Monday, 2016-12-12

« Sunday, 2016-12-11 Index Tuesday, 2016-12-13 »
*** ducttape_ has quit IRC00:12
*** amotoki has joined #openstack-lbaas00:14
*** tovin07 has joined #openstack-lbaas00:48
*** tovin07_ has joined #openstack-lbaas00:51
*** tovin07_ has quit IRC00:52
*** tovin07 has quit IRC00:56
*** tovin07 has joined #openstack-lbaas00:57
*** tovin07_ has joined #openstack-lbaas00:58
*** yuanying has joined #openstack-lbaas01:03
*** kevo has joined #openstack-lbaas01:08
*** ducttape_ has joined #openstack-lbaas01:13
*** ducttape_ has quit IRC01:18
*** ducttape_ has joined #openstack-lbaas01:42
*** ducttape_ has quit IRC01:53
*** armax has joined #openstack-lbaas02:06
*** armax has quit IRC02:08
*** amotoki has quit IRC02:11
*** kevo has quit IRC02:13
*** yatin_on_leave has quit IRC02:16
*** yatin_on_leave has joined #openstack-lbaas02:27
*** ducttape_ has joined #openstack-lbaas02:41
*** yuanying has quit IRC02:49
*** ducttape_ has quit IRC03:05
*** yatin_on_leave has quit IRC03:18
*** ducttape_ has joined #openstack-lbaas03:31
*** yatin_on_leave has joined #openstack-lbaas03:31
*** Alex_Stef has joined #openstack-lbaas03:49
openstackgerritSindhu Devale proposed openstack/octavia: [WIP] Align Octavia API to n-lbaasv2 for listeners  https://review.openstack.org/40506003:59
*** ducttape_ has quit IRC04:10
openstackgerritSindhu Devale proposed openstack/octavia: [WIP] Align Octavia API to n-lbaasv2 for listeners  https://review.openstack.org/40506004:17
*** ducttape_ has joined #openstack-lbaas04:19
openstackgerritSindhu Devale proposed openstack/octavia: [WIP] Align Octavia API to nlbaasv2 for pools  https://review.openstack.org/40592204:28
*** ducttape_ has quit IRC04:29
openstackgerritSindhu Devale proposed openstack/octavia: [WIP] Align Octavia API to nlbaasv2 for pools  https://review.openstack.org/40592204:31
*** korean101 has joined #openstack-lbaas04:34
korean101hi guys04:35
korean101anyone knows this error? (http://paste.openstack.org/show/592061/)04:35
korean101"generate-subunit 1481514746 2344 fail" on devstack04:35
openstackgerritSindhu Devale proposed openstack/octavia: [WIP] Align Octavia API to n-lbaasv2 for members  https://review.openstack.org/40593204:39
openstackgerritSindhu Devale proposed openstack/octavia: Align Octavia API to n-lbaasv2 for L7Policy  https://review.openstack.org/40632804:49
*** rstarmer has joined #openstack-lbaas04:52
*** Alex_Stef has quit IRC04:53
*** Alex_Stef has joined #openstack-lbaas04:55
*** yuanying has joined #openstack-lbaas04:58
openstackgerritSindhu Devale proposed openstack/octavia: Align Octavia API to n-lbaasv2 for L7Rules  https://review.openstack.org/40633604:58
*** chenhb has joined #openstack-lbaas05:19
chenhbHi,every body. We create a ci job in our gate05:26
chenhbWe start the lbaas and octavia service05:27
chenhbIt was failed, this is error info link  http://paste.openstack.org/show/59206205:29
*** ducttape_ has joined #openstack-lbaas05:30
chenhbit occurred error when cloning diskimage-builder in octavia_install05:31
chenhbWho can help me? Thanks05:33
*** ducttape_ has quit IRC05:35
openstackgerritSindhu Devale proposed openstack/octavia: [WIP] Align Octavia API to n-lbaasv2 for listeners  https://review.openstack.org/40506005:44
openstackgerritSindhu Devale proposed openstack/octavia: [WIP] Align Octavia API to nlbaasv2 for pools  https://review.openstack.org/40592205:50
openstackgerritSindhu Devale proposed openstack/octavia: [WIP] Align Octavia API to n-lbaasv2 for listeners  https://review.openstack.org/40506005:51
openstackgerritSindhu Devale proposed openstack/octavia: [WIP] Align Octavia API to n-lbaasv2 for members  https://review.openstack.org/40593205:52
*** reedip has quit IRC05:53
openstackgerritSindhu Devale proposed openstack/octavia: Align Octavia API to n-lbaasv2 for L7Policy  https://review.openstack.org/40632805:53
openstackgerritSindhu Devale proposed openstack/octavia: Align Octavia API to n-lbaasv2 for L7Rules  https://review.openstack.org/40633605:54
*** reedip has joined #openstack-lbaas06:03
openstackgerritAbhishek G M proposed openstack/neutron-lbaas: Fix for error "no attribute tenant_id"  https://review.openstack.org/40603306:08
*** openstackgerrit_ has joined #openstack-lbaas06:24
*** kevo has joined #openstack-lbaas06:24
*** openstackgerrit_ has quit IRC06:25
*** openstackgerrit_ has joined #openstack-lbaas06:26
*** openstackgerrit_ has quit IRC06:27
*** ducttape_ has joined #openstack-lbaas06:31
*** openstackgerrit_ has joined #openstack-lbaas06:34
*** openstackgerrit_ has quit IRC06:35
*** ducttape_ has quit IRC06:36
*** openstackgerrit_ has joined #openstack-lbaas06:59
*** openstackgerrit_ has quit IRC07:00
*** eezhova has joined #openstack-lbaas07:15
*** ihrachys has joined #openstack-lbaas07:19
*** kevo has quit IRC07:23
*** yamamoto has quit IRC07:26
*** kobis has joined #openstack-lbaas07:31
*** ducttape_ has joined #openstack-lbaas07:32
*** ducttape_ has quit IRC07:37
*** eezhova has quit IRC07:43
rm_workkorean101: that isn't the real error, the real error happened before that somewhere... let me check07:50
rm_workah yeah, korean101 and chenhb you are both probably having the same problem, which is that currently we have a problem with the image build process in devstack07:50
rm_workwe're working on resolving it, hopefully early this week07:51
rm_workso, at the moment, that is going to happen :(07:51
rm_workthere is a series of patches in both diskimage-builder and octavia that kinda works, but i think still has an issue07:51
*** yamamoto has joined #openstack-lbaas08:09
*** yamamoto_ has joined #openstack-lbaas08:10
korean101rm_work: OMG! thank you for your qucik answer~08:12
korean101rm_work: really appreciate08:13
*** yamamoto has quit IRC08:13
*** yamamoto_ has quit IRC08:20
chenhb@rm_work, thanks.waiting..... :)08:22
rm_workkorean101: np, if you follow the chat here you will see when we figure it out, hopefully08:22
rm_workchenhb / korean101: this is the main patch: https://review.openstack.org/#/c/356590/08:23
rm_workthe others (in diskimage-builder) are linked as "depends-on" in the commit message08:24
*** yamamoto has joined #openstack-lbaas08:24
rm_workwith those it will build the image successfully, though it seems like it may still have an issue that prevents the agent from running correctly, so it looks like it's still a little bit broken08:25
rm_workI'll be looking at it in the morning (still midnight here)08:25
chenhbWe have added the related project to the variable $PROJECT.  Is it useful to resolve the git_clone error?08:29
*** jsheeren has joined #openstack-lbaas08:29
*** pcaruana has joined #openstack-lbaas08:30
*** amoralej|off is now known as amoralej08:31
*** yamamoto has quit IRC08:34
*** pcaruana has quit IRC08:37
*** pcaruana has joined #openstack-lbaas08:38
*** jsheeren has quit IRC08:39
*** jsheeren has joined #openstack-lbaas08:40
*** eezhova has joined #openstack-lbaas08:41
*** nmagnezi has joined #openstack-lbaas08:52
*** ducttape_ has joined #openstack-lbaas09:34
*** ducttape_ has quit IRC09:38
*** yuanying has quit IRC09:43
*** yamamoto has joined #openstack-lbaas09:48
*** yamamoto has quit IRC09:49
*** strigazi_AFK is now known as strigazi09:49
*** ihrachys has quit IRC09:57
*** tovin07_ has quit IRC10:09
*** yamamoto has joined #openstack-lbaas10:27
*** yamamoto has quit IRC10:30
*** ducttape_ has joined #openstack-lbaas10:35
*** ducttape_ has quit IRC10:39
*** yamamoto has joined #openstack-lbaas10:43
*** ducttape_ has joined #openstack-lbaas10:50
*** yamamoto has quit IRC10:51
*** ihrachys has joined #openstack-lbaas11:01
*** yamamoto has joined #openstack-lbaas11:03
*** yamamoto has quit IRC11:08
*** ducttape_ has quit IRC11:15
*** ducttape_ has joined #openstack-lbaas11:18
*** ducttape_ has quit IRC11:44
*** yamamoto has joined #openstack-lbaas12:04
*** yamamoto has quit IRC12:05
openstackgerritValleriya Perelman proposed openstack/octavia: ACTIVE-ACTIVE Topology - create distributor network flow  https://review.openstack.org/40976312:22
openstackgerritValleriya Perelman proposed openstack/octavia: Active-Active Topology - Cluster DB Tasks  https://review.openstack.org/40976412:22
openstackgerritValleriya Perelman proposed openstack/octavia: Active-Active Topology - register/uregister amphorae tasks  https://review.openstack.org/40976512:22
*** rstarmer has quit IRC12:30
*** ducttape_ has joined #openstack-lbaas12:45
*** ducttape_ has quit IRC12:50
*** chenhb has quit IRC12:58
openstackgerritNir Magnezi proposed openstack/neutron-lbaas: Auto reschedule loadbalancers from dead agents  https://review.openstack.org/29999813:27
*** xgerman has quit IRC13:28
*** xgerman has joined #openstack-lbaas13:29
ajopsutil/_psutil_linux.c:13:20: fatal error: Python.h: No such file or directory13:41
ajohmm13:41
nmagnezijohnsom, ping: a question about octavia/neutron-lbaas codebases squash13:41
*** jsheeren has quit IRC13:41
nmagneziajo, ^^ I got the same thing an hour ago :(13:42
*** amoralej is now known as amoralej|lunch13:45
bcafarelhttps://review.openstack.org/#/c/356590/ in progress for the Python.h fun13:49
*** amotoki has joined #openstack-lbaas13:50
*** amotoki has quit IRC14:01
*** ducttape_ has joined #openstack-lbaas14:10
*** eezhova_ has joined #openstack-lbaas14:16
*** eezhova has quit IRC14:18
*** Matias has quit IRC14:18
*** Matias has joined #openstack-lbaas14:23
*** amoralej|lunch is now known as amoralej14:25
openstackgerritNir Magnezi proposed openstack/neutron-lbaas: Auto reschedule loadbalancers from dead agents  https://review.openstack.org/29999814:33
johnsomnmagnezi ajo That error about python.h is caused by the ongoing issues the diskimage-builder project is having after a change they merged two Fridays ago.14:37
nmagnezijohnsom, Michael we should come up with something to pre-block such issues14:37
nmagnezijohnsom, i will send a mail to the mailing list with some possible suggestions :)14:38
nmagnezijohnsom, maybe we can follow up on the weekly meeting14:38
johnsomGreg and Adam have been working on it.  I put up some patches that fixed it, but the DIB team are attempting to fix it another way.14:38
johnsomnmagnezi yes, we should probably add a simple octavia gate to their project.  This isn't the first time they have merged a broken change....14:39
*** ducttape_ has quit IRC14:40
openstackgerritValleriya Perelman proposed openstack/octavia: ACTIVE-ACTIVE Topology - Initial Cluster Manager  https://review.openstack.org/40523814:42
openstackgerritValleriya Perelman proposed openstack/octavia: ACTIVE-ACTIVE Topology Initial Distributor Noop Driver  https://review.openstack.org/31300614:42
openstackgerritValleriya Perelman proposed openstack/octavia: ACTIVE-ACTIVE Topology - create distributor network flow  https://review.openstack.org/40976314:42
openstackgerritValleriya Perelman proposed openstack/octavia: Active-Active Topology - register/uregister amphorae tasks  https://review.openstack.org/40976514:42
openstackgerritValleriya Perelman proposed openstack/octavia: Active-Active Topology - Cluster DB Tasks  https://review.openstack.org/40976414:42
openstackgerritValleriya Perelman proposed openstack/octavia: ACTIVE-ACTIVE Topology OVS-based Distributor Driver  https://review.openstack.org/31762914:42
openstackgerritValleriya Perelman proposed openstack/octavia: ACTIVE-ACTIVE Topology OVS-based Distributor Backend  https://review.openstack.org/32042214:42
openstackgerritValleriya Perelman proposed openstack/octavia: ACTIVE-ACTIVE Topology - Distributor related tasks  https://review.openstack.org/40695114:42
openstackgerritValleriya Perelman proposed openstack/octavia: ACTIVE-ACTIVE - controller network tasks  https://review.openstack.org/32348114:42
openstackgerritValleriya Perelman proposed openstack/octavia: ACTIVE-ACTIVE Topology - Distributor image creation  https://review.openstack.org/40359414:42
openstackgerritValleriya Perelman proposed openstack/octavia: ACTIVE-ACTIVE - network driver related changes  https://review.openstack.org/32249414:42
openstackgerritValleriya Perelman proposed openstack/octavia: ACTIVE-ACTIVE - distributor certificate tasks  https://review.openstack.org/40695214:42
openstackgerritValleriya Perelman proposed openstack/octavia: ACTIVE-ACTIVE Topology - distributor creation flow  https://review.openstack.org/40695314:42
openstackgerritValleriya Perelman proposed openstack/octavia: ACTIVE-ACTIVE Topology - create shared distributor  https://review.openstack.org/40695414:42
*** ducttape_ has joined #openstack-lbaas14:42
johnsomnmagnezi What is your question about the merge?14:43
nmagnezijohnsom, just want to verify i get the roadmap for this14:44
nmagnezijohnsom, so first i started to look at https://etherpad.openstack.org/p/octavia-ocata-merge-review-priority14:44
nmagnezijohnsom, in order to review related patches14:44
nmagnezijohnsom, now, do we expect to finish API alignment on Ocata?14:45
johnsomYes, first up is getting the lbaasv2 API moved into Octavia14:45
johnsomnmagnezi We are still trying for Ocata yes14:45
nmagnezijohnsom, ack. so if we make it to Ocata, what is expected to happen with the neutron-lbaas codebase?14:46
johnsomThen we are working on making the drivers work in Octavia and adding a proxy to neutron API14:47
johnsomIt will be deprecated with the pass through proxy code allowing compatibility14:48
*** ducttape_ has quit IRC14:48
nmagneziin Ocata or afterwards?14:49
johnsomTo be determined based on how14:49
johnsomFar we get with the code14:49
nmagneziack. what i'm trying to figure out here is a usecase in which an operator is still using haproxy in namespace for Ocata. will is still work assuming he did not deploy Octavia at all?14:51
johnsomThis week is Ocata-2 milestone, so, frankly we are running short on time14:51
johnsomProbably.  We need to move the namespace driver over to Octavia14:52
*** cody-somerville has joined #openstack-lbaas14:53
*** cody-somerville has quit IRC14:53
*** cody-somerville has joined #openstack-lbaas14:53
johnsomI suspect the pass through proxy will need to be an optional setting in Ocata, just due to the timing.14:53
nmagnezijohnsom, how will that work? will you add the lbaas agent to octavia?14:53
johnsomCorrect.  I think it will need some re-work to break some dependencies on neutron code14:54
nmagnezijohnsom, by pass through proxy you mean the API endpoint will be part of Octavia, but the driver is in neutron-lbaas codebase?14:55
nmagnezijohnsom, yeah i think i blogan said it would need re-factoring in the last weekly meeting14:55
johnsomRight, code in neutron that forwards the request through to the Octavia API endpoint14:56
*** csomerville has quit IRC14:56
johnsomTo maintain backward compatibility via the neutron API14:57
nmagnezisounds like a lot of fun :)14:58
johnsomHa, yeah, a lot of work, but in the long run it will be good for the project.14:59
johnsomNo more editing there repos for some API changes and just one database instead of two we have now, etc.15:00
johnsomThere->three15:01
johnsomOn my mobile as it is still early here15:01
nmagneziIRCing on mobile? nice :D15:02
nmagnezire: ^^, totally agree, this move makes a lot of sense, and we will be able to move faster afterwards15:03
nmagnezijohnsom, to we have an estimate about when the broken diskimage stuff is going to get fixed? I can't stack :<15:08
johnsomYeah, well, I had a fix up for review the same day, but the 'new' fixes didn't work all week.  Adam is working with the DIB team.  We really do need to get DIB fixed today.15:10
johnsomIt is slowing our merge work too15:10
openstackgerritNir Magnezi proposed openstack/neutron-lbaas: Auto reschedule loadbalancers from dead agents  https://review.openstack.org/29999815:18
nmagnezijohnsom, ack.15:18
*** ducttape_ has joined #openstack-lbaas15:19
nmagnezijohnsom, sorry for asking this again, but we are in Ocata-2. any chance to get some cycles for https://review.openstack.org/#/c/344658/ and https://review.openstack.org/299998 ? those are pending from Newton15:19
johnsomYeah, don't be sorry for bugging us on that.  They need to be reviewed15:20
nmagnezijohnsom, I rebased to verify they are in a good shape. https://review.openstack.org/#/c/344658/ is. as for https://review.openstack.org/299998 some changes were needed (detailed in the bug and in the patch) - fixed now  as well.15:20
nmagnezijohnsom, in the future, maybe we can use process monitor for haproxy in the amphora? (just suggesting)15:23
johnsomWe already have that capability built in15:24
*** fnaval has joined #openstack-lbaas15:29
benoithi guys15:46
benoiti m testing lbaasv2 and i wonder if the SSL terminaison could be managed without barbican ?15:47
benoiti saw there is a local_cert_manager class, but couldnt find if it's usable15:48
*** armax has joined #openstack-lbaas15:48
xgermanThat one is defunct15:49
xgermanAt this point we only support babican15:50
*** armax has quit IRC15:54
benoitthanks xgerman for your reply, so for now, if i want to do ssl ofloading with lbaasv2, i must install barbican15:55
*** _ducttape_ has joined #openstack-lbaas15:56
xgermanYes, we felt that is best since we didn’t want to manage certificates15:56
benoityes it sounds to be the appropriate solution15:56
benoitjust i dont have barbican installed :)15:56
xgermanBenoit not sure how current this is but https://wiki.openstack.org/wiki/Network/LBaaS/docs/how-to-create-tls-loadbalancer15:57
*** ducttap__ has joined #openstack-lbaas15:58
*** _ducttape_ has quit IRC15:58
*** ducttape_ has quit IRC15:59
benoitxgerman, thanks for sharing it but i'm evaluating it outside devstack (on a rhosp labs witch do not have barbican for now)16:01
xgermanK, might still be useful - since the steps should work similarly in a bigger env.16:02
johnsomThere is also a description here: http://docs.openstack.org/developer/octavia/guides/basic-cookbook.html#deploy-a-tls-terminated-https-load-balancer16:02
ankur-gupta-fMorning16:04
*** armax has joined #openstack-lbaas16:04
benoityes indeed, both links should be useful16:05
nmagnezixgerman, hi German, could you please spend some cycles on https://review.openstack.org/#/c/344658/ and https://review.openstack.org/#/c/299998/ ? i also pinged johnsom about those, which are pending for review since Newton16:06
xgermanThey took my +2 away but will look ;-)16:07
johnsomSounds like a great way to start building back that review history to get the +2 back...  Grin16:07
*** Alex_Stef has quit IRC16:07
nmagnezijohnsom, lol16:07
nmagnezixgerman, nevertheless, much appreciated :)16:08
xgerman:-)16:08
*** kobis has quit IRC16:09
*** Alex_Stef has joined #openstack-lbaas16:25
*** kobis has joined #openstack-lbaas16:34
*** diogogmt has joined #openstack-lbaas16:35
openstackgerritLubosz Kosnik (diltram) proposed openstack/octavia: Implement sorting and pagination for octavia  https://review.openstack.org/38214716:35
*** sshank has joined #openstack-lbaas16:39
nmagnezixgerman, thank you for the review :)16:40
xgermanYeah, I am not sure where this fits in our jigsaw puzzle of moving out of Neutron…16:41
nmagnezixgerman, why is that?16:42
nmagnezialso, we didn't move yet :)16:42
xgermanIt looked to me like you are importing from neutron and not neutron_lib… but I haven’t kept up on the neutron discussions16:43
johnsomYeah, work hasn't started on moving that yet, so still valid to merge in neutron-lbaas16:43
johnsomAh, yeah, the neutron-lib thing is a different issue.  If it's available in neutron-lib, it should import from there16:44
johnsomThat is also a moving target though16:45
diltramxgerman: may you take a look again into review?16:45
xgermandiltram will do16:46
diltramthx16:46
nmagnezijohnsom, well xgerman reviewed one of my patches. please give him his superpowers back :D16:46
johnsomhahahaa16:47
diltramrm_work, blogan: https://review.openstack.org/#/c/265065/ - this review also needs a little of yours love16:47
xgermandiltram sonar didn’t like your patch - FYI16:47
*** eezhova_ has quit IRC16:48
*** ducttap__ has quit IRC16:48
diltramehhhh :/16:48
*** ducttape_ has joined #openstack-lbaas16:48
johnsomWhich one?  265065?16:49
openstackgerritLubosz Kosnik (diltram) proposed openstack/octavia: Implement sorting and pagination for octavia  https://review.openstack.org/38214716:49
*** kobis has quit IRC16:50
diltramjohnsom: 38214716:55
diltramrebased and working16:55
*** nmagnezi has quit IRC16:57
*** pcaruana has quit IRC16:59
*** kobis has joined #openstack-lbaas17:01
*** kobis has quit IRC17:03
rm_workmorning17:05
rm_workyeah i know, really need to get stacks working today17:05
*** kevo has joined #openstack-lbaas17:15
rm_workdiltram: i THINK that patch is fine, i just wanted to test first in devstack, and that means basically everything is blocked for me by this dib issue17:21
rm_workgreghaynes: i'm hoping we can power through getting this stuff fixed today -- there appears to be something broken even with a "successful build"17:22
greghaynesNice17:45
greghaynesSystemd?17:45
johnsomI added systemd support a while ago17:57
greghayneshrm17:57
johnsomhttps://review.openstack.org/39643817:57
*** openstack has joined #openstack-lbaas18:01
diltramjohnsom: https://review.openstack.org/#/c/38214718:03
diltramokay keystone merged18:17
johnsomWahoo18:18
*** ihrachys has quit IRC18:18
diltramfinally after 11 months it's done :P18:18
*** harlowja has joined #openstack-lbaas18:20
*** reedip has quit IRC18:32
*** yatin_on_leave has quit IRC18:32
*** cody-somerville has quit IRC18:35
*** cody-somerville has joined #openstack-lbaas18:36
*** cody-somerville has quit IRC18:36
*** cody-somerville has joined #openstack-lbaas18:36
johnsomgreghaynes I'm building a new devstack now to take a look at what is going on in the image18:38
greghaynesOk. I have a dr appt and then meeting but I'll have phone irc18:38
greghaynesJust won't be able to hack on fixes until late afternoon18:39
johnsomOk, if you don't mind and it's something obvious I can push up a change.18:39
*** csomerville has joined #openstack-lbaas18:40
*** cody-somerville has quit IRC18:43
*** reedip has joined #openstack-lbaas18:44
*** yatin_on_leave has joined #openstack-lbaas18:45
*** yatin_on_leave has quit IRC18:50
greghaynesDefinitely18:53
*** ihrachys has joined #openstack-lbaas18:53
greghaynes(go for it)18:53
*** ihrachys has quit IRC18:57
rm_workheh, you found someone else willing to +A diltram? :P18:59
diltramyep :)18:59
johnsomdiltram lol -dirty....19:00
rm_workwe might again be at a point where we just need to shove some of this stuff through (as long as it doesn't *totally break everything*) and then fix bugs later19:01
rm_workif we have any hope of making ocata19:01
rm_work*if we want to have any hope19:01
diltram+119:01
rm_worki really wish we could at least see the basic scenarios passing before we merge a TON of stuff tho19:02
rm_workT_T19:02
johnsomNo, I really don't want to have a *broken* release during out inaugural release19:02
johnsomI would rather just not get as far as we hoped.  I.e. get the API in, but maybe not the driver move, etc.19:02
rm_workjohnsom: I SAID, "as long as it doesn't *totally break everything*"19:02
rm_workso that's covered :P19:02
johnsomYeah, I'm just saying this didn't end well for us in the past19:03
rm_workit ended with us getting *this far*19:03
diltrambut we need to merge the code and start merging tests from nlbaas to verify that all is workingg19:03
diltramand based on this we can start fixing any issues19:03
rm_work^^ +119:03
diltrambut without policy, tests we're not able to do anything19:03
*** yatin_on_leave has joined #openstack-lbaas19:03
rm_worki think things would move a lot faster that way19:03
johnsomYeah, I was thinking about that this morning, that I need to get the tests moved over and a gate setup19:03
diltramfirst of all merge the policy and sorting pagination19:04
johnsomRight now, my focus is getting our builds working again19:04
diltrambecause the whole v2 will fail after merging this code19:04
johnsomI'm pretty worried and think we have already broken master19:04
ankur-gupta-fAll we have merged in from the V2 API is the directory init... How is it breaking scenario tests?19:06
johnsomankur-gupta-f http://logs.openstack.org/90/356590/12/check/gate-octavia-v1-dsvm-scenario-ubuntu-xenial-nv/4072b4e/console.html.gz#_2016-12-08_22_47_07_10370519:08
johnsomThat worries me19:08
ankur-gupta-fwhoa. But I still fail how to see how exposing the base of v2 prevents accessing of v1 endpoint. Digging in.19:10
johnsomThanks19:11
openstackgerritMerged openstack/octavia: Use pip install rather than setup.py  https://review.openstack.org/40823819:17
diltramjohnsom: building server to test that19:17
johnsomOk, so our latest DIB issue is cloud-init failing19:27
johnsom[    7.392458] cloud-init[ESC[0;1;31mFAILEDESC[0m] Failed to start Apply the settings specified in cloud-config.19:27
*** _ducttape_ has joined #openstack-lbaas19:27
johnsom[    7.487929] cloud-init[948]: AttributeError: module 'lib' has no attribute 'Cryptography_HAS_SSL_ST'19:27
johnsomhttp://paste.openstack.org/show/592141/19:29
*** rstarmer has joined #openstack-lbaas19:30
*** ducttape_ has quit IRC19:31
*** _ducttape_ has quit IRC19:39
*** ducttape_ has joined #openstack-lbaas19:40
*** Alex_Stef has quit IRC19:41
*** rstarmer has quit IRC19:44
*** rstarmer has joined #openstack-lbaas19:44
*** rstarmer has quit IRC19:44
*** rstarmer has joined #openstack-lbaas19:45
*** rstarmer has quit IRC19:45
*** rstarmer has joined #openstack-lbaas19:46
*** rstarmer has quit IRC19:46
*** rstarmer has joined #openstack-lbaas19:46
*** rstarmer has quit IRC19:47
*** rstarmer has joined #openstack-lbaas19:47
*** rstarmer has quit IRC19:47
*** rstarmer has joined #openstack-lbaas19:48
*** rstarmer has quit IRC19:48
*** rstarmer has joined #openstack-lbaas19:49
*** rstarmer has quit IRC19:49
*** kevo has quit IRC19:51
*** amoralej is now known as amoralej|off19:51
*** kevo has joined #openstack-lbaas19:58
*** woodster_ has joined #openstack-lbaas20:01
johnsomI am pretty sure it's that the cloud image has {"package": "python3-cryptography","arch": "amd64","version": "1.2.3-1ubuntu0.1","status": "ii "} so we get: Requirement already satisfied: cryptography!=1.3.0,>=1.0 in /usr/lib/python3/dist-packages (from octavia==0.9.1.dev109)20:16
johnsomBut 1.2.3 of cryptography is very old20:16
johnsomOk, march, but in python years that is like 5 years old....20:16
rm_workyeah hmmm20:18
johnsomYep.  That line came into cryptography in version 1.3....20:18
rm_workso we need to update our reqs20:18
johnsomSo, that is the problem.20:18
johnsomYeah, I am thinking so20:18
rm_work... which means a g-r check20:18
rm_work*change20:18
rm_workf20:18
rm_workffffff20:18
rm_worki can do it...20:19
johnsomYes, I thought the same, f's included20:19
rm_workwe just need >=1.3.120:19
rm_work?20:19
johnsomWell, that I actually have no idea20:19
rm_workhmm20:19
rm_worki'll just set it to ... latest? :/20:20
johnsomAt least greater than 1.3, but I haven't tested through with that version.20:20
rm_work<_<20:20
rm_workwell since it has specifically !=1.3.020:20
rm_worki assume >=1.3.1 makes sense20:20
johnsomFunny thing here is it's cloud-init that is dying, so it's not really *our* code that is failing/needs something new.20:20
rm_workyeah...20:20
johnsomNot sure why other's aren't seeing this20:20
rm_workuhh20:20
rm_workhmm20:20
rm_workwhelp20:20
johnsomIs it all the py3.x stuff DIB did?20:21
rm_workshould i do the change?20:21
rm_workprobably yes20:21
rm_workmy guess is prior it'd just be using py2 for all the stuff that anyone else touches20:21
johnsomLet me do a local change, spin an image and see if it boots20:21
johnsomSigh, blazing a trail once again....   A trail of fire20:21
rm_workjohnsom: is Octavia doing the PTG thing?20:29
johnsomYes, there will be a few of us there.20:29
johnsomWe will be using the neutron room though.  I opted to not have our own as I didn't know we would have enough folks to warrant it.20:30
xgermanWe can always congregate in the closest pub :-)20:31
greghaynesoh, hah20:43
greghayneswhat installs cloud-init?20:43
greghaynesthats really odd it'd be able to be installed in a broken way like that20:44
greghaynesjohnsom: I bet thats actually a missing libssl-dev dep20:45
johnsomNo, we load that20:46
greghaynesalthough again, if its installed from package that should be pulled in20:46
greghayneshr20:46
johnsomI tracked it back to the actual version20:46
greghayneswell being old shouldnt mean its broken - I suspect its old because its being pulled in as a cloud-init dep (which is terrifying, but w/e) and therefore its using the latest ubuntu packaged version rather than pip20:47
rm_workoh, looks like I probably won't make PTG... timing is bad for me20:47
johnsomThe old version is what ships with xenial20:47
rm_workSo I mean... updating the dep WILL fix this, I believe? right? since it'll be updated throughout openstack20:48
johnsomwe must be getting something that requires the new version that cloud-init uses20:48
rm_workSo do we want to just do that or try to dig in deeper here (like we did with the other thing and that has taken an extra week)20:48
johnsomI'm booting a test image with the requirement change now20:48
greghaynesright, so a wierd thing about cryptography specifically is it uses cffi/pycparsing to generate bindings for libssl at runtime20:49
greghaynesand I suspect that error is basically saying it failed to do so20:49
greghayneswhich generally means something along the lines of headers are missing20:49
*** ihrachys has joined #openstack-lbaas20:50
johnsomNope, literally the code was added to cryptography in 1.320:51
greghaynesoh wait, I see what youre saying - theres no way it could work20:51
johnsomhttps://github.com/pyca/cryptography/blame/master/src/_cffi_src/openssl/ssl.py#L1420:52
greghaynessorry, I was thining the issue was 'why dont we have SSL_ST=', not 'why are we checking that value', youre right20:52
johnsomOk, it looks like that fixed cloud-init, but the amp still failed and reverted.  I need to grab lunch, will be back in a few to debug deeper20:52
greghaynesso the question I still have is - whats installing cryptography and cloud-init? Seems like both of those should be installed from ubuntu packages20:54
greghayneswhich, if they dont work, would mean that ubuntu packaging is broken20:54
rm_workyeah I don't know that much about the cloud-init setup20:58
rm_workwe don't do anything custom during the init process right?20:59
rm_workjust dropping some config files in place20:59
greghaynescan you grab the package versions of python-openssl and cloud-init from dpkg?20:59
greghaynesreally its a mismatch between python-openssl and cryptography thats going on20:59
greghaynesoh!21:00
rm_workif i can actually get one of the amps up in my environment to ssh into >_>21:00
greghaynesdo you all install python packages at the system level?21:00
johnsomThey are both coming in from the cloud image it looks like.  So, yeah, broken with python321:00
greghaynesI bet I know whats going on - something probably does an upgrade of python-openssl21:00
greghaynesas in a pip install -U pyopenssl21:01
*** chlong has joined #openstack-lbaas21:01
johnsomLet me look for that package.  Both cloud-init and cryptography are the ubuntu packages.21:01
greghaynesso that gets upgraded, but the cryptography (which it depends on) doesnt21:01
johnsomI don't see python-openssl in the DIB output21:02
johnsomnot in the manifests either21:03
greghayneshrmm. I have to meeting but I might be able to poke at it duringd21:03
greghaynesis there a log up somewhere (ci maybe?)21:05
johnsomOh, you are looking for the pythhon package OpenSSL, just a sec21:05
greghaynesI think its called pyopenssl21:06
greghaynesin pypi21:06
johnsomOk, there is a version coming in from our requirements, pyOpenSSL>=0.14 # Apache-2.021:07
johnsomCollecting pyOpenSSL>=0.14 (from octavia==0.9.1.dev109)21:07
johnsom  Downloading pyOpenSSL-16.2.0-py2.py3-none-any.whl (43kB)21:07
*** rstarmer has joined #openstack-lbaas21:09
johnsomSo, hmnm, I guess bumping the cryptography version requirement in g-r is the right answer.  I just need to figure out why it is still failing, we might need something in addition21:09
*** rstarmer has quit IRC21:10
johnsomWhy is it still using flask?21:15
rm_workthe change never merged21:15
rm_workto fix that21:15
rm_worki think?21:16
johnsomAh,21:16
johnsomhttp://paste.openstack.org/show/592144/21:16
openstackgerritAnkur proposed openstack/octavia: Introduce API Decorators  https://review.openstack.org/40562121:16
johnsomHere is the agent error21:16
rm_workoh, i guess it did merge21:17
rm_workhmm21:17
rm_workthen i wonder why21:17
openstackgerritAnkur proposed openstack/octavia: Introduce Handlers for V2 API  https://review.openstack.org/40556221:17
rm_workjohnsom: so it shouldn't be using flask still... is it?21:17
johnsomrm_work According to that traceback it is...21:19
rm_workuhhh21:19
rm_workwtf21:20
rm_workhmm well i do notice i didn't actually *remove* flask from our requirements.txt which I should have21:22
rm_workbut i don't see how it'd *use* flask21:23
rm_workonce i can get it actually building the image correctly in devstack i'll verify we're good to yank that and put up a quick change to remove that21:25
johnsomI am spinning a new image with what I think is a fix for the agent.  No idea how that was running before with a scoping issue21:30
rm_workhmm21:31
rm_worklet me know if you trip over how it's using flaask21:31
rm_worki'm looking and it seems like that shouldn't be possible21:31
rm_workthe cloud-init issue would prevent it from even booting to where I can ssh, right?21:32
johnsomcorrect21:32
johnsomThe IP never gets setup21:32
rm_workk21:32
johnsomI think the flask stuff is this: https://github.com/openstack/octavia/blob/master/octavia/amphorae/backends/agent/api_server/plug.py#L18021:32
rm_workaaah21:33
rm_workrofl21:33
rm_workso that needs to be removed21:33
rm_workI guess it "works" still O_o21:33
rm_workbut21:33
rm_workyeah I'll fix that21:34
rm_workand remove flask from reqs21:34
johnsomThis is the line with bad indention that I have no idea how it has been working: https://github.com/openstack/octavia/blob/master/octavia/amphorae/backends/agent/api_server/plug.py#L17121:34
rm_workhmm21:34
rm_workmerged on oct 10...21:35
rm_workwut21:35
johnsomOk, spinning another image as I forgot to commit the fix before building it last time21:40
xgermanFlask is gone?21:52
johnsomOk, yeah, that was the bug.  With that I can create an lb21:53
greghaynesjohnsom: oh but I think cryptography is being installed from a system package?21:54
johnsomxgerman YES.  The port getting lost issue never got fixed and it would return 404's on startup for valid paths, and there was some other strange thing it was doing, so we switched to gunicorn21:54
xgermanAnd everybody is happier21:55
johnsomSo far it looks that way21:55
xgermanAt least we didn’t do pecan ;-)21:56
rm_workjohnsom: ok, if you push up those changes I'll be able to test locally now, i fixed all my weird centOS issues21:56
rm_work"locally"21:56
*** eezhova_ has joined #openstack-lbaas21:57
*** amotoki has joined #openstack-lbaas21:57
greghaynesok, done with meetings21:59
rm_workit's just the two, right?21:59
openstackgerritMichael Johnson proposed openstack/octavia: The amphora agent plug_vip has an indention issue  https://review.openstack.org/40995321:59
johnsomrm_work  ^^^22:00
rm_workkk22:00
johnsomThe requirements.txt we have to do through g-r right?22:00
rm_workyes22:00
rm_workmost of those are quick if it's just a version bump22:00
rm_workdid >=1.3.1 work?22:00
rm_workor, what did you end up doing22:00
johnsomYes22:00
rm_workkk22:00
johnsomI can post it as I can articulate the versions/issues22:01
greghaynesjohnsom: where is that pip install called from?22:01
johnsomcryptography!=1.3.0,>=1.3.1 # BSD/Apache-2.022:01
rm_workok22:01
rm_workerr22:01
rm_workyou can remove != now22:01
rm_workbecause >=1.3.1 handles it22:02
greghaynesbecause id expect a pip install -U to cause cryptography to upgrade...22:02
johnsomha, yes22:02
rm_workthat's why i said 1.3.1 :P22:02
rm_workor maybe >1.3.0 would be more technically accurate22:02
rm_work1.3.1 is a guess22:02
johnsomgreghaynes The image comes with the ubuntu version that is old.  Currently this requirements from global will accept that version and will not upgrade.  We are going to put in a patch to bump global requirements22:03
greghaynesyea, but pip install -U (which is what id expect us to do) should transitively upgrade deps is what im getting at22:03
greghaynesalso, it sounds like youre not using upper-constraints as well22:03
johnsomIt says this: Requirement already satisfied: cryptography!=1.3.0,>=1.0 in /usr/lib/python3/dist-packages (from octavia==0.9.1.dev109)22:04
greghaynesyea, so must just be pip install without -U22:04
johnsomThe pip install is the one you just changed22:04
greghaynesah yea, so try adding a -U and it may fix everything :p22:04
greghayneswithout changin gg-r22:04
greghaynesbut also, we should make that use upper-constraints22:04
johnsomThis patch: https://review.openstack.org/#/c/408238/22:04
johnsomgreghaynes So that patch should have been: pip install -U -c{env:UPPER_CONSTRAINTS_FILE:https://git.openstack.org/cgit/openstack/requirements/plain/upper-constraints.txt} /opt/amphora-agent22:08
johnsomCorrect?22:08
greghayneshuh, I havent ever used a url as -c, usually I just use an on-disk file22:08
greghaynesbut otherwise yes22:08
greghaynes(if that url deal works then that seems awesome)22:09
johnsomIt's what they put in our tox.ini when they added the upper-constraints22:09
greghaynesah, awesome22:09
*** amotoki has quit IRC22:10
*** ihrachys has quit IRC22:11
*** amotoki has joined #openstack-lbaas22:14
*** amotoki has quit IRC22:15
openstackgerritMichael Johnson proposed openstack/octavia: Fix the agent install to use pip -U  https://review.openstack.org/40995622:16
johnsomOk, let's give that a spin22:16
rm_workcool, that's at the end of the chain22:17
rm_worktesting22:17
*** yamamoto has joined #openstack-lbaas22:22
*** eezhova_ has quit IRC22:24
*** kevo has quit IRC22:33
johnsomWell, nope, that didn't work: 2016-12-12 22:44:06.558 | Could not open requirements file: [Errno 2] No such file or directory: '{env:UPPER_CONSTRAINTS_FILE:https://git.openstack.org/cgit/openstack/requirements/plain/upper-constraints.txt}'22:47
johnsomhttp://logs.openstack.org/56/409956/1/check/gate-neutron-lbaasv2-dsvm-scenario-ubuntu-xenial-nv/6be80a3/logs/devstacklog.txt.gz#_2016-12-12_22_44_06_55822:47
rm_workuhh22:48
greghaynesyea that may be a tox thing22:52
greghaynesnot a pip thing22:52
greghaynesso you probably need to use the on-disk version of it22:52
johnsomYeah, I'm just trying to think about the right way to reference that....22:52
greghayneshrm, it might be cached on disk already in the images22:55
greghaynesin /opt/stack/wherever22:55
rm_workyeah i imagine it SHOULD be...22:55
johnsomWell in devstack it is in /opt/stack/requirements but I'm not sure it's the "right" answer to just code that path22:56
rm_workprolly /opt/stack/requirements/upper-constraints.txt22:56
rm_workah22:56
rm_workyeah22:56
greghaynesso youd probably want to make a source-repositories to clone the repo from git.o.o to /tmp/requirements or something, then use DIB_REPOREF_requirements=/opt/wherever22:56
greghaynesso then itll work in the general case by using git.o.o and in infra itll use the cached version, I think you all did this somewhere else...22:58
*** _ducttape_ has joined #openstack-lbaas22:59
*** ducttape_ has quit IRC23:02
*** _ducttape_ has quit IRC23:03
openstackgerritMichael Johnson proposed openstack/octavia: Fix the agent install to use pip -U  https://review.openstack.org/40995623:05
greghaynesoh, for the amphora-agent you all do this23:05
johnsomI just added it to soure-repos as a file and dropped it in the /opt/amphora-agent directory where we drop the agent code23:06
johnsomUsed the URL they use in tox to get it.23:06
rm_workhmm, seems fine, though i wonder if there's some devstack black-magic where it's already in an ENV var somewhere23:07
greghaynesah that should work23:07
rm_workif it is, someone might pop by and fix it later, or not and we might get bit again a year down the road, but i'm not sure it's worth driving ourselves crazy now trying to find out23:07
johnsomYeah, black magic is right.  I searched the elements and didn't see anyone else using it23:07
*** yuanying has joined #openstack-lbaas23:13
*** diogogmt has quit IRC23:18
*** yuanying has quit IRC23:20
*** yuanying has joined #openstack-lbaas23:25
rm_workhmm, i didn't realize how much magic flask was doing just to return responses23:25
openstackgerritMichael Johnson proposed openstack/octavia: Fix the agent install to use pip -U  https://review.openstack.org/40995623:35
*** ducttape_ has joined #openstack-lbaas23:37
greghayneswhat did you all switch to from flask?23:39
*** reedip has quit IRC23:39
*** yatin_on_leave has quit IRC23:39
rm_workgunicorn23:40
greghaynesah23:40
*** ducttape_ has quit IRC23:42
*** sticker has joined #openstack-lbaas23:44
rm_workbut yeah, i missed the part where flask was doing some response generation23:46
rm_workwhich technically still works because it's just wsgi app context stuff23:46
rm_workbut we should probably remove23:46
*** yatin_on_leave has joined #openstack-lbaas23:51
*** fnaval has quit IRC23:52
*** reedip has joined #openstack-lbaas23:53
« Sunday, 2016-12-11 Index Tuesday, 2016-12-13 »

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