IRC14:11
*** amitry_ is now known as amitry14:11
*** sindhu has quit IRC14:11
*** sindhu has joined #openstack-lbaas14:13
openstackgerritReedip proposed openstack/octavia master: Introduce Octavia v2 API for Health Monitor  https://review.openstack.org/40625614:14
reedipxgerman , rm_work, ankur-gupta-f4, sindhu : rebased to the master, review pending ^14:25
xgermanthx14:25
*** ducttape_ has joined #openstack-lbaas14:41
*** aojea_ has quit IRC14:43
*** gcheresh_ has quit IRC14:44
*** armax has joined #openstack-lbaas14:50
*** voelzmo has quit IRC14:51
*** voelzmo has joined #openstack-lbaas14:51
*** voelzmo has quit IRC14:52
*** voelzmo has joined #openstack-lbaas14:53
*** fnaval has joined #openstack-lbaas15:01
*** armax has quit IRC15:04
*** ducttape_ has quit IRC15:06
*** fnaval has quit IRC15:06
*** voelzmo has quit IRC15:07
*** cody-somerville has quit IRC15:08
*** cody-somerville has joined #openstack-lbaas15:09
*** cody-somerville has quit IRC15:09
*** cody-somerville has joined #openstack-lbaas15:09
*** fnaval has joined #openstack-lbaas15:09
*** fnaval has quit IRC15:10
*** fnaval has joined #openstack-lbaas15:11
*** ducttape_ has joined #openstack-lbaas15:12
*** ducttape_ has quit IRC15:13
*** bzhao has joined #openstack-lbaas15:13
*** cody-somerville has quit IRC15:19
*** foutatoro has quit IRC15:22
*** pcaruana has quit IRC15:28
*** kobis has quit IRC15:29
*** voelzmo has joined #openstack-lbaas15:44
rm_workk15:55
rm_worko/ back15:55
johnsomo/15:56
rm_workdisappeared unexpectedly yesterday as well >_<15:56
rm_worknmagnezi / xgerman: yeah since listeners merged first, v1 were left in as placeholders until the v2 versions made it in, and those were supposed to be changed as the next models down the line merged, but look to have been forgotten as it was a separate file :P15:57
ankur-gupta-f4hallo15:58
ankur-gupta-f4good mornings15:58
*** voelzmo has quit IRC16:02
*** voelzmo has joined #openstack-lbaas16:08
*** voelzmo has quit IRC16:16
*** armax has joined #openstack-lbaas16:27
reedipits 0030 am ... good morning to u as well :D16:32
reedipjohnsom : regarding th DNS patch, I confirmed with Nova, they dont have any explicit reason but they have always kept the DNS Name as host-<ip address> so the same can be probably used in Octavi16:34
johnsomCool, I like keeping it simple16:34
reedipagreed !16:35
*** voelzmo has joined #openstack-lbaas16:39
*** voelzmo has quit IRC16:49
*** tesseract has quit IRC16:50
rm_workreedip / sindhu: either of you working on HM patch today?16:55
rm_workit still needs a bit of work16:55
sindhurm_work: nope16:55
rm_workk... i might have time16:56
reediprm_work : if you hve comments, then please let us know, I will , once I wake up in another 5-6 hours16:56
reedipor if u have time, then do it16:56
reedipoops u beat me to it :)16:56
rm_worklol k16:56
rm_workin 5-6 hours i'll probably be off16:57
rm_workso you can take over if it isn't *ready* still by then16:57
reedipsure thingy16:57
reedipgotta go now... dont know whats being typed ... too drowsy now16:57
reedip:)16:57
sindhurm_work: wanted to ask u abt this:  https://review.openstack.org/#/c/453721/16:58
rm_worklol k night reedip16:58
rm_workI commented16:58
rm_workyou mean, ask about my comment?16:58
sindhurm_work: yeah, can u explain a lil more what needs to be looked at. will work on it today16:59
rm_workit is a good catch, but the catch is more than just a bad import16:59
rm_worki think we need to remove that completely16:59
rm_worki mean, either we need to update the controller to also create the members16:59
rm_workor we need to not allow sending members into the POST16:59
rm_workI think the latter is the case, as I don't believe we should be doing *partial* tree creates17:00
rm_workleave that to the "create lb graph" stuff17:00
rm_workjohnsom: ^^ thoughts? does Neutron-lbaas allow this?17:00
sindhurm_work: remove only from the POST?17:01
johnsomrm_work isn't this for single-call-create?17:02
rm_workjohnsom: i thought single-call-create goes through a different mechanism17:11
rm_worknot POST /pools/17:11
rm_workfor example17:11
rm_workI guess it might still use the PoolPOST type?17:12
rm_workbut, we shouldn't be accepting an array of members to create on a POST to /pools/ right?17:12
johnsomhttps://github.com/openstack/neutron-lbaas/blob/master/neutron_lbaas/extensions/lb_graph.py17:12
johnsomhttps://github.com/openstack/neutron-lbaas/blob/master/neutron_lbaas/extensions/lb_graph.py#L6617:12
rm_workLooking briefly at the controller for Pools, I don't think it'll DO anything with the members17:12
rm_workhmm17:13
rm_workwell there's two options: remove members from pool POST, or fix the controller to actually handle them, I think17:13
johnsomYeah, I don't know that we have single-call wired up yet17:13
rm_workso, should we just leave it?17:14
rm_workfix the import but have it continue to not function?17:14
johnsomIt does seem odd to have that there.  Maybe there should be separate types for that.17:14
rm_workyeah17:14
rm_workI think there need to be17:14
rm_workotherwise it will also accept members on a non-single-call17:14
johnsomRight, which could be good, could be bad17:15
*** harlowja has joined #openstack-lbaas17:23
rm_workerm... k17:27
rm_workI mean if we want to allow for that, I'm not against it17:28
rm_workpartial trees would be fine17:28
rm_workand keeping the models like that would make things simpler in a way17:28
rm_workbut we'd need to actually make sure it works17:28
johnsomYep17:28
rm_workI can do a pass once we get HM going I guess...17:28
rm_workideally we don't have to duplicate a ton of logic17:28
rm_workI was actually working in the background on decomposing all the controllers17:28
rm_workto make 90% of the boilerplate go away17:28
rm_workIf i can get my personal life sorted so I'm less distracted, and then actually get all the currently pending stuff through... :/17:29
eanderssonjohnsom, is the api stuff currently in neutron-lbaas being completely re-done for Pike?17:29
johnsomYep17:30
johnsomIt will go deprecated17:30
eanderssonThe stuff I found yesterday seem kinda crazy.17:30
eanderssonSo much uncessary work happening in the background.17:30
rm_workYeah I want to say the octavia version of the API won't do this...17:31
rm_work(very briefly reviewed your comments earlier in scrollback)17:31
rm_workall our models lazy-load17:31
eanderssonMaybe I am just ignorant, but I don't understand why listing lbs would need to pull and build data for listeners and pools as well.17:31
rm_workand I don't believe LB actually displays much17:32
rm_workI am not sure it needs to unless we are doing a full-tree display17:32
rm_workwhich is possible17:32
rm_workbut, maybe it was not properly isolated from regular display17:32
eanderssonIf you want pools and listeners just perform all three calls :D17:32
rm_workeugh17:32
rm_worki mean by default yes17:32
rm_workbut we support graph-create17:32
rm_workand so we need to be able to generate a full tree17:33
rm_workbut17:33
rm_workI don't think a normal GET needs that, no17:33
rm_worki don't know how it was implemented in n-lbaas at all though17:33
rm_workwe're black-boxing it for Octavia17:33
rm_workthe actual implementation is totally disconnected17:33
rm_workand we use a drastically different DB-interaction model, I feel like17:34
rm_workmy hope is that you will try in Octavia and notice that it is not a problem :)17:34
eanderssonWe are pretty lock in with vendors as well, so that does not help :p17:34
rm_workif you have a solution for the neutron-lbaas side, feel free to post it, we're in feature freeze soon but I would consider that a bugfix17:34
rm_workeandersson: Octavia will support vendor plugins very soon17:35
eanderssonI am testing a solution at the moment, but will need to do a lot of testing to make sure that there are no unforseen side-effects.17:35
rm_workas it will be replacing neutron-lbaas completely17:35
rm_workit is no longer just an appliance17:35
rm_workyeah17:35
openstackgerritAnkur proposed openstack/python-octaviaclient master: Initialize plugin for OSC  https://review.openstack.org/44622317:36
eanderssonI am afraid to see what is going to happen once we add another 200 LBs to our environment :p17:37
eanderssonSince we are already at 62s for listening lbs.17:37
rm_workeandersson: it is interesting that no one else is noticing this T_T17:37
rm_workit makes me wonder if no one actually uses this, or cares, or has already just fixed it internally17:37
rm_workor if there's something specific about your deployment that's making this manifest badly17:37
eanderssonMaybe they are using v1?17:38
rm_worklol maybe17:38
rm_workT_T17:38
rm_workplease don't make me depressed this early in the work day17:38
eanderssonAfter seeing the crazy amount of stuff happening there is no way that this can't be an issue at scale :p17:38
eanderssonDon't worry LBAAS is the best <317:38
eanderssonIt could just be that they have smaller setups.17:38
rm_workmaybe17:38
rm_workI know Yahoo/Paypal were working on v2 stuff...17:39
eanderssonMany regions/sites.17:39
rm_worksince they were doing the UI work for lbaasv2 many moons ago17:39
rm_workbut maybe they pivoted, or given their propensity to do stuff internal-only, maybe they fixed it and didn't say anything? lol17:39
rm_workjohnsom: or wait, was their UI thing a weird v1 hack?17:39
eanderssonI think it's pretty common to keep the number of LBs low as well.17:40
rm_workthat was so long ago17:40
rm_workhmm maybe17:40
rm_workyeah RAX definitely would have run into this issue if they went to v2 in their public cloud, rofl17:40
johnsomI think they had a v2 api but never released it publicly17:40
eanderssonAlthough this work is spread out over tenants.17:41
*** cody-somerville has joined #openstack-lbaas17:41
*** cody-somerville has quit IRC17:41
*** cody-somerville has joined #openstack-lbaas17:41
eanderssonSo it's only slow if you have a lot of lbs in a tenant (or if you are an admin and list all lbs for all tenants)17:42
rm_workah right, it'd be per-tenant17:44
*** foutatoro has joined #openstack-lbaas18:01
*** cody-somerville has quit IRC18:03
*** chlong has quit IRC18:04
*** bread has joined #openstack-lbaas18:08
*** bread is now known as Guest1018218:08
*** armax has quit IRC18:18
*** armax has joined #openstack-lbaas18:18
*** chlong has joined #openstack-lbaas18:19
*** armax has quit IRC18:22
openstackgerritIhar Hrachyshka proposed openstack/neutron-lbaas master: Remove subunit-trace fork  https://review.openstack.org/45428418:23
*** armax has joined #openstack-lbaas18:23
*** aojea has joined #openstack-lbaas18:25
*** armax has quit IRC18:25
*** voelzmo has joined #openstack-lbaas18:41
openstackgerritAlexander Amerine proposed openstack/octavia master: blueprint LBaaS Alternative Monitoring IP/Port  https://review.openstack.org/45300118:41
*** gcheresh_ has joined #openstack-lbaas18:41
openstackgerritAlexander Amerine proposed openstack/octavia master: blueprint LBaaS Alternative Monitoring IP/Port  https://review.openstack.org/45300118:42
eanderssonrm_work: So my current fix is silly. I added a new parameter to from_sqlalchemy_model called exclude18:47
eanderssonWhen running the lbaas-loadbalancer-list command I simply pass listeners, pools and stats to exclude18:48
rm_workI would maybe go the other way kinda18:48
rm_workmake it "show_tree=False"18:48
eanderssonyea18:48
rm_workand then if show_tree, include them18:48
eanderssongood idea I'll try that out as well18:48
rm_workand then go to the specific places we might need to actually generate the whole tree18:49
rm_workand set it to true explicitly there18:49
rm_work*pass it as True18:49
eanderssonwith this it's down to 3s from 30s :D18:49
rm_workcool18:49
rm_workyeah you can feel free to push up as early as you want18:50
rm_workwe don't judge (too harshly) :P18:50
eanderssonhaha18:50
rm_workbut it helps us give direction earlier in case it looks like you're going in a very different direction than we'd recommend18:50
nmagnezio/18:50
rm_workor in nmagnezi's case, he pushes up code early anyway and we don't look at it for months and then tell him he went the wrong direction >_>18:51
rm_worksorry nmagnezi18:51
rm_workit helps to poke us :P18:51
nmagneziHA18:51
rm_workmost of us are busy enough that unless people say "hey, please look at this review: <link>" we don't get around to it T_T18:52
rm_workexcept very periodically18:52
openstackgerritAdam Harwell proposed openstack/octavia master: Install DIB from source so depends-on will work  https://review.openstack.org/44470718:53
nmagnezirm_work, so it will be a good idea to poke you and johnsom about this bug that currently blocks user_group auto detect https://bugs.launchpad.net/octavia/+bug/168043518:53
openstackLaunchpad bug 1680435 in octavia "Amphora images based on RHEL flavors should use Python 3 interpreter same as in Ubuntu" [Undecided,New]18:53
rm_workyeah i did actually see that nmagnezi :P18:53
rm_workwhich is why i just looked at my DIB patch again18:53
rm_workthough it is unfortunately unrelated18:53
rm_work*different* py2/3 issue18:53
rm_workbut i had to chec18:54
rm_work*check18:54
* nmagnezi takes a note that he should add a crone job that reminds him to poke Adam once a week18:54
rm_workit's not a bad idea :/18:54
rm_workI've discovered that generally not putting a lot of effort into tracking things (which would probably make me go mad) works fine, because if stuff is important it tends to stay on my radar by itself, and if it isn't, it falls off automatically :P18:55
nmagnezirm_work, also see https://review.openstack.org/#/c/429398/32/octavia/amphorae/backends/agent/api_server/listener.py18:55
nmagnezirm_work, it actually worked :D18:55
nmagnezi(for ubuntu..)18:55
rm_worklol18:55
rm_worki'd caution about the use of bytes / b.decode and such there18:55
rm_workjust because of py2/3 issues18:55
rm_workwhich might be why you're running into this?18:56
rm_workIMO we still need the agent to be 2/3 compat18:56
nmagneziyes, this does not work on python 218:56
rm_workright18:56
rm_workso maybe the key is18:56
rm_workto make it work on either :P18:56
rm_workI can help maybe18:56
rm_workI have done a LOT of this18:56
nmagneziif you have an idea, please speak :) I18:56
rm_worklet me look briefly18:56
nmagneziI'll try to implement it18:56
rm_workwill run some tests18:57
nmagnezii wanted to refrain from "if python2 -> run this else (python3) run that"18:57
nmagnezisince it's... ugly..18:57
rm_workyeah18:57
nmagnezi:)18:57
rm_worki would not18:57
rm_worksec18:57
nmagnezirm_work,  i will be back in ~30min18:58
rm_workkk18:58
rm_workgives me time :)18:58
nmagnezihah18:58
nmagnezior you can wait for me to return :D18:58
*** gcheresh_ has quit IRC18:59
*** gcheresh has joined #openstack-lbaas18:59
*** blogan_ has joined #openstack-lbaas19:03
*** cody-somerville has joined #openstack-lbaas19:05
foutatorohi all, did someone use lbaasv2 in ocata ? it seems that the default ha driver for lbaas in no longer working19:05
*** blogan has quit IRC19:06
*** csomerville has joined #openstack-lbaas19:07
rm_workfoutatoro: that being, Octavia?19:09
rm_workdefault driver for neutrin-lbaas in Ocata should be Octavia19:09
rm_workor do you mean, the haproxy_agent driver?19:10
*** cody-somerville has quit IRC19:10
foutatororm_work: I've installed openstak with openstack-ansible deployer an the default lbaas in this project in ha19:10
rm_workcan you pastebin the neutron-lbaas configuration you are using? with any passwords removed of course19:11
johnsomxgerman I thought the default driver in OSA was octavia?19:13
xgermanYes19:14
xgermanBut Ocata does not have Octavia - so not sure19:15
xgermanOcata Andi LE19:15
xgermanAndi LE19:15
xgermanHate autocorrect19:15
rm_workhmm19:15
johnsomAh, so maybe it's a neutron-lbaas role that used the old haproxy namespace driver19:16
*** voelzmo has quit IRC19:16
foutatorojohnsom: yes the neutron-lbaas role is using the old one19:17
openstackgerritIhar Hrachyshka proposed openstack/neutron-lbaas master: Configure api_extensions= in tempest.conf  https://review.openstack.org/45431519:19
johnsomWe do have a gate that tests the old driver19:19
openstackgerritIhar Hrachyshka proposed openstack/neutron-lbaas master: Configure api_extensions= in tempest.conf  https://review.openstack.org/45431519:19
johnsomBut most of use don't use that driver19:19
openstackgerritBrent Eagles proposed openstack/neutron-lbaas master: Create header with authtoken for requests without args  https://review.openstack.org/45417219:21
johnsomfoutatoro What problem are you seeing?19:22
*** fnaval has quit IRC19:22
foutatorojohnsom: i get this error http://paste.openstack.org/show/605659/19:24
foutatoroI get that in lbaas namespace19:26
rm_worknmagnezi: ok got a suggested course of action19:31
rm_workthat is ... when you try to connect to the VIP?19:31
rm_workthat just looks like you don't have any members in the haproxy config that was generated19:31
rm_worki am not actually sure why the regex i posted in that comment works, but it seems to <_<19:36
rm_worknmagnezi: http://www.regexpal.com/?fam=9734319:38
*** mdurrant has joined #openstack-lbaas19:39
mdurrantI've looked all over the web and can't find a spec for how flavors will be implemented in LBaaS or Octavia. Could someone in the know please point me in the right direction? Thank you.19:39
rm_worknmagnezi: anyway I tested that stringio thing in py2 and py3 assuming bytes as input, and it should work19:39
xgermanmdurrant: https://review.openstack.org/#/c/392485/19:42
xgermanwe have a spec but it needs some more ironing out19:42
rm_workI guess the issue might be if py2 doesn't return bytes from the stream object?19:42
xgermanso please comment19:42
johnsom^^^ Faster on the paste than I was19:42
xgermanback from my coffee run…19:42
mdurrantThank you xgerman19:42
mdurrant:)19:42
*** openstack has joined #openstack-lbaas19:48
openstackgerritMerged openstack/neutron-lbaas master: Updated from global requirements  https://review.openstack.org/45389419:50
openstackgerritMerged openstack/octavia master: Updated from global requirements  https://review.openstack.org/45389719:52
foutatororm_work johnsom: take a look to these outputs http://paste.openstack.org/show/605664/19:52
foutatoroit seems that health is working correctly19:53
openstackgerritBrent Eagles proposed openstack/neutron-lbaas master: Create header with authtoken for requests without args  https://review.openstack.org/45417219:54
*** armax has joined #openstack-lbaas19:56
*** voelzmo has joined #openstack-lbaas19:57
*** voelzmo has joined #openstack-lbaas19:57
johnsomfoutatoro On the host with the haproxy process and the lbaas agent, do a "sudo ip netns"19:58
johnsomThere should be a netns named "lbaas-<uuid>"19:58
johnsomIs there more than one?19:58
foutatorojohnsom: there is only one20:00
*** voelzmo has quit IRC20:01
*** harlowja has quit IRC20:03
johnsomfoutatoro Ok, good.  Run "ip netns exec lbaas-<uuid> curl http://192.168.74.11:8080" and see if it responds20:03
foutatorojohnsom: yes it works I can reach both members from lbaas namespace20:06
foutatorobut not with VIP20:06
johnsomOk, that is odd then.20:06
johnsomfoutatoro Can you pastebin the haproxy config file it is using?  The path should be visible in the process list20:07
foutatorojohsom: netstat -tulpn20:07
foutatorotcp        0      0 192.168.74.5:80         0.0.0.0:*               LISTEN      13880/haproxy20:07
johnsomAlso, there should be a log file for haproxy somewhere20:07
rm_worknmagnezi: let me know when you're back around20:07
foutatorojohnsom: where can I finf these config because the /etc/haproxy/haproxy.cfg contains default configs20:09
johnsomfoutatoro do a "ps -ef | grep 13880"20:10
foutatorojohnsom: nobody   13880     1  0 20:49 ?        00:00:01 haproxy -f /var/lib/neutron/lbaas/v2/6a79a66a-a383-4f54-ab92-62179948a362/haproxy.conf -p /var/lib/neutron/lbaas/v2/6a79a66a-a383-4f54-ab92-62179948a362/haproxy.pid -sf 1384620:11
johnsomCool, it is this file: /var/lib/neutron/lbaas/v2/6a79a66a-a383-4f54-ab92-62179948a362/haproxy.conf20:11
foutatorojohnsom: http://paste.openstack.org/show/605668/20:13
johnsomfoutatoro That config file looks good to me.20:15
foutatorojohnsom: do you have a working heat template ?20:16
johnsomI have never used heat myself20:16
johnsomfoutatoro If you don't mind installing socat could you run this?20:20
johnsomecho "show servers state" | socat unix-connect:/var/lib/neutron/lbaas/v2/6a79a66a-a383-4f54-ab92-62179948a362/haproxy_stats.sock stdio20:20
foutatorojohnsom: I just delete the stack. i wil try again20:22
johnsomfoutatoro Thanks, curious what the issue is20:26
foutatorojohnsom: I will run again using openstack cli then retry with heat20:27
*** armax has quit IRC20:30
openstackgerritAlexander Amerine proposed openstack/octavia master: blueprint LBaaS Alternative Monitoring IP/Port  https://review.openstack.org/45300120:31
*** rcernin has quit IRC20:44
*** gcheresh has quit IRC20:46
rm_workreedip: not sure what you did with the HM patch... was that a merge?20:54
rm_workseems like it reintroduced some old code somehow, and a few pep8 errors20:54
rm_worki'm fixing it up now20:54
*** chlong has quit IRC20:57
*** openstackstatus has quit IRC20:59
*** openstackstatus has joined #openstack-lbaas21:01
*** ChanServ sets mode: +v openstackstatus21:01
*** harlowja has joined #openstack-lbaas21:03
nmagnezirm_work, hey21:04
nmagnezirm_work, sorry it took longer than expected :D21:05
rm_worknp21:06
rm_workmade comments, and see my pings21:06
*** openstackgerrit has quit IRC21:07
*** chlong has joined #openstack-lbaas21:10
*** openstackgerrit has joined #openstack-lbaas21:17
openstackgerritAdam Harwell proposed openstack/octavia master: Introduce Octavia v2 API for Health Monitor  https://review.openstack.org/40625621:17
rm_workok still a few failing functional tests... still haven't had time to track down why it isn't flushing out the statuses properly sometimes O_o21:17
rm_workI hope tomorrow I can do that, if reedip doesn't have time before me21:18
*** sputnik13 has quit IRC21:33
*** sputnik13 has joined #openstack-lbaas21:34
*** catintheroof has quit IRC21:38
*** bzhao_ has joined #openstack-lbaas21:43
rm_workwhelp, looks like my logs are outdated...21:50
rm_worknmagnezi: let me know if you have questions, i might have missed stuff, my IRC client closed and didn't actually store logs21:53
*** fnaval has joined #openstack-lbaas21:54
openstackgerritGerman Eichberger proposed openstack/octavia master: ACTIVE-ACTIVE Topology: Initial Distributor Noop Driver  https://review.openstack.org/31300621:58
nmagnezixgerman, that was a quick adoption..21:59
nmagnezi:)21:59
*** openstack has joined #openstack-lbaas22:04
*** chlong has quit IRC22:05
*** aojea has quit IRC22:07
*** sputnik13 has quit IRC22:10
foutatorojohnsom: still here22:11
johnsomYes22:11
johnsomTrying to figure out why my venv isn't playing nice with the openstack plugin22:11
*** sputnik13 has joined #openstack-lbaas22:12
foutatorojohnson: I'm lbaas ha driver by creting env with neutron. now the lbaas works when I run wget to the private ip of q-lbaas22:14
foutatorobut not floating ip22:14
johnsomAh, it was the floating IP that wasn't working?22:15
foutatorojohnsom: this is not the previous lab with heat. I buid resources using CLI22:18
johnsomfoutatoro Ok22:18
foutatorojohnsom: I think that the official heat template has some mistakes22:19
johnsomfoutatoro Ok, can you file a bug for the heat template?  https://bugs.launchpad.net/heat-templates/+filebug22:23
foutatoroyeah22:24
johnsomThanks22:24
*** bzhao has quit IRC22:27
*** bzhao has joined #openstack-lbaas22:27
*** sputnik13 has quit IRC22:37
*** sputnik13 has joined #openstack-lbaas22:38
*** sputnik13 has quit IRC22:39
*** sputnik13 has joined #openstack-lbaas22:40
*** fnaval has quit IRC23:00
openstackgerritAnkur proposed openstack/python-octaviaclient master: Initialize plugin for OSC  https://review.openstack.org/44622323:05
rm_workok hopefully soon py3 gates will be passing23:06
rm_worko/23:06
rm_workunless there's ANOTHER bug in DIB down the line, but hopefully this is it23:06
johnsom+123:06
*** netchord has quit IRC23:12
*** gongysh has joined #openstack-lbaas23:18
ankur-gupta-f4rm_work: When you get a chance. A lil love https://review.openstack.org/#/c/446223/23:23
ankur-gupta-f4xgerman: Also you ^^^ please and thank you :)23:25
openstackgerritAlexander Amerine proposed openstack/octavia master: blueprint LBaaS Alternative Monitoring IP/Port  https://review.openstack.org/45300123:53
« Wednesday, 2017-04-05 Index Friday, 2017-04-07 »

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!