Tuesday, 2015-09-08

« Monday, 2015-09-07 Index Wednesday, 2015-09-09 »
*** mlavalle has quit IRC01:01
*** Varun_Lodaya has quit IRC01:03
openstackgerritZhaoBo proposed openstack/neutron-lbaas: neutron lbaas should add the new appended timestamp field  https://review.openstack.org/22080401:13
*** ganeshna has joined #openstack-lbaas01:14
*** apuimedo_ has quit IRC01:15
*** apuimedo has quit IRC01:15
*** apuimedo_ has joined #openstack-lbaas01:16
*** apuimedo has joined #openstack-lbaas01:16
*** vivek-ebay has joined #openstack-lbaas01:17
*** vivek-ebay has quit IRC01:22
*** minwang2 has quit IRC01:32
*** yuanying has quit IRC01:32
*** ganeshna has quit IRC01:33
openstackgerritPhillip Toohill proposed openstack/octavia: Adding new network driver for containers  https://review.openstack.org/19785801:34
*** yuanying has joined #openstack-lbaas01:35
openstackgerritPhillip Toohill proposed openstack/octavia: Updates for containers functionality  https://review.openstack.org/19995401:43
*** sballe has quit IRC01:46
*** yuanying has quit IRC01:52
*** yuanying has joined #openstack-lbaas01:53
*** yuanying has quit IRC02:01
*** yuanying has joined #openstack-lbaas02:01
*** yuanying has quit IRC02:02
*** blogan_ has joined #openstack-lbaas02:02
*** Varun_Lodaya has joined #openstack-lbaas02:05
*** yuanying has joined #openstack-lbaas02:13
*** yuanying has quit IRC02:15
*** apuimedo_ has quit IRC02:23
*** yuanying has joined #openstack-lbaas02:38
*** blogan_ has quit IRC02:48
*** amotoki has joined #openstack-lbaas02:48
*** blogan_ has joined #openstack-lbaas02:49
*** vivek-ebay has joined #openstack-lbaas02:49
*** yuanying has quit IRC02:50
*** yuanying has joined #openstack-lbaas02:54
*** yuanying_ has joined #openstack-lbaas03:42
*** yuanying has quit IRC03:44
*** blogan_ has quit IRC03:57
*** blogan_ has joined #openstack-lbaas03:57
*** minwang2 has joined #openstack-lbaas04:00
*** vivek-eb_ has joined #openstack-lbaas04:10
*** vivek-ebay has quit IRC04:13
*** vivek-eb_ is now known as vivek04:14
*** vivek is now known as Guest7918804:14
*** Guest79188 has quit IRC04:23
*** ganeshna has joined #openstack-lbaas05:09
*** nmagnezi has joined #openstack-lbaas05:14
*** numans has joined #openstack-lbaas05:15
*** nmagnezi has quit IRC05:15
*** vivek has joined #openstack-lbaas05:20
*** vivek is now known as Guest4609505:20
*** ganeshna has quit IRC05:22
*** Guest46095 has quit IRC05:24
*** ganeshna has joined #openstack-lbaas05:30
*** vivek has joined #openstack-lbaas05:54
*** vivek is now known as Guest107105:54
santosh__Hi All06:08
*** Varun_Lodaya has quit IRC06:15
*** Guest1071 has quit IRC06:17
*** Alex_Stef has joined #openstack-lbaas06:41
santosh__Hi All, with current neutron-lbaas , service is always trying to access barbican tls containers of admin tenants as admin user.06:42
*** minwang2 has quit IRC06:48
*** apuimedo_ has joined #openstack-lbaas07:03
*** ganeshna has quit IRC07:03
rm_worksantosh__: we're going to be using Barbican ACLs to handle auth for that, so it should be ok07:21
rm_worksantosh__: we'll attempt to access any user's containers/secrets as our admin user, the idea is that they will have granted our account access to their secrets via Barbican ACL endpoint07:21
*** ganeshna has joined #openstack-lbaas07:51
*** chlong has quit IRC07:56
*** ganeshna has quit IRC07:57
*** ganeshna has joined #openstack-lbaas07:59
*** ganeshna has quit IRC08:00
*** ganeshna has joined #openstack-lbaas08:11
*** ganeshna has quit IRC08:29
*** ganeshna has joined #openstack-lbaas08:40
*** vjay13 has joined #openstack-lbaas08:48
*** vjay13 has quit IRC08:57
*** ganeshna has quit IRC09:00
*** ganeshna has joined #openstack-lbaas09:03
*** kiran-r has joined #openstack-lbaas09:11
*** ganeshna has quit IRC09:31
*** ganeshna has joined #openstack-lbaas09:32
*** ganeshna has quit IRC09:34
*** ganeshna has joined #openstack-lbaas09:48
*** blogan_ has quit IRC10:15
*** blogan_ has joined #openstack-lbaas10:15
openstackgerritZhaoBo proposed openstack/neutron-lbaas: neutron lbaas should add the new appended timestamp field  https://review.openstack.org/22080410:24
*** _kiran_ has joined #openstack-lbaas10:29
*** kiran-r has quit IRC10:30
*** ganeshna has quit IRC11:05
*** _kiran_ is now known as kiran-r11:11
*** ebagdasa has joined #openstack-lbaas11:55
*** chlong has joined #openstack-lbaas12:22
*** amotoki has quit IRC12:54
*** ganeshna has joined #openstack-lbaas13:03
*** woodster_ has joined #openstack-lbaas13:13
*** ganeshna has quit IRC13:15
*** ganeshna has joined #openstack-lbaas13:23
*** kiran-r has quit IRC13:40
*** vjay13 has joined #openstack-lbaas13:42
*** ganeshna has quit IRC13:46
*** vjay13 has quit IRC13:49
*** vjay13 has joined #openstack-lbaas13:49
*** rohara has quit IRC13:52
*** pai15 has joined #openstack-lbaas13:53
*** vjay13 has quit IRC13:59
*** amotoki has joined #openstack-lbaas14:00
*** localloop127 has joined #openstack-lbaas14:09
openstackgerritMerged openstack/octavia: Updated from global requirements  https://review.openstack.org/22111414:12
*** localloop127 has quit IRC14:14
*** ajmiller has joined #openstack-lbaas14:17
*** pai15 has quit IRC14:25
*** pai15 has joined #openstack-lbaas14:26
*** blogan_ has quit IRC14:30
*** pai15 has quit IRC14:30
*** blogan_ has joined #openstack-lbaas14:30
*** pai15 has joined #openstack-lbaas14:31
*** blogan_ has quit IRC14:33
*** sballe has joined #openstack-lbaas14:40
*** apuimedo_ has quit IRC15:00
*** Alex_Stef has quit IRC15:22
*** bradjones has joined #openstack-lbaas15:28
*** bradjones has quit IRC15:28
*** bradjones has joined #openstack-lbaas15:28
*** Varun_Lodaya has joined #openstack-lbaas15:40
*** devlaps has joined #openstack-lbaas15:43
*** numans has quit IRC15:49
*** pai15 has quit IRC15:54
*** madhu_ak has joined #openstack-lbaas16:01
*** Varun_Lodaya has quit IRC16:02
*** pai15 has joined #openstack-lbaas16:03
*** Aish has joined #openstack-lbaas16:05
*** vivek has joined #openstack-lbaas16:06
*** vivek is now known as Guest346016:06
*** rm_mobile has joined #openstack-lbaas16:19
*** rm_mobile has joined #openstack-lbaas16:19
rm_mobileajmiller: you in the office?16:19
pai15hello, one question regarding lbaas v2 fip association to vip. it gets associated to default security group16:19
rm_mobilesbalukoff: are you in your office? :P16:20
pai15there is no way to tie it to a custom security group - workaround is to open the default group - which customer may not want to do in production16:20
pai15any thoughts?16:20
sbalukoffrm_mobile: Not today. Was planning a work-from-home day. I'll probably be in my office tomorrow.16:21
rm_mobileHeh k16:21
sbalukoffAre you still in Seattle? Did you want to come work from our office?16:21
rm_mobileI'm about to be downtown and looking for a desk and wifi16:21
rm_mobileMin mentioned she'd be in so I'm probably good16:21
rm_mobileJust covering based16:21
rm_mobile*bases16:21
sbalukoffOk. You can still work from the Blue Box offices if you want. I can let them know you're coming, eh.16:22
rm_mobileHehe16:22
johnsomrm_mobile Ping me if you can't find someone in the office16:22
rm_mobileBig Blue Box office? :P16:22
sbalukoffHaha!16:22
rm_mobileYou there johnsom?16:22
sbalukoffIt's still mostly the Blue Box office. But it's more crowded now. XD16:23
johnsomrm_mobile No, but I can ping people there16:23
xgermanrm_mobile ajmiller is there — just talked with him on hangout16:23
rm_mobileI assumed you'd be in corvalis16:23
rm_mobileAh yes16:23
xgermanyeah, we know people in that office :-)16:23
rm_mobileI should be good, still on the bus and need to ship something at the FedEx there16:24
*** Aish has quit IRC16:25
*** mlavalle has joined #openstack-lbaas16:25
*** Guest3460 has quit IRC16:26
*** Aish has joined #openstack-lbaas16:26
*** minwang2 has joined #openstack-lbaas16:30
ajmillerrm_mobile16:35
ajmillerHi16:35
rm_mobileYo16:36
rm_mobileProlly be there around 10:15 depending on how long shipping at the FedEx takes16:37
*** Varun_Lodaya has joined #openstack-lbaas16:38
rm_mobileI'll put you as my escort when I do get there I guess16:39
pai15anyone has any input on https://bugs.launchpad.net/neutron/+bug/1295424 ?16:40
openstackLaunchpad bug 1295424 in neutron "lbaas security group" [Low,In progress] - Assigned to Jaume Devesa (devvesa)16:40
pai15openstack: thanks for confirming16:43
*** amotoki has quit IRC16:46
xgermanpai15 agreed16:47
xgermanyou have a fix?16:47
*** madhu_ak has quit IRC16:48
pai15xgerman: for the vip issue? no, I don't have a fix. wanted to check what's the pulse on it and if it would be worthwhile to increase importance/priority on that.16:52
xgermanok, I think we are busy with reaching our Liberty goal (Octavia being reference implementation)16:53
pai15so it is not slated for Liberty I guess.16:55
*** Aish has quit IRC16:56
*** harlowja has joined #openstack-lbaas17:03
*** KunalGandhi has joined #openstack-lbaas17:05
*** nmagnezi has joined #openstack-lbaas17:09
*** KunalGandhi has quit IRC17:10
*** vivek has joined #openstack-lbaas17:12
*** vivek has quit IRC17:12
rm_mobileajmiller: elevator17:13
rm_mobileajmiller,  minwang2: reception17:13
*** KunalGandhi has joined #openstack-lbaas17:14
*** pai15 has quit IRC17:15
*** rm_mobile has quit IRC17:18
rm_workcool, melody still works :P17:18
*** Aish has joined #openstack-lbaas17:24
rm_workblogan: you there?17:36
*** bana_k has joined #openstack-lbaas17:37
*** Varun_Lodaya_ has joined #openstack-lbaas17:42
*** Varun_Lodaya has quit IRC17:43
*** Varun_Lodaya_ is now known as Varun_Lodaya17:43
*** pai15 has joined #openstack-lbaas17:45
*** KunalGandhi has quit IRC18:06
*** pai15_ has joined #openstack-lbaas18:10
*** madhu_ak has joined #openstack-lbaas18:12
*** pai15 has quit IRC18:14
bana_klbaas v2 agent is not listed in the neutron agent-list command output18:17
*** openstackgerrit has quit IRC18:31
*** openstackgerrit has joined #openstack-lbaas18:31
*** KunalGandhi has joined #openstack-lbaas18:53
*** nmagnezi has quit IRC18:57
*** Aish has quit IRC19:00
*** minwang2 has quit IRC19:02
*** Aish has joined #openstack-lbaas19:02
*** apuimedo has quit IRC19:05
openstackgerritPhillip Toohill proposed openstack/octavia: Adding new network driver for containers  https://review.openstack.org/19785819:40
openstackgerritPhillip Toohill proposed openstack/octavia: Updates for containers functionality  https://review.openstack.org/19995419:40
*** crc32 has joined #openstack-lbaas19:43
*** minwang2 has joined #openstack-lbaas19:53
rm_workptoohill: hey19:56
ptoohillhey19:57
ptoohillrm_work:19:58
rm_workptoohill: so, the containers code you're working on19:58
rm_workis that kinda RAX specific or would it work in devstack?19:58
ptoohillit could work in devstack19:58
rm_workawesome, ok19:59
rm_workhow is that coming along19:59
ptoohill?19:59
rm_workis it like... working end-to-end? :P19:59
ptoohillits been there for a while19:59
rm_worki haven't heard anything about it19:59
rm_workmust just be out of the loop19:59
rm_worki saw a lot of reviews still in flight19:59
ptoohillthough, with all these changes and rebases im about to retest it19:59
rm_workok19:59
rm_worki am interested in your test script / test config19:59
ptoohillyea, no one has really cared to try it out19:59
rm_worki am about to be doing similar testing19:59
ptoohillno script really, the config is what makes it because were connecting to lab and using idfferent images20:00
rm_workright so i mean20:00
rm_workI use https://gist.githubusercontent.com/rm-you/f7585ca4932b3ee1eed9/raw/30243d33dd1803a4c7b773460fddcbb4c4d26134/new_octavia_devstack.sh20:01
rm_workwhat would I have to do20:01
*** Aish has quit IRC20:02
ptoohillSo, I was also going to work on a way so we can have multiple octavias using n-lbaas set up in lab. I'll take this to a pm20:02
*** bharath has joined #openstack-lbaas20:06
rm_workminwang2: so to recreate the tests you are doing (with 15 minute amphora boot time!) what do I need to do? just set up devstack with octavia/neutron-lbaas with default settings and then ... kick off tempest somehow?20:18
crc32anyone else getting "impl.MismatchError: 404 != 400" on CR   -> 220747 -> 202336 -> 160061 -> 2018825 -> 218735?20:18
rm_workis that the HM/UDP chain?20:18
crc32yea it seems the UDP chain keeps getting pushed further and further down the stack. :(20:18
rm_work:(20:19
crc32So I get to inherit a lot of bugs. :/20:19
johnsomrm_work We are going to talk about this tomorrow.  The issue is with hosts that don't emulate vt-x, qemu switches to TCG engine which is software only.  The kernel alone takes ~7 minutes20:19
minwang2rm_work, before that you might wantt o create flavor with 1g, the default flavor size is big, then you change the flavor_id in etc/octavia/octavia.conf20:19
rm_workjohnsom: yeah, i am going to try it on RAX and see if it is similarly bad or if it is ok20:19
rm_workminwang2: ok will do that20:19
johnsomrm_work Just devstack with Octavia, create a load balancer.  Watch how long it takes to go active20:20
rm_workjohnsom: worst case if it works on RAX, we could try to specify RAX only for that gate?20:20
johnsomThe RAX infra instances don't work20:20
johnsomWe have tried that20:20
rm_workjohnsom: ah ok20:20
johnsomAt least not the main ones.20:21
*** nmagnezi has joined #openstack-lbaas20:21
rm_worki'll time it on my own instances and see if it's any better (probably it would be the same)20:21
johnsomIn linux, the /proc/cpuinfo needs to have vmx or svm listed20:21
johnsomIt's less than a minute on hosts with vt-x20:22
johnsomWhich kicks qemu into kvm mode20:23
*** Aish has joined #openstack-lbaas20:23
johnsomcrc32 For a bit we were having the problem due to the new wsme.  This fixed it: https://review.openstack.org/#/c/217021/20:24
johnsomcrc32 but if you are up to date on that failover stack, it's bugs in the network stuff Brandon added last week20:25
johnsomhere: https://review.openstack.org/#/c/220747/20:25
crc32I testing straight from tox20:25
johnsomIt's probably the bugs then.  We just need to wait for those to get fixed and rebase20:26
xgermanouch — wish things were more stable20:31
ptoohillgetting there, need more eyes and testing happening on these thigns i guess.20:32
*** tiny-hands has joined #openstack-lbaas20:35
crc32testing thighs?20:42
crc32I guess I'll just review -R for now. I can't catch up with the rebase waves.20:43
*** clduser_ has quit IRC20:44
*** nmagnezi has quit IRC20:49
*** nmagnezi has joined #openstack-lbaas20:49
*** nmagnezi has quit IRC20:50
*** nmagnezi has joined #openstack-lbaas20:51
rm_workblogan: do you know what the workflow is for the tempest tests for neutron-lbaas+octavia?20:55
rm_workblogan: is it: Create via Neutron-LBaaS; NLBaaS -> Octavia -> Test; Cleanup via NLBaaS; NLBaaS -> Octavia ?20:57
rm_workblogan: neutron seems to create the port and it seems octavia therefore can't clean up the port or subnet during the cleanup phase (per design), but then is nlbaas supposed to clean it up?20:57
rm_workwatching min run the tests and it dies trying to remove the subnet on the octavia side during cleanup, which it should know isn't possible20:58
ptoohillblogan is out today i believe21:02
ptoohillTheres a patch for the ports out there somewhere21:02
ptoohillMost of these gotchas have been discovered and reviews for them out there. Just need people to put them together and get them through. Problem is, many are adding up and its becoming quite daunting imo21:03
*** minwang2 has quit IRC21:03
*** minwang2 has joined #openstack-lbaas21:04
xgermanok, will see if I can right bologna’s patch21:05
*** clduser has joined #openstack-lbaas21:07
ptoohillrm_work: https://review.openstack.org/#/c/214058/ and this one https://review.openstack.org/#/c/217439/ address your concerns i believe.21:08
ptoohillcrc32: yes, we need to test thighs!21:08
rm_workptoohill: do you know how to run only a single test with testr or whatever tox uses?21:15
rm_workinstead of a whole directory21:15
bana_ksomething like this21:18
bana_ktox -e py27 neutron.tests.unit.test_manager.NeutronManagerTestCase.test_service_plugin_is_loaded21:18
bana_ki think21:18
bana_krm_work  : if this is what you are looking for21:19
crc32rm_work where do I drop the EvenStream class and can you confirm the emit() method will just take a UpdateDb Object that we discussed before right?21:19
rm_workbana_k: yes21:20
rm_workbana_k: thanks21:20
rm_workcrc32: i believe so -- not sure exactly where to put it though... er... let me think a sec21:20
rm_workemit() would need to take either a StatsUpdate or a HealthUpdate -- or just an Update if you merged them somehow21:21
crc32I did.21:21
rm_workk21:21
*** nmagnezi has quit IRC21:22
rm_workjohnsom: flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat clflush mmx fxsr sse sse2 syscall nx rdtscp lm constant_tsc rep_good nopl pni pclmulqdq ssse3 cx16 sse4_1 sse4_2 x2apic popcnt tsc_deadline_timer aes hypervisor lahf_lm21:25
rm_workjohnsom: don't see the two you mentioned but maybe "vme" is it or something? because... nova instance booted in like 20 seconds21:25
johnsomNope21:26
johnsomHmm, if you ps -ef | grep qemu  What is the accel?21:26
rm_workhmm ok min is showing me that maybe it isn't ACTUALLY done yet21:27
crc32I got the vmx flag on my CPU set.21:28
crc32whats it for?21:28
johnsomyeah, check the console log21:28
rm_workhmm it is tcg... so yeah i guess it is not done :/21:28
rm_workyeah looking. nova says it's ACTIVE + Running21:28
rm_workwhich apparently doesn't mean what you'd think it means21:29
johnsomcrc32 the vmx flag means vt-x is available under linux which makes qemu use kvm and can boot instances quickly21:29
rm_workas console still says it's building21:29
crc32I also have the flag f00fbug21:29
crc32just kidding21:30
*** tiny-hands has quit IRC21:31
johnsomYeah, that would mean you are running on an old proc21:31
crc32do they really have a flag for that? lol21:31
crc32oh wow they do.21:32
rm_workanyone else notice that this doesn't read correctly from config? https://github.com/openstack/neutron-lbaas/blob/master/neutron_lbaas/drivers/octavia/driver.py#L4321:32
crc32I'll put the Queing code in the controller for now.21:34
crc32I'll figure out where to stuff EvenStreamer later21:34
*** pai15_ has quit IRC21:34
*** pai15 has joined #openstack-lbaas21:35
*** devlaps has quit IRC21:38
rm_workweird, looks like it goes in neutron.conf and not neutron-lbaas.conf :(21:39
rm_workweird21:39
*** pai15 has quit IRC21:39
rm_workjohnsom: ok, LB create seems to go active in ~4-5 min21:41
rm_workwith tcg21:41
openstackgerritSherif Abdelwahab proposed openstack/octavia: Amphora Flows and Service Drivers for Active Standby  https://review.openstack.org/20625221:41
rm_workanyone know what this means?21:42
rm_work{0} setUpClass (neutron_lbaas.tests.tempest.v2.api.test_health_monitor_admin.TestHealthMonitors) ... SKIPPED: Neutron support is required21:42
rm_work... i can run "neutron lbaas-loadbalancer-create" and it works fine to create LBs21:42
rm_workneutron is running...21:42
rm_workit says that for 100% of the tests21:45
ptoohillmight be a config option thing21:45
ptoohill?21:46
rm_worki mean... what config option would "neutron support" be21:46
rm_worknetwork_driver = allowed_address_pairs_driver21:47
rm_workfor octavia21:47
rm_workfnaval: you there?21:47
crc32yes hes here21:51
fnavalyea21:51
fnavali'm here21:51
crc32see21:51
fnavalsup21:51
rm_workerk i might have figured it out21:52
rm_workhad commented out the enable_service tempest21:52
fnavalyea21:52
rm_workin my devstack config and didn't notice21:52
fnavalyou need that line in your localrc21:52
fnavalyou'll need it for all of the neutron ones too21:53
fnavalrm_work: here's my updated devstack script https://gist.github.com/fnaval/ea396b405fd6447568c821:53
fnavalas of last night21:53
fnavalworks on a local ubuntu 14.04.03 system21:54
*** crc32 has quit IRC21:55
fnavalrm_work: btw, do you know why the barbican folks removed running of the functional tests outside of localhost?21:55
fnavaltheir configs are all staticly definied21:55
rm_workyeah21:56
rm_worknot sure what the changes happening right now are21:56
rm_workbut was listening to a discussion of changes to the functional tests even more just today21:57
*** woodster_ has quit IRC21:59
openstackgerritGerman Eichberger proposed openstack/octavia: Plug vip and networks by port mac address  https://review.openstack.org/22074722:21
xgerman^ hope that will do some good22:22
rm_workjohnsom: so22:28
rm_workflags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat clflush mmx fxsr sse sse2 syscall nx rdtscp lm constant_tsc rep_good nopl pni pclmulqdq ssse3 cx16 sse4_1 sse4_2 x2apic popcnt tsc_deadline_timer aes hypervisor lahf_lm22:28
rm_workjohnsom: you sure nothing else there would make it faster? I know it says accel=tcg22:28
rm_workbut22:28
rm_workit is passing 3 tests in under 2 minutes22:29
xgermanthat is RAX hardware?22:30
xgermanI know who will sponsor our 3rd party CI for that ;-[)22:30
rm_workheh22:30
rm_workyes22:30
rm_workthis is RAX public cloud22:30
rm_workjohnsom: what would be different from the last time someone tried it?22:30
rm_workjohnsom: you said someone else already tested on RAX cloud, didn't you? and that they also got bad results?22:31
*** Varun_Lodaya has quit IRC22:36
rm_workjohnsom: who did the HP and RAX tests, and what size instances were they run on? which flavor?22:38
johnsomIt was a week or two ago22:49
johnsomIt was an infra instance launched by the gate.  the hostname said rax something22:49
johnsomrm_work how can it pass three tests in two minutes if the amp boot is 4-5?22:52
rm_workamp boot is < 1m22:52
rm_workonce i got all the config fixed22:53
johnsomNot the nova claimed boot, but actually running22:53
rm_workyeah22:53
rm_worktests are actually passing22:53
johnsomWhat config did you change?22:53
rm_work{0} neutron_lbaas.tests.tempest.v2.api.test_health_monitor_admin.TestHealthMonitors.test_create_health_monitor_empty_tenant_id_field [21.306326s] ... ok22:53
rm_work{0} neutron_lbaas.tests.tempest.v2.api.test_health_monitor_admin.TestHealthMonitors.test_create_health_monitor_for_another_tenant_id_field [21.408491s] ... ok22:53
rm_work{0} neutron_lbaas.tests.tempest.v2.api.test_health_monitor_admin.TestHealthMonitors.test_create_health_monitor_missing_tenant_id_field [11.242169s] ... ok22:53
rm_workfor example22:53
rm_workalthough can't run a whole lot more because of the port/subnet deletion bug22:54
rm_workjohnsom: the ones that min showed me22:54
rm_workrunning another manual test now just to verify22:55
johnsomJust the timeouts?  But above you said it took 4-5 minutes for the LB create (which is a two mintues faster than I have seen before without vt-x)22:55
rm_workstart time: Tue Sep  8 22:54:53 UTC 201522:55
rm_workyeah this one is taking longer again22:58
rm_workfinish time: Tue Sep  8 22:58:30 UTC 201522:58
rm_workso that's 3:40 or so22:58
rm_workstarting another Tue Sep  8 22:59:42 UTC 201522:59
rm_workmaybe the tests don't log the full time?23:00
*** crc32 has joined #openstack-lbaas23:00
rm_workjohnsom: although infra also said to do any tests on HP nodes on the 30G flavor due to CPU restrictions on the 8G23:01
rm_workwhich they think would be the cause of VMs taking so long23:01
johnsomWe don't get to pick the flavor for their gates, right?23:01
rm_workright23:02
rm_worktheir gates use 30G HP, 8G RAX23:02
johnsomIt was a RAX node for the last test run that bombed anyway23:02
rm_workthe job exists already?23:02
rm_workis it an experimental job?23:02
rm_workcan you link me to some of the runs?23:02
rm_worki didn't think it actually existed yet23:02
johnsomYeah, this is what we have been testing on.  It's broken at the very moment, but for the was two weeks we have been adjusting up timeouts23:03
rm_workjohnsom: https://gist.github.com/rm-you/f7585ca4932b3ee1eed9#file-new_octavia_devstack-sh-L70-L7623:03
rm_workjohnsom: that is what i'm using in my testing23:03
johnsomrm_work THIS ONE IS TOTALLY BROKEN and not representative: https://review.openstack.org/#/c/209675/23:03
johnsomBut that is the patchset with the gate in it.23:04
rm_workthe job is gate-neutron-lbaasv2-dsvm-scenario ?23:04
johnsomIt is paired to: https://review.openstack.org/#/c/218613/23:04
johnsomLikewise broken23:04
johnsomgate-neutron-lbaasv2-octavia-dsvm-api23:05
rm_workk23:10
rm_workyeah23:10
rm_workwe need this to *pass*, or at least get it partially working23:10
rm_workeven if it takes 30 hours23:10
rm_workjust so we can have a baseline to point the infra people at for debugging, because they are interested in why this is not working well for us23:10
johnsomYep23:11
rm_worksince there are other projects that do spin up VMs internally without issue23:11
johnsomYeah, we talked to two of those teams.23:11
rm_workso we need to fix the port/subnet issue23:11
johnsomThey only spin up one or two, so the time isn't important to them.23:11
rm_workhmm23:11
openstackgerritMichael Johnson proposed openstack/octavia: Adding amphora failover flows  https://review.openstack.org/20233623:12
rm_workit still shouldn't take that long23:12
openstackgerritMichael Johnson proposed openstack/octavia: health manager service  https://review.openstack.org/16006123:12
rm_workis https://review.openstack.org/#/c/220747/ ready? xgerman23:12
openstackgerritMichael Johnson proposed openstack/octavia: Implement UDP heartbeat sender and receiver  https://review.openstack.org/20188223:12
rm_workI can test it but do I need to also test it alongside a change on the neutron side?23:12
openstackgerritMichael Johnson proposed openstack/octavia: Adds the NO_MONITOR operational status for members  https://review.openstack.org/22023223:13
minwang2ajmiller create a bigger instance for me, i will test it there see if it speed up23:13
crc32olso_messaging Notifeir objects contain a emit() method. am I supposed to implement a notifier or wrap around it with a "has_a" notifier object. Or should I just stick to plain old RPC?23:26
rm_workhmm23:27
rm_workinteresting, i wonder if oslo_messaging already has something for us23:27
crc32just wondering what made you pick the "emit()" method name.23:27
rm_worki picked it because it's the term for what we're doing23:27
rm_workprobably the same reason they picked it23:27
crc32Notfifers look more like there for logs but they have a emit(record) entry.23:27
rm_worki hadn't heard of theirs before23:27
*** chlong has quit IRC23:27
rm_workit was not related23:28
crc32um. I've heard of send but never crossed my mind to call it emit23:28
crc32http://docs.openstack.org/developer/oslo.messaging/notifier.html23:28
rm_workit's a pretty standard term for this pattern23:28
crc32looks like they have there own little nameing scheme for the message.23:28
rm_workI may have gotten it from one of my college programming patterns textbooks23:28
*** chlong has joined #openstack-lbaas23:28
crc32And what pattern would that be?23:29
crc32its not the observer pattern right?23:29
rm_worki forget the exact name, but it's like... a thing to observe changes, but no, not observer23:29
rm_worknot base observer at least23:29
rm_workbut it is kinda related23:29
rm_workGTK also uses it extensively with their event system i think23:30
rm_worklike this: http://www.marco.panizza.name/dispenseTM/slides/exerc/eventNotifier/eventNotifier.html23:30
rm_workEvent Notifier23:30
rm_workthat uses emit23:30
rm_workpublisher/subscriber model maybe?23:31
rm_workhttps://en.wikipedia.org/wiki/Publish%E2%80%93subscribe_pattern23:31
rm_workIE, a queue23:31
rm_workbut, i don't see emit() on that page23:31
crc32well like I said these notifiers look like there log related. Even the doc for emit(record) implies its to emit the log record to the messaging notification system.23:31
crc32look for the word emi(record)23:31
rm_workright so i guess maybe that one isn't for us23:31
rm_workbut we're essentially using the same pattern23:32
crc32its right below warn and warning.23:32
rm_workyeah, i mean, it doesn't really matter what they are doing23:32
crc32yea so should I just use Notifier?23:33
rm_workunless it's exactly what we want, in which case cool, but if not, ignore it and just keep writing your thing23:33
rm_workwell, would it *work* if we tried to use it for what we're doing?23:33
crc32eventStream = oslo_messaging.Notifier(target=blah, blah, blibbidyblah)23:33
rm_workor would it be square peg round hole23:33
rm_workif it "just works", then great23:33
crc32I don't even know if it works period. I'm just gonna give it a shot.23:33
rm_workkk23:34
rm_workworth investigating before we possibly reinvent the wheel23:34
crc32we had barely gotten oslo to work via RPC.23:34
rm_worknice find23:34
rm_worklol23:34
crc32even though RPC is totally not what we are doing. :/23:34
crc32I'll just assume Notifier was supposed to be general but the docs and some methods are specific to logging.23:36
rm_workyeah so this is probably where i got the term emit stuck in my head for this kind of pattern: http://www.gtk.org/tutorial1.2/gtk_tut-18.html23:36
rm_workfrom using GTK way back in the day23:36
rm_workand it stuck23:36
rm_workyeah that seems correct, i can't imagine it'd be logging only23:36
*** vivek-ebay has joined #openstack-lbaas23:36
crc32so your using the term in the event driven context. Odd that the terms stuck for non event driven contexts.23:38
*** Varun_Lodaya has joined #openstack-lbaas23:38
rm_workthe design i had in mind for this *is* event-driven23:38
rm_workit's supposed to hook in to the "event" function triggered by a UDP message coming in with data23:38
rm_workthat seems event-driven to me23:38
crc32like event driven meaning the reciever is going to use callbacks on certain messages?23:39
rm_workyes, if the receiver gets a health message, it should emit() for the notifier23:39
rm_workat the same time that it also does an UpdateDB23:39
rm_workprobably immediately following the UpdateDB I guess23:40
rm_workat least that was the design in my head23:40
rm_workjohnsom: ok so this can't be right23:44
rm_work{0} neutron_lbaas.tests.tempest.v2.api.test_health_monitor_admin.TestHealthMonitors.test_create_health_monitor_empty_tenant_id_field [21.462123s] ... ok23:44
rm_work{0} neutron_lbaas.tests.tempest.v2.api.test_health_monitor_admin.TestHealthMonitors.test_create_health_monitor_for_another_tenant_id_field [21.228237s] ... ok23:44
rm_work{0} neutron_lbaas.tests.tempest.v2.api.test_health_monitor_admin.TestHealthMonitors.test_create_health_monitor_missing_tenant_id_field [21.518437s] ... ok23:44
rm_workbecause it did take longer than 2 minutes just to spin up one VM23:45
rm_workso maybe those tests don't count the VM spinup in their time, and maybe they only require one VM for the whole set?23:45
johnsomMin tells me they create and delete the lb PER test, which is why we are concerned23:46
rm_workit doesn't look that way23:48
rm_worklooks like the setup creates one LB for that suite of tests23:48
rm_workand it is done before the test timers start23:48
johnsomThat would excellent23:48
rm_workso yeah i really want to see how long even a "slow" run takes to finish23:49
rm_workso we need to get this shit fixed T_T23:49
rm_worki am trying to review https://review.openstack.org/#/c/220747/ now23:49
rm_workI think it is related?23:49
johnsomIt should certainly help, but I haven't reviewed it yet23:50
johnsomThis might also be related: https://review.openstack.org/#/c/214058/23:52
*** ajmiller has quit IRC23:55
*** woodster_ has joined #openstack-lbaas23:57
« Monday, 2015-09-07 Index Wednesday, 2015-09-09 »

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