Tuesday, 2017-10-03

« Monday, 2017-10-02 Index Wednesday, 2017-10-04 »
johnsomrm_work So, running out of ideas.  My host gets CPU bound quickly. Running in TCP mode greatly increases the CPS.00:00
rm_workhmmm00:01
rm_worklike i said, i think we have different issues00:01
johnsomI did a quick look at perf, but nothing jumps out there.  I might play with oprofile tomorrow00:01
johnsomhttps://www.irccloud.com/pastebin/BY31DGw8/00:03
*** belharar has quit IRC00:05
*** sticker has joined #openstack-lbaas00:16
*** tongl has quit IRC00:26
*** SumitNaiksatam_ has joined #openstack-lbaas01:32
*** SumitNaiksatam has quit IRC01:35
*** SumitNaiksatam_ is now known as SumitNaiksatam01:35
*** tongl has joined #openstack-lbaas02:17
*** yamamoto has joined #openstack-lbaas02:45
*** tongl has quit IRC03:59
*** tongl has joined #openstack-lbaas04:00
*** oanson has quit IRC04:21
*** oanson has joined #openstack-lbaas04:30
*** armax has quit IRC04:35
*** pcaruana has joined #openstack-lbaas05:01
*** gcheresh_ has joined #openstack-lbaas05:18
*** pcaruana has quit IRC05:23
*** tongl has quit IRC05:24
*** gcheresh_ has quit IRC05:34
*** oanson has quit IRC05:36
*** oanson has joined #openstack-lbaas05:36
*** rcernin has joined #openstack-lbaas05:53
*** sanfern has joined #openstack-lbaas05:57
*** eezhova has joined #openstack-lbaas06:16
*** belharar has joined #openstack-lbaas06:17
*** slaweq has quit IRC06:35
*** slaweq has joined #openstack-lbaas06:36
*** slaweq has quit IRC06:40
*** slaweq has joined #openstack-lbaas06:42
*** pcaruana has joined #openstack-lbaas06:44
*** eezhova has quit IRC06:56
*** tesseract has joined #openstack-lbaas07:17
*** aojea has joined #openstack-lbaas07:31
*** aojea has quit IRC07:33
*** eezhova has joined #openstack-lbaas07:33
*** sticker has quit IRC07:49
*** Alex_Staf has joined #openstack-lbaas08:01
*** oanson has quit IRC08:21
*** oanson has joined #openstack-lbaas08:22
*** spectr has joined #openstack-lbaas08:41
nmagnezijohnsom, about https://review.openstack.org/#/c/507986/ , I want review it but had a question (in the patch) ; as for dashboard patches i reviewed them (one of them needs some more work)08:48
*** salmankhan has joined #openstack-lbaas08:58
*** nrado has joined #openstack-lbaas09:05
*** gcheresh_ has joined #openstack-lbaas09:22
*** yamamoto has quit IRC09:24
*** yamamoto has joined #openstack-lbaas09:41
*** yamamoto has quit IRC09:46
*** yamamoto has joined #openstack-lbaas09:46
*** yamamoto has quit IRC09:51
*** gcheresh_ has quit IRC09:57
*** yamamoto has joined #openstack-lbaas10:00
*** sanfern has quit IRC10:07
isantospIm trying to set a dev environment with octavia conf, the steps I did are here: https://stikked.web.cern.ch/stikked/view/b1537594, anyway something is wrong, I cant see how is balancing the load.. any idea?10:16
*** sanfern has joined #openstack-lbaas10:21
nmagneziisantosp, this url looks like an internal cern thing. it asks for login10:30
*** cyberde has joined #openstack-lbaas10:38
*** oanson has quit IRC10:38
*** oanson has joined #openstack-lbaas10:40
*** sanfern has quit IRC10:55
*** spectr has quit IRC11:07
*** belharar has quit IRC11:07
isantospmy fault, sorry11:18
isantosphttps://pastebin.com/ix4GdNkS11:19
nmagneziisantosp, what happens after neutron lbaas-loadbalancer-create --name lb1 private-subnet ?11:19
isantospnmagnezi, everything is created and working apparently, the status is active when I do show11:27
nmagneziisantosp, so what does not work for you exactly?11:37
isantospnmagnezi, actually the aim of this question is if u sow something wrong on that steps, when I do "while true; do echo -e 'HTTP/1.0 200 OK\r\nContent-Length: 8\r\n\r\n<servername>' | sudo nc -l -p 80 ; done " I should see the balancing, but actually nothing happens. probably something unrelated to it..11:41
*** belharar has joined #openstack-lbaas11:57
*** spectr has joined #openstack-lbaas12:24
*** atoth has quit IRC12:33
*** aojea has joined #openstack-lbaas12:33
*** catintheroof has joined #openstack-lbaas12:36
*** catintheroof has quit IRC12:36
*** gcheresh_ has joined #openstack-lbaas12:36
*** catintheroof has joined #openstack-lbaas12:36
*** aojea has quit IRC12:38
*** leitan has joined #openstack-lbaas12:40
*** catintheroof has quit IRC12:41
*** atoth has joined #openstack-lbaas12:47
*** catintheroof has joined #openstack-lbaas12:54
xgerman_o/13:00
*** eezhova_ has joined #openstack-lbaas13:03
*** eezhova has quit IRC13:05
*** chlong has joined #openstack-lbaas13:10
*** chlong_ has joined #openstack-lbaas13:11
*** yamamoto has quit IRC13:16
*** sanfern has joined #openstack-lbaas13:19
*** yamamoto has joined #openstack-lbaas13:35
*** fnaval has joined #openstack-lbaas13:55
*** spectr has quit IRC13:58
*** spectr has joined #openstack-lbaas13:59
nmagnezio/14:04
*** gcheresh_ has quit IRC14:12
nmagneziisantosp, what happens if you simply use 'curl <vip_address>' ?14:13
*** jniesz has joined #openstack-lbaas14:21
*** yamamoto has quit IRC14:26
*** eezhova_ has quit IRC14:26
*** tongl has joined #openstack-lbaas14:30
*** gcheresh_ has joined #openstack-lbaas14:31
*** chlong has quit IRC14:31
*** chlong_ has quit IRC14:31
*** chlong has joined #openstack-lbaas14:32
openstackgerritBar RH proposed openstack/octavia master: [WIP] Assign bind_host ip address per amphora  https://review.openstack.org/50515814:36
*** tongl has quit IRC14:37
*** gcheresh_ has quit IRC14:37
*** belharar has quit IRC14:42
*** belharar has joined #openstack-lbaas14:46
*** leitan has quit IRC14:47
*** sanfern has quit IRC14:55
*** sanfern has joined #openstack-lbaas14:56
*** leitan has joined #openstack-lbaas15:05
openstackgerritBar RH proposed openstack/octavia master: [WIP] Assign bind_host ip address per amphora  https://review.openstack.org/50515815:06
johnsomisantosp It might be that you need to update the security group on the web servers to allow port 8015:07
johnsomI don't see that in your script15:07
*** armax has joined #openstack-lbaas15:10
*** rcernin has quit IRC15:12
*** pcaruana has quit IRC15:20
*** yamamoto has joined #openstack-lbaas15:26
*** yamamoto has quit IRC15:34
*** belharar has quit IRC15:36
*** chlong has quit IRC15:36
*** leitan has quit IRC15:44
*** leitan has joined #openstack-lbaas15:45
*** leitan has quit IRC15:46
*** tongl has joined #openstack-lbaas15:51
*** tonygunk has joined #openstack-lbaas16:14
*** sshank has joined #openstack-lbaas16:24
*** leitan has joined #openstack-lbaas16:29
*** sshank has quit IRC16:29
*** longstaff has joined #openstack-lbaas16:29
*** tesseract has quit IRC16:37
*** sshank has joined #openstack-lbaas16:39
*** eezhova has joined #openstack-lbaas16:41
johnsomFYI: zuulv3 update: http://lists.openstack.org/pipermail/openstack-dev/2017-October/123049.html16:42
johnsomSummary, rolling back to zuul v2 for now16:44
*** SumitNaiksatam has quit IRC16:48
longstaffjohnsom: Hi Michael. I'm working on the provider driver spec and have a quick question: the Octavia docs say that users cannot specifiy both provder and flavor. Why are these mutually exclusive? Can't a provider support different flavors?16:55
johnsomYeah, I think we can allow both, but the have to both be valid16:57
johnsomIt's mutually exclusive currently because a flavor includes the provider, so no point in allowing a provider16:57
*** salmankhan has quit IRC16:59
longstaffThanks.17:00
*** SumitNaiksatam has joined #openstack-lbaas17:15
*** eezhova has quit IRC17:23
*** sanfern has quit IRC17:30
*** Alex_Staf has quit IRC17:31
*** sanfern has joined #openstack-lbaas17:31
*** sshank has quit IRC18:15
*** sshank has joined #openstack-lbaas18:19
*** sanfern has quit IRC18:30
*** aojea has joined #openstack-lbaas18:34
*** aojea has quit IRC18:38
*** sshank has quit IRC18:40
*** sanfern has joined #openstack-lbaas18:53
*** eezhova has joined #openstack-lbaas19:15
*** sanfern has quit IRC19:17
*** catintheroof has quit IRC19:32
*** catintheroof has joined #openstack-lbaas19:32
*** chlong_ has joined #openstack-lbaas19:43
*** chlong has joined #openstack-lbaas19:44
*** chlong has quit IRC19:46
*** eezhova has quit IRC20:09
*** catintheroof has quit IRC20:10
*** catintheroof has joined #openstack-lbaas20:11
*** tomtomtom has joined #openstack-lbaas20:14
*** catintheroof has quit IRC20:42
*** catintheroof has joined #openstack-lbaas20:42
*** sshank has joined #openstack-lbaas20:49
*** belharar has joined #openstack-lbaas20:49
*** chlong_ has quit IRC21:03
*** numans has quit IRC21:03
*** numans has joined #openstack-lbaas21:07
*** sshank has quit IRC21:15
*** sshank has joined #openstack-lbaas21:16
*** armax has quit IRC21:37
*** fnaval_ has joined #openstack-lbaas21:45
*** fnaval has quit IRC21:48
*** fnaval_ has quit IRC22:04
rm_workwtf22:05
rm_work2017-10-03 21:57:25.131 | sed: -e expression #1, char 36: expected newer version of sed22:05
rm_workthat's a new one22:05
rm_worknot sure there IS a newer version of sed T_T22:06
rm_work(during a stack.sh run)22:06
rm_workmaybe caused by something else being null or something22:06
*** belharar has quit IRC22:13
*** tonygunk has quit IRC22:15
*** tonygunk has joined #openstack-lbaas22:16
johnsomCentOS?22:18
johnsomsed (GNU sed) 4.2.2 here22:19
*** fnaval has joined #openstack-lbaas22:20
rm_workthis is actually ubuntu22:20
rm_workalso 4.2.222:20
rm_workso my guess is that it's a red herring, and actually the issue is some arg it normally has is filled by a var and the var is being filled wrong, or something like that22:20
rm_workignoring that for now as i'm switching gears away from perf testing, unfortunately22:21
rm_workgot retasked22:21
rm_workkong: still not sure I can rebase my WIP failover test yet, since I need to have members, and ... that's not done yet?22:27
kongrm_work: alll resource operations methods are inside octavia_tempest_plugin.tests.v2.base.BaseLoadbalancerTest22:28
rm_workhmm22:28
kongrm_work: i put them there instead of service client because  there is assertion in each method22:29
kongand only basic methods are provided in service client22:29
rm_workerrr22:29
kongrm_work: btw, for specific resource test, do i need to move them to api folder?22:29
rm_worknot sure which tests you're thinking of22:30
rm_workbut one sec22:30
kongoctavia_tempest_plugin.tests.v2.scenario.test_loadbalancer.LoadbalancerTest22:30
rm_workyeah... what is in this?22:31
kongcrud for lb22:31
rm_workI only see create/delete loadbalancer22:31
rm_workyes but a LB can't pass traffic without members22:31
rm_workand the failover test is a traffic test22:31
kongok, so you also prefer it's api test right?22:31
rm_workerr22:31
rm_workI guess? i'm not sure what you mean there22:32
rm_workit's a scenario test...22:32
kong1 sec22:32
rm_workwhere would "api" tests go22:32
kongrm_work: https://review.openstack.org/#/c/486775/20/octavia_tempest_plugin/tests/v2/scenario/test_loadbalancer.py22:33
kongi mean, should test_loadbalancer.py reside in api dir instead of scenario dir?22:33
rm_worki don't think so22:34
rm_workit's not just testing the API22:34
rm_workit's testing to make sure the stuff actually is created22:34
rm_workthat's a scenario test22:35
*** longstaff has quit IRC22:40
kongthat's what i thought when i started to write the code, i am asking because someone disagree with that https://review.openstack.org/#/c/492311/1122:40
kongand also lb tests rely on compute/image/network, according to tempest definition https://docs.openstack.org/tempest/latest/field_guide/index.html#scenario-field-guide22:41
*** jniesz has quit IRC22:41
kongthey are all scenario tests, except what i understand is wrong22:41
kongrm_work: and you can not simply rebase your failover test, maybe you need a little bit 'refactor'22:43
rm_workI don't think I agree with Alex22:44
rm_workwe *do* have API Tests ... they're in the main tree, and they're called "functional"22:45
rm_workanything that actually creates an LB and has it go to active, is a scenario22:45
kong+122:46
kongok, so i will keep the tests there for now, maybe need more feedback before i actually change it22:46
kongcould you please review https://review.openstack.org/#/c/486775 when you are available? It is the root patch all other patches depend on22:47
rm_workyeah i just need to wait for Member rebase I think22:47
rm_workto do my rebase22:48
rm_workah yeah i'll look, but again, I really will just end up giving the whole chain a +2 when I verify it works from the end22:48
konghah, ok22:48
rm_workfor these kinda of patches (dumping the whole initial state) it's really tedious to constantly test each patch level22:49
johnsomThere is value to API level tests in the plugin as it's supposed to be branchless22:49
rm_workif the whole thing works by the end, i'm good22:49
rm_workhmm, that's interesting johnsom22:49
rm_workI guess that makes sense22:49
rm_workbut I see a test that creates a LB and a listener on it and deletes them as being ... a kind of scenario, just because it's testing a ton more than just "the api" in octavia22:50
johnsomBut, that said, scenario tests are important here as we do have functional tests that cover the API22:50
johnsomYeah, totally right22:50
kongjohnsom: do you think the test in https://review.openstack.org/#/c/486775/20/octavia_tempest_plugin/tests/v2/scenario/test_loadbalancer.py is api test?22:50
johnsomHmm, I guess, but an incomplete one22:51
johnsomIt's like a "smoke" API test22:51
kongyeah, i know, it's just basic lb api functionality, not cover every test case for lb api22:52
*** sshank has quit IRC22:52
*** armax has joined #openstack-lbaas22:53
kongbut do you think i should move the file to tests/v2/api/test_loadbalancer.py?22:53
*** fnaval has quit IRC22:54
johnsomYes, probably22:55
rm_workeugh i guess if we really want to classify them as such22:55
*** nrado has quit IRC22:55
rm_worki still feel like they're not just API22:55
rm_workthey rely on a bunch of other services to work22:55
johnsomTrue22:55
rm_workwhich seems very scenario22:55
johnsomI guess we need to make a call22:55
johnsomDo we want tempest API tests that use noop drivers?22:56
johnsomOr should they actually drive an LB?22:56
rm_workwe could do that22:56
rm_workbut that'd be ... not what these are22:56
rm_workand also, that'd require us to reconfigure and restart the octavia services22:56
rm_workbetween tests22:56
rm_work(until we get Flavors!)22:56
johnsomYes, it would be different runs by filter22:56
*** sshank has joined #openstack-lbaas22:57
johnsomI think we just move forward with "scenario" tests, stick with our functional API tests.22:57
rm_workit'd be worth looking at later22:57
johnsomIf we need to move the functional API tests into a branchless repo, that is fine22:57
johnsomI mean, it's 100% overlap when you are running real drivers22:58
*** leitan has quit IRC22:59
johnsomSo, yeah, leave it as scenario IMO22:59
*** catintheroof has quit IRC23:01
johnsomSo, it looks like things are back to zuulv2 and gates are working again23:03
*** sshank has quit IRC23:06
*** SumitNaiksatam has quit IRC23:06
kongjohnsom, rm_work, sorry was distracted by something else. I see your points23:07
rm_workyeah so i think with flavor support, we can have a noop flavor and do "api" tests like that23:10
rm_workbut until then, they're all going to be scenario unless we make the whole test run be noop23:10
rm_workbut, again, we already do noop api tests in the main tree23:10
johnsomYeah, and frankly, the scenarios should cover the API right?23:10
rm_workyes23:22
rm_workkong: ah i see, i think i had been looking at an old patchset or something... i see the member clients there now23:25
rm_workso i will see about rebasing23:25
rm_workkong: what happened to the waiters stuff? like, how do I wait for a LB to change status now?23:37
rm_workah nm I see it, you moved it into the main thing and named it await_loadbalancer_active23:38
*** tongl has quit IRC23:41
rm_workbrb23:44
« Monday, 2017-10-02 Index Wednesday, 2017-10-04 »

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