Monday, 2014-02-10

*** xuhanp has joined #openstack-neutron00:00
*** yamahata has quit IRC00:04
*** WackoRobie has quit IRC00:07
*** WackoRobie has joined #openstack-neutron00:07
*** ihrachyshka has quit IRC00:08
*** xuhanp has quit IRC00:25
*** markmcclain has joined #openstack-neutron00:38
*** sputnik13net has joined #openstack-neutron00:41
*** markmcclain has quit IRC00:43
*** yamahata has joined #openstack-neutron00:54
*** yongli has joined #openstack-neutron00:57
*** sp has joined #openstack-neutron00:59
*** mwagner_lap has joined #openstack-neutron01:00
*** alagalah has quit IRC01:01
*** idella4 has joined #openstack-neutron01:13
*** dims has joined #openstack-neutron01:21
*** dkehn__ has quit IRC01:24
*** dkehn__ has joined #openstack-neutron01:25
*** idella4 has left #openstack-neutron01:27
*** iwamoto has joined #openstack-neutron01:28
*** yamahata has quit IRC01:33
*** ijw has quit IRC01:33
*** WackoRobie has quit IRC01:37
*** yamahata has joined #openstack-neutron01:48
*** WackoRobie has joined #openstack-neutron01:50
*** RajeshMohan has quit IRC01:51
*** RajeshMohan has joined #openstack-neutron01:53
*** yamahata has quit IRC01:53
*** morganfainberg_Z is now known as morganfainberg01:56
*** WackoRob_ has joined #openstack-neutron01:57
*** WackoRobie has quit IRC01:57
*** networkstatic has joined #openstack-neutron02:08
*** xianghui has joined #openstack-neutron02:12
*** yongli has quit IRC02:31
*** yamahata has joined #openstack-neutron02:36
*** dkehn__ is now known as dkehn_02:45
*** ijw has joined #openstack-neutron03:00
*** morganfainberg is now known as morganfainberg_Z03:19
*** morganfainberg_Z is now known as morganfainberg03:20
*** sputnik13net has quit IRC03:27
*** harlowja is now known as harlowja_away03:30
*** suresh12 has joined #openstack-neutron03:31
*** markmcclain has joined #openstack-neutron03:31
openstackgerritXu Han Peng proposed a change to openstack/neutron: Permit ICMPv6 RAs only from known routers  https://review.openstack.org/7225203:35
*** banix has quit IRC03:40
*** alagalah has joined #openstack-neutron03:41
*** irenab has joined #openstack-neutron03:46
irenabLooking for core team member for question regarding patch approval03:54
*** sp has quit IRC03:58
*** WackoRob_ has quit IRC04:02
openstackgerritChengli Xu proposed a change to openstack/python-neutronclient: str2dict support commas within quotes  https://review.openstack.org/6509704:05
*** dguitarbite has joined #openstack-neutron04:06
*** baoli has quit IRC04:06
*** MM_at_HP has quit IRC04:07
*** MM_at_HP has joined #openstack-neutron04:07
markmcclainirenab: I'm around.. which patch?04:10
irenabmarkmcclain: https://review.openstack.org/#/c/53609/04:15
irenabThanks!04:15
* markmcclain looking04:16
markmcclainirenab: I'm going to run a re-check since the last one was almost two week ago.. it is passes, I'll approve04:17
markmcclains/it is/if it/04:18
irenabmarkmcclain: Thanks a lot04:18
*** oda-g has quit IRC04:22
*** oda-g has joined #openstack-neutron04:24
*** gongysh has joined #openstack-neutron04:25
openstackgerritA change was merged to openstack/neutron: Imported Translations from Transifex  https://review.openstack.org/7176104:26
openstackgerritmark mcclain proposed a change to openstack/neutron: Delete DHCP port without DHCP server on a network node  https://review.openstack.org/5674004:29
openstackgerritmark mcclain proposed a change to openstack/neutron: Delete DHCP port without DHCP server on a net node  https://review.openstack.org/5674004:31
*** WackoRobie has joined #openstack-neutron04:35
*** irenab has quit IRC04:38
openstackgerritA change was merged to openstack/neutron: Fix typo in rootwrap files: neuton -> neutron  https://review.openstack.org/7213404:40
*** WackoRobie has quit IRC04:41
*** yfried has joined #openstack-neutron04:56
*** alagalah has quit IRC05:13
*** chandankumar_ has joined #openstack-neutron05:17
*** irenab has joined #openstack-neutron05:20
*** aymenfrikha has joined #openstack-neutron05:34
*** gdubreui has quit IRC05:35
*** gdubreui has joined #openstack-neutron05:35
*** WackoRobie has joined #openstack-neutron05:38
*** dguitarbite has quit IRC05:42
*** markwash has joined #openstack-neutron05:44
*** rotbeard has joined #openstack-neutron05:44
*** WackoRobie has quit IRC05:46
*** markwash has quit IRC05:48
*** markwash has joined #openstack-neutron05:49
*** dguitarbite has joined #openstack-neutron05:49
openstackgerritA change was merged to openstack/neutron: Return request-id in API response  https://review.openstack.org/5827005:53
openstackgerritBerezovsky Irena proposed a change to openstack/neutron: Change tenant network type usage for IB Fabric  https://review.openstack.org/6718905:55
*** vmtrooper has quit IRC05:59
*** vmtrooper has joined #openstack-neutron06:00
*** amotoki has joined #openstack-neutron06:05
*** aymenfrikha has quit IRC06:06
*** WackoRobie has joined #openstack-neutron06:13
*** markmcclain has quit IRC06:13
*** markmcclain has joined #openstack-neutron06:16
*** WackoRobie has quit IRC06:18
*** markmcclain1 has joined #openstack-neutron06:18
*** markmcclain has quit IRC06:21
openstackgerritJenkins proposed a change to openstack/neutron: Imported Translations from Transifex  https://review.openstack.org/7227106:24
*** pbeskow has quit IRC06:36
*** gdubreui has quit IRC06:37
*** pbeskow has joined #openstack-neutron06:37
openstackgerritHyunsun Moon proposed a change to openstack/neutron: Fix binding:host_id is set to None when port update  https://review.openstack.org/6282906:41
*** markmcclain1 has quit IRC06:42
*** WackoRobie has joined #openstack-neutron06:43
*** ihrachyshka has joined #openstack-neutron06:47
*** WackoRobie has quit IRC06:48
openstackgerritIsaku Yamahata proposed a change to openstack/neutron: mailmap: update .mailmap  https://review.openstack.org/7227406:49
*** saju_m has joined #openstack-neutron06:53
*** saju_m has quit IRC07:00
openstackgerritIsaku Yamahata proposed a change to openstack/neutron: tests/service: consolidate setUp/tearDown logic  https://review.openstack.org/7066807:03
*** ajo has joined #openstack-neutron07:04
*** chandankumar_ has quit IRC07:07
*** chandan_kumar has joined #openstack-neutron07:07
openstackgerritBerezovsky Irena proposed a change to openstack/neutron: Add update from agent to plugin on device up  https://review.openstack.org/5360907:08
*** uaberme has joined #openstack-neutron07:08
*** akamyshnikova has joined #openstack-neutron07:12
*** WackoRobie has joined #openstack-neutron07:14
*** saju_m has joined #openstack-neutron07:16
*** igor has joined #openstack-neutron07:16
*** dims has quit IRC07:18
*** dims has joined #openstack-neutron07:19
*** WackoRobie has quit IRC07:20
*** igor has quit IRC07:20
*** igor has joined #openstack-neutron07:21
doudegongysh: hi07:21
gongyshdoude: hi07:21
*** ijw has quit IRC07:21
doudegongysh: Do you have time to have a look of my draft https://review.openstack.org/#/c/68694/ ?07:21
doudegongysh: this is not urgent, it's just to know07:23
gongyshdoude: why do u need to do this?07:24
gongyshdoude: all the dhcp notification is in https://review.openstack.org/#/c/68694/2/neutron/api/v2/base.py07:25
doudegongysh: I propose that in order thaht DHCP agent is aware of l3 port delete, update or create07:25
gongyshdoude: why do u delete them there?07:25
doudegongysh: why the DHCP notification are in neutron/api/v2/base.py ?07:25
gongyshdoude: because this is a common feature for all plugins.07:26
doudewhen a service like l3 , lbaas, fwaas... make a change on a ressource, the agent DHCP doesn't have any notifications07:26
doudegongysh: all plugin implements inevitably DHCP agent?07:27
gongyshdoude:  almost.07:27
amotokidoude: gongysh: i don't think DHCP notification is a common feature, although most plugins/drivers use dhcp-agent now.07:27
doudegongysh: ok07:28
gongyshamotoki:  this is decided when we implemented the dhcp agent notification.07:28
doudegongysh: but I propose that in order the DHCP agent know l3 port modification and activate or desactivate metadata proxy service07:28
doudeactually it starts inevitably a metadata proxy07:29
doudeevent if a gw port is configured on  the network07:29
amotokigongysh: i am okay that it is placed in api.v2.base at now, but I believe it is not a requirement.07:30
doudegongysh, amotoki : and we lost all ressource modifications notifications when they are created by the plugin (an not thourgh the API)07:31
gongyshdoude: previoiusly, we thought only the compute's port should be dhcp supported.07:32
doudegongysh: ha ok, I understand07:33
gongyshdoude: why do u want the l3 ports be dhcp supported?07:33
doudegongysh: it's not to be DHCP supported but DHCP agent need to know if l3 port status on the network to know if it needs to activate or not (update) metadata proxy service07:34
doudegongysh: I like to propose this improveme07:35
gongyshdoude: then you can send an ad-hoc dhcp notification.07:37
doudegongysh: I like to propose this improvement because in our production we saw that metadata proxy consume lot of memory and actually a metadata proxy was started by the l3 agent and DHCP agent at the same time for a same network07:38
gongyshdoude: why not disable the dhcp agent started metadata proxy?07:39
doudeand lot of memory are consume uselessly on the network and decrease the Neutron network node scalability07:39
doudegongysh: in the case of the network is isolated, the DHCP agent should provide the metadata proxy07:40
openstackgerritAkihiro Motoki proposed a change to openstack/neutron: Raise an error from ovs_lib list operations  https://review.openstack.org/7228207:41
openstackgerritXu Han Peng proposed a change to openstack/neutron: Permit ICMPv6 RAs only from known routers  https://review.openstack.org/7225207:41
gongyshdoude: yes07:43
doudegongysh: it's a proposal and just want to know your opinion07:44
openstackgerritOfer Barkai proposed a change to openstack/neutron: Enhance floating IP router lookup  https://review.openstack.org/5598707:45
*** WackoRobie has joined #openstack-neutron07:46
*** ihrachyshka has quit IRC07:46
*** mflobo has joined #openstack-neutron07:46
*** HenryG has quit IRC07:47
*** ihrachyshka has joined #openstack-neutron07:47
gongyshdoude: what is your plan?  you want to shutdown the dhcp started proxy when the network is changed from isolated into router connected?07:47
*** HenryG has joined #openstack-neutron07:48
doudegongysh: yes and remove DHCP static route from the lease07:48
gongyshdoude: sound good.  but don't change the overall dhcp notification framework, which will be a big change.07:50
*** WackoRobie has quit IRC07:51
*** ihrachyshka has quit IRC07:51
doudegongysh: but I require to change DHCP notification framework to notify l3 port update07:52
gongyshdoude: no need, you just need to send an ad-hoc notification to dhcp agent if the port is l3 port.07:53
doudegongysh: ok, but do you think we could have a new service (XaaS) that needs to be dhcp supported?07:55
*** sweston_ has joined #openstack-neutron07:57
*** evgenyf has joined #openstack-neutron07:58
*** sweston has quit IRC07:58
*** salv-orlando has joined #openstack-neutron07:59
*** markwash has quit IRC07:59
*** ijw has joined #openstack-neutron08:00
*** ekarlso has quit IRC08:04
*** ekarlso has joined #openstack-neutron08:05
*** ekarlso has quit IRC08:12
openstackgerritIsaku Yamahata proposed a change to openstack/neutron: tests/service: consolidate setUp/tearDown logic  https://review.openstack.org/7066808:12
*** ekarlso has joined #openstack-neutron08:13
*** pasquier-s has joined #openstack-neutron08:14
*** WackoRobie has joined #openstack-neutron08:17
*** joshh has quit IRC08:21
*** morganfainberg has quit IRC08:22
*** joshh has joined #openstack-neutron08:23
*** WackoRobie has quit IRC08:23
*** morganfainberg has joined #openstack-neutron08:23
*** ihrachyshka has joined #openstack-neutron08:25
openstackgerritOleg Bondarev proposed a change to openstack/neutron: LBaaS: move agent based driver files into a separate dir  https://review.openstack.org/6133908:26
*** jlibosva has joined #openstack-neutron08:32
openstackgerritIWAMOTO Toshihiro proposed a change to openstack/neutron: LBaaS: implements LVS driver (WORKINPROGRESS)  https://review.openstack.org/6740508:37
*** iwamoto has quit IRC08:38
*** suresh12 has quit IRC08:41
ihrachysamotoki: could you please check the following simple fix for neutronclient: https://review.openstack.org/#/c/71551/08:42
ihrachysamotoki: sorry to reach directly, but I need this to be checked by core devs asap to proceed with another performance fix for neutron08:43
amotokiihrachys: sure. looking now.08:45
ajohmmm08:49
*** jistr has joined #openstack-neutron08:49
ajoI'm seeing failures after only changing the commit message here: https://review.openstack.org/#/c/68601/  , it's a backport to havana08:49
*** ygbo has joined #openstack-neutron08:50
*** feleouet has joined #openstack-neutron08:50
ajo SSHTimeout: Connection to the 172.24.4.228 via SSH timed out.08:50
*** WackoRobie has joined #openstack-neutron08:50
*** ijw has quit IRC08:51
*** WackoRobie has quit IRC08:54
*** matrohon has joined #openstack-neutron08:58
openstackgerritEdgar Magana proposed a change to openstack/neutron: Implements provider network support in PLUMgrid plugin  https://review.openstack.org/6038308:59
*** safchain has joined #openstack-neutron09:00
openstackgerritliusheng proposed a change to openstack/python-neutronclient: Remove vi modelines  https://review.openstack.org/7229709:02
*** jpich has joined #openstack-neutron09:04
openstackgerritA change was merged to openstack/neutron: Fix misspellings in neutron  https://review.openstack.org/7178109:06
openstackgerritA change was merged to openstack/neutron: Disallow non-admin users update net's shared attribute  https://review.openstack.org/6909509:08
*** jlibosva has quit IRC09:14
*** jlibosva has joined #openstack-neutron09:17
ihrachysamotoki: thanks for +2! any other core dev for neutronclient willing to add another +2 to obvious bug fix with great impact on neutron scalability? :) https://review.openstack.org/#/c/71551/09:20
openstackgerritOleg Bondarev proposed a change to openstack/neutron: Introduce Loadbalancer instance  https://review.openstack.org/6020709:20
openstackgerritOleg Bondarev proposed a change to openstack/neutron: LBaaS: apply agent scheduling for loadbalancer instance  https://review.openstack.org/7147209:20
*** WackoRobie has joined #openstack-neutron09:21
*** rotbeard has quit IRC09:25
*** WackoRobie has quit IRC09:25
*** rotbeard has joined #openstack-neutron09:28
*** morganfainberg is now known as morganfainberg_Z09:28
*** mjbot has joined #openstack-neutron09:28
*** afazekas has joined #openstack-neutron09:29
*** rotbeard has quit IRC09:32
openstackgerritJun Jie Nan proposed a change to openstack/neutron: Check vxlan enablement via modinfo  https://review.openstack.org/6590209:35
*** rotbeard has joined #openstack-neutron09:37
*** jlibosva has quit IRC09:39
*** bvandenh has joined #openstack-neutron09:39
openstackgerritA change was merged to openstack/neutron: add quota support for ryu plugin  https://review.openstack.org/6524509:40
openstackgerritA change was merged to openstack/neutron: ML2 plugin cannot raise NoResultFound exception  https://review.openstack.org/7209609:44
*** pradipta` has joined #openstack-neutron09:50
*** WackoRobie has joined #openstack-neutron09:52
*** pradipta has quit IRC09:54
*** dguitarbite has quit IRC09:55
*** WackoRobie has quit IRC09:56
*** jlibosva has joined #openstack-neutron09:56
*** networkstatic has quit IRC09:57
*** dguitarbite has joined #openstack-neutron10:00
*** jp_at_hp has joined #openstack-neutron10:01
*** ihrachyshka has quit IRC10:08
openstackgerritOleg Bondarev proposed a change to openstack/neutron: LBaaS: apply agent scheduling for loadbalancer instance  https://review.openstack.org/7147210:10
openstackgerritXiang Hui proposed a change to openstack/neutron: Remove legacy quantum config path  https://review.openstack.org/6884610:13
*** gongysh has quit IRC10:20
*** networkstatic has joined #openstack-neutron10:20
*** WackoRobie has joined #openstack-neutron10:22
*** WackoRobie has quit IRC10:27
EmilienMenikanorov: hi Eugene, I've abandoned https://review.openstack.org/#/c/63240/ - you can see why in the last comment10:29
*** xianghui has quit IRC10:31
*** RajeshMohan has quit IRC10:33
*** RajeshMohan has joined #openstack-neutron10:34
*** amotoki has quit IRC10:38
*** yamahata has quit IRC10:41
*** djoreilly has joined #openstack-neutron10:51
*** irenab has quit IRC11:08
*** safchain has quit IRC11:11
*** safchain has joined #openstack-neutron11:12
enikanorovEmilienM: ok, but i see Nach's patch is abandoned as well11:17
enikanorov*Nachi's11:17
openstackgerritChristoph Thiel proposed a change to openstack/python-neutronclient: Allow allocating specific floating ip for admins  https://review.openstack.org/2356211:22
*** WackoRobie has joined #openstack-neutron11:24
*** pcm_ has joined #openstack-neutron11:26
*** pcm_ has quit IRC11:29
*** pcm_ has joined #openstack-neutron11:30
*** dguitarbite has quit IRC11:30
*** sweston_ has quit IRC11:31
*** WackoRobie has quit IRC11:32
akamyshnikovaanteaya, Hello! I'm writing to you because you did review of my change https://review.openstack.org/55411 some weeks ago. At that time there was problems with passing tests. Now everything works correct. Please, visit my change https://review.openstack.org/55411 one more time and remove -1 if nothing else confuses you.11:38
*** RajeshMohan has quit IRC11:40
*** RajeshMohan has joined #openstack-neutron11:41
openstackgerritSalvatore Orlando proposed a change to openstack/neutron: DB Mappings for NSX security groups  https://review.openstack.org/7169211:44
openstackgerritSalvatore Orlando proposed a change to openstack/neutron: NSX plugin: Introduce operational status for security groups  https://review.openstack.org/7169311:44
openstackgerritChristoph Thiel proposed a change to openstack/python-neutronclient: Allow allocating specific floating ip for admins  https://review.openstack.org/2356211:50
openstackgerritSylvain Afchain proposed a change to openstack/neutron: Fix import NeutronManager  https://review.openstack.org/6888711:51
*** pradipta` is now known as pradipta11:55
openstackgerritA change was merged to openstack/neutron: Fix binding:host_id is set to None when port update  https://review.openstack.org/6282911:56
*** HenryG has quit IRC11:58
*** WackoRobie has joined #openstack-neutron11:59
*** jroovers has joined #openstack-neutron12:01
*** joshh has quit IRC12:01
jroovershi yfried12:03
*** WackoRobie has quit IRC12:03
*** cburgess has quit IRC12:04
*** cburgess has joined #openstack-neutron12:11
*** joshh has joined #openstack-neutron12:13
*** irenab has joined #openstack-neutron12:14
*** HenryG has joined #openstack-neutron12:14
HenryGenikanorov: ping12:15
enikanorovHenryG: pong12:19
HenryGenikanorov: Hi, you gave a -1 here and rcurran answered your question, can you check?  https://review.openstack.org/#/c/56478/412:21
enikanorovHenryG: sure, let me look12:22
HenryGenikanorov: thanks!12:22
*** yamahata has joined #openstack-neutron12:22
yfriedjroovers: looking for me?12:22
jrooversyep, how are you, it's been a while since I've been here?12:23
yfriedI'm ok. how can I help you?12:23
jrooversyfried, https://review.openstack.org/#/c/66854/12:23
jrooversthis patch set is currently abandoned, but that is not my question12:24
jrooverscan you have a look at Masayuki's comment12:24
yfriedjroovers: what about it?12:25
jrooversyfried, well, I was wondering whether you have any insight into this.12:25
jrooversis there a guideline around scenario/functional tests and neutron moving forward?12:26
enikanorovHenryG: ok, so I'm still abit confused about this part of code. it calls ..get_bininG() and expects bindingS12:27
enikanorovand checks for amount of bindings12:27
enikanorovHenryG: do you know what is expected at line 107?12:28
HenryGenikanorov: checking ...12:29
yfriedjroovers: I understand his point - the strict definition of scenario tests is at least 2 components working together12:30
jrooversyfried, I agree12:30
jrooversyfried, I do think that a tests like the one this patchset provides can be very relevant.12:30
jrooversyfried, so wondering where it would belong12:30
yfriedjroovers: I'm not seeing any actual tests in the patch - just moving network related scenario to a new folder12:31
jrooverssorry (working on too much stuff at the same time), I based a different one on it: https://review.openstack.org/#/c/6551512:32
yfriedjroovers: However, in practice, I agree with your approach - there can be a lot of network targeted tests that conform to that definition but actually test only neutron functionality12:32
*** IanGovett has joined #openstack-neutron12:33
yfriedjroovers: so to what comment are you referring?12:34
jrooversyfried, so I based this VPN scenario test on the directory restructuring12:35
jrooversso 65515 is based on 6685412:35
jrooversas I wanted to bring some more structure12:35
HenryGenikanorov: I agree get_..._binding() should be plural (and the comment in the method fixed)12:36
jrooversI asked sdague about this, and he agreed12:36
jrooversyfried, but I also agree Masayuki's point12:36
jrooversby create a separate network directory, it does seems as if the tests are only focussed on the network. While this is not really so. In the case of the VPN scenario tests, we do spin up vm's in the networks (=nova).12:37
yfriedjroovers: tell you what - push the test independently, as it's a good idea, then, once you have 3 network tests, file a bug and get support for the new dir12:38
*** akamyshnikova has quit IRC12:38
jrooversyfried, sounds good, thanks for the guidance :-)12:38
HenryGenikanorov: however, changing that would ripple out to 6 places where it is called. I think rcurran may want to avoid that for his bug.12:39
yfrieda directory for 2 tests is not really that strong.12:39
yfriedjroovers: however, you should add LBaaS and FWaaS to such directory12:39
enikanorovsure, no prob, the question is about L107, is that 'else' expects more than 1 binding?12:39
enikanorovor no bindings?12:39
yfriedjroovers: because, disregarding the strict definition for a moment, once we get too many scenario modules we will need some order in that directory12:40
jrooversyfried, yes, I agree :-)12:41
HenryGenikanorov: ah, ok. Are you prepared to learn about the quirks of configuring nexus TOR switches?12:41
yfriedjroovers: maybe you'll get other ideas on how to classify scenario tests, aside from service based12:41
enikanorovHenryG: nope :) I'm just trying to check correspondence between client code (the patch) and get_bindings12:42
yfriedjroovers: also - I don't really have time for review (I'm on leave for exams) but if you are using methods from other modules (such as connectivity tests), try to get the code to manager.py12:43
yfriedjroovers: IMO - copy-paste makes for bad coding, and difficult maintenance12:43
jrooversyfried, I agree. Sorry, this code is one of my first contributions, I wanted to keep the patch rather simple.12:45
jrooversyfried, but I will follow up on that, thanks!12:45
*** banix has joined #openstack-neutron12:45
yfriedjroovers: I understand the reluctance, but you will be thanked for it later.12:45
openstackgerritSascha Peilicke proposed a change to openstack/neutron: Support passing 'neutron_insecure' to neutronclient  https://review.openstack.org/6569612:47
yfriedjroovers: good job so far - keep it on12:48
jrooversyfried, thanks :-)12:48
*** banix has quit IRC12:50
*** bada has joined #openstack-neutron12:51
openstackgerritAnn Kamyshnikova proposed a change to openstack/neutron: Sync models with migrations  https://review.openstack.org/5541112:54
*** baoli has joined #openstack-neutron12:55
HenryGenikanorov: The method _configure_switch_entry() is called at post_commit, so bindings cannot be 0.12:58
HenryGenikanorov: does that answer your question?12:58
enikanorovHenryG: yes. thanks!12:59
enikanorovHenryG: i've removed -113:00
openstackgerritBerezovsky Irena proposed a change to openstack/neutron: Add support to request vnic type on port  https://review.openstack.org/7233413:01
openstackgerritChristoph Thiel proposed a change to openstack/python-neutronclient: Allow allocating specific floating ip for admins  https://review.openstack.org/2356213:02
ajohttp://logs.openstack.org/01/68601/2/check/gate-neutron-python26/69a7a57/console.html#_2014-02-10_10_31_41_31113:03
ajowow13:03
ajohas anybody seen a failure like this before?13:03
ajoit failed with a timeout on testools/subunit itself for py2613:03
EmilienMobondarev: ping13:03
EmilienMenikanorov: does LBaaS need a public IP address ? or does it use L3 agent to manage public VIPs?13:05
HenryGenikanorov: great! thanks!13:05
enikanorovEmilienM: haproxy vips are on tenant networks so they require floating ip to be associated to be reachable13:06
EmilienMenikanorov: but if i install the lbaas agent on another node than l3 agent, do I need public network access?13:07
enikanorovthat doesn't matter where lbaas & l3 agents are13:08
enikanorovl3 agent will NAT from external network to a vip on tenant network13:08
EmilienMenikanorov: i just wonder if lbaas server can be in internal network13:08
EmilienMso yes, it seems to be, thanks :-)13:08
enikanorovit is13:08
EmilienMenikanorov: thx :()13:08
*** saju_m has quit IRC13:12
*** ilyashakhat has quit IRC13:14
*** dave_tucker is now known as dave_tucker_zzz13:17
anteayaakamyshnikova: my -1 disappeared when you submitted patchset 13, I hope it passes check tests for you13:17
openstackgerritA change was merged to openstack/neutron: Support building wheels (PEP-427)  https://review.openstack.org/5712413:20
ajohttps://bugs.launchpad.net/neutron/+bug/1278430 I submited this bug, it seems that unit test fails when sending a possitive test result...13:20
ajomay be the testing systems are a little bit too stressed13:21
ajoin fact, it seems that most gate-nova-python26 at test jobs13:23
ajoremain in "queued" in the gate13:23
openstackgerritenikanorov proposed a change to openstack/neutron: Introduce Loadbalancer instance  https://review.openstack.org/6020713:25
ajoanteaya ^^13:25
ajobtw, what does mean that a test is marked as "LOST" ? :? :)13:26
*** WackoRobie has joined #openstack-neutron13:30
*** alagalah has joined #openstack-neutron13:31
*** ilyashakhat has joined #openstack-neutron13:32
anteayaajo: so you are asking me what it means if the logs from a test are marked as logged, is that correct?13:33
ajono13:33
openstackgerritAnn Kamyshnikova proposed a change to openstack/neutron: Sync models with migrations  https://review.openstack.org/5541113:33
*** jgallard has joined #openstack-neutron13:33
ajoIm asking about the bug I submitted13:33
ajoIt seems that we're running into timeouts for py26 tests,13:33
ajonormal passing tests13:34
anteayado you have the url for a patch that shows this?13:34
ajosure, wait, it's a cherry pick backport13:34
ajohttps://review.openstack.org/#/c/68601/13:35
ajoIf I didn't understand the traceback wrong, it seems that the unit test passed, but when unittest tried to report the success of this test (via it's internal mechanism... it ended in a timeout)13:37
*** tongli has joined #openstack-neutron13:38
*** baoli_ has joined #openstack-neutron13:38
*** baoli has quit IRC13:40
*** dave_tucker_zzz is now known as dave_tucker13:46
*** dims has quit IRC13:47
openstackgerritSylvain Afchain proposed a change to openstack/neutron: Fix Metering doesn't respect the l3 agent binding  https://review.openstack.org/6001613:48
anteayaajo infra is having random failure issues this morning, it appears, I am trying to understand why13:48
ajooh, ok, thanks anteaya , If they're are aware, then it will be ok13:49
anteayawe are trying to understand and compare notes regarding what we are aware of13:50
anteayaright now we seem to be aware of different things13:50
anteayatrying to share and sort out what is relevant13:50
*** baoli_ is now known as baoli13:50
*** dims has joined #openstack-neutron13:50
*** yamahata has quit IRC13:56
*** safchain has quit IRC13:58
*** yamahata has joined #openstack-neutron13:59
*** safchain has joined #openstack-neutron14:00
openstackgerritAvishay Balderman proposed a change to openstack/neutron: LBaaS L7 model (WIP)  https://review.openstack.org/6172114:00
*** xuhanp has joined #openstack-neutron14:01
*** xuhanp has quit IRC14:25
*** rossella_s has joined #openstack-neutron14:25
*** jecarey has joined #openstack-neutron14:28
*** jecarey_ has joined #openstack-neutron14:29
*** jistr has quit IRC14:29
openstackgerritBerezovsky Irena proposed a change to openstack/neutron: Add support to request vnic type on port  https://review.openstack.org/7233414:34
*** ramishra has joined #openstack-neutron14:35
*** jistr has joined #openstack-neutron14:39
*** obondarev_ has joined #openstack-neutron14:41
*** peristeri has joined #openstack-neutron14:42
*** matsuhashi has joined #openstack-neutron14:49
openstackgerritJakub Libosvar proposed a change to openstack/neutron: Metadata agent caches networks for routers  https://review.openstack.org/7068914:54
*** ramishra has quit IRC14:54
*** ramishra has joined #openstack-neutron14:55
*** yamahata has quit IRC14:59
*** ramishra_ has joined #openstack-neutron15:00
ihrachyslet's assume we get a fix in neutronclient that unlocks review for another fix to neutron. so to be able to pass jenkins checks, we probably need a new neutronclient version to depend on for neutron (I guess we can't depend on nightly releases). Does it mean that the sequence of steps to get neutron fix that depends on neutronclient fix is as follows: 1. release new neutronclient with the fix; 2.15:00
*** banix has joined #openstack-neutron15:00
ihrachysupdate global requirements for neutronclient to enforce new release version; 3. update requirements in neutron based on new global requirements; 4. finally, send the fix for neutron and get it merged.15:00
ihrachysdoes anyone follow me? :)15:01
*** ramishra has quit IRC15:01
*** AramakiHiroki has joined #openstack-neutron15:01
*** irenab has quit IRC15:01
*** saju_m has joined #openstack-neutron15:01
*** yamahata has joined #openstack-neutron15:03
*** saju_m has quit IRC15:03
*** saju_m has joined #openstack-neutron15:04
*** clev has joined #openstack-neutron15:07
ajoihrachys, I follow you :)15:07
*** markwash has joined #openstack-neutron15:07
*** otherwiseguy has joined #openstack-neutron15:08
*** sdague has quit IRC15:12
*** jgrimm has joined #openstack-neutron15:16
*** rwsu has joined #openstack-neutron15:17
*** chandan_kumar has quit IRC15:20
*** sdague has joined #openstack-neutron15:29
*** AramakiHiroki has quit IRC15:31
*** AramakiHiroki has joined #openstack-neutron15:32
*** markwash has quit IRC15:32
*** sweston has joined #openstack-neutron15:33
*** armax has joined #openstack-neutron15:34
*** AramakiHiroki has quit IRC15:36
*** matsuhashi has quit IRC15:37
*** yamahata has quit IRC15:37
*** matsuhashi has joined #openstack-neutron15:37
*** saju_m has quit IRC15:37
*** sweston has quit IRC15:38
*** saju_m has joined #openstack-neutron15:39
*** dims has quit IRC15:41
*** markwash has joined #openstack-neutron15:42
*** dims has joined #openstack-neutron15:45
*** markwash has quit IRC15:46
*** afazekas has quit IRC15:47
*** rotbeard has quit IRC15:58
*** AlexF has joined #openstack-neutron15:59
*** carl_baldwin has joined #openstack-neutron15:59
*** yfried has quit IRC15:59
*** MM_at_HP has quit IRC16:01
*** AlexF has quit IRC16:01
otherwiseguyanteaya: do you know the status of the gate for the havana branches? Didn't know if I should try to recheck the polling minimization patches yet or not.16:02
*** dave_tucker is now known as dave_tucker_zzz16:05
*** aymenfrikha has joined #openstack-neutron16:06
*** matsuhashi has quit IRC16:07
*** AlexF has joined #openstack-neutron16:08
*** sweston has joined #openstack-neutron16:09
anteayaotherwiseguy: the status of the gate right now for everything is poor16:09
anteayanodepool was eating up swap this morning and we don't know why16:09
*** sweston is now known as sweston_16:09
otherwiseguyanteaya: so a day like any other! :)16:09
anteayaso the host rebooted the vm nodepool was on16:09
anteayawhich puts us in a bad spot16:09
otherwiseguyindeed.16:09
anteayathe weekend was good16:10
*** sbalukoff has quit IRC16:12
*** terrylhowe has joined #openstack-neutron16:12
*** networkstatic has joined #openstack-neutron16:17
*** dave_tucker_zzz is now known as dave_tucker16:17
*** thedodd has joined #openstack-neutron16:19
*** bvandenh has quit IRC16:21
openstackgerritEdgar Magana proposed a change to openstack/neutron: Implements provider network support in PLUMgrid plugin  https://review.openstack.org/6038316:21
openstackgerritJakub Libosvar proposed a change to openstack/neutron: Metadata agent caches networks for routers  https://review.openstack.org/7068916:22
*** ijw has joined #openstack-neutron16:23
anteayaotherwiseguy: jobs are running again though we do have a back log16:24
anteayaas for status of the gate for just havana branches, I'm not sure what to look for to evaluate that16:24
anteayalet me give it some thought16:24
*** ramishra_ has quit IRC16:25
*** _gk_ has joined #openstack-neutron16:27
openstackgerritEdgar Magana proposed a change to openstack/neutron: Implements provider network support in PLUMgrid plugin  https://review.openstack.org/6038316:28
*** ramishra has joined #openstack-neutron16:29
*** arosen has joined #openstack-neutron16:32
*** jistr has quit IRC16:33
*** jistr has joined #openstack-neutron16:34
anteayaotherwiseguy: just looked at a bunch of links16:37
anteayastarted here: https://review.openstack.org/#/q/status:merged+project:openstack/neutron+branch:stable/havana,n,z16:37
anteayaa patch merged to stable/havana yesterday16:37
*** mlavalle has joined #openstack-neutron16:37
anteayathen looking at job builds: https://jenkins03.openstack.org/job/gate-tempest-dsvm-neutron/ seems to be building fine16:38
anteayahttps://jenkins01.openstack.org/job/gate-tempest-dsvm-neutron-pg/ seems good16:38
*** thuc has joined #openstack-neutron16:38
anteayaditto for: https://jenkins04.openstack.org/job/gate-tempest-dsvm-neutron-isolated/16:39
*** thuc has quit IRC16:39
anteayalast successful build was 4 days ago for https://jenkins05.openstack.org/job/gate-tempest-dsvm-neutron-pg-isolated/16:39
otherwiseguyanteaya: ok. thanks. I know that mark was looking at something related to gate stability with havana last week. maybe things got resolved? I just know that the polling minimization patches seemed to take a couple of tries to get through a while ago.16:39
anteayaworking on a build now16:39
*** thuc has joined #openstack-neutron16:39
anteayaI haven't talked to mark in a few days16:40
anteayabest get an update from him16:40
otherwiseguyanteaya: ok. thanks.16:40
anteayanp16:40
*** terrylhowe_ has joined #openstack-neutron16:40
otherwiseguyjust wanted to see if we could get the patches in in time for build. :/16:41
*** terrylhowe_ has left #openstack-neutron16:41
*** jroovers has quit IRC16:42
anteayayes a worthy goal16:42
*** markmcclain has joined #openstack-neutron16:43
anteayaspeak of the devil16:43
*** jgallard has quit IRC16:46
*** markmcclain1 has joined #openstack-neutron16:46
* markmcclain1 reads scrollback16:47
*** igor has quit IRC16:47
*** AlexF has quit IRC16:47
*** markmcclain has quit IRC16:47
*** igor has joined #openstack-neutron16:47
*** evgenyf has quit IRC16:50
openstackgerritKyle Mestery proposed a change to openstack/neutron: Add OpenDaylight ML2 MechanismDriver  https://review.openstack.org/6977516:50
*** igor has quit IRC16:52
*** _gk_ has quit IRC16:55
markmcclain1so stable branch still is seeing several failures16:58
*** saju_m has quit IRC16:59
anteayamarkmcclain1: are any of those failures trackable for those wanting to follow along at home?17:00
markmcclain1we still have to dig through the latest batch of failures17:00
markmcclain1and figure out if they're neutron items or caused by something else17:00
*** _gk_ has joined #openstack-neutron17:01
*** dave_tucker is now known as dave_tucker_zzz17:01
markmcclain1we're still not consistent +1s from the checks17:03
markmcclain1https://review.openstack.org/#/q/status:open+project:openstack/neutron+branch:stable/havana,n,z17:03
*** _gk_ has quit IRC17:03
*** baoli has quit IRC17:04
*** saju_m has joined #openstack-neutron17:04
*** dguitarbite has joined #openstack-neutron17:04
*** yamahata has joined #openstack-neutron17:04
*** dguitarbite has left #openstack-neutron17:04
*** baoli has joined #openstack-neutron17:04
*** markwash has joined #openstack-neutron17:05
*** yfried has joined #openstack-neutron17:06
openstackgerritJakub Libosvar proposed a change to openstack/neutron: Metadata agent caches networks for routers  https://review.openstack.org/7068917:06
anteayamarkmcclain1: marun was telling me last week about some of the havana check jobs being noop17:07
anteayaI'm not sure if I fully understood his point though17:07
anteayathere is a large chance I didn't17:07
markmcclain1yeah they're def not a no-op if we're seeing -1s17:07
anteayathat was what didn't make sense to me17:08
anteayamy brain is still mid-foggy though so I haven't gone back to try to understand it17:08
*** yamahata has quit IRC17:08
anteayaokay, I'll leave it for now17:08
*** dave_tucker_zzz is now known as dave_tucker17:09
*** yamahata has joined #openstack-neutron17:09
*** markmcclain1 has quit IRC17:09
*** networkstatic has quit IRC17:09
*** markmcclain has joined #openstack-neutron17:10
*** jlibosva has quit IRC17:11
openstackgerritTerry Wilson proposed a change to openstack/neutron: Add multiple RPC workers  https://review.openstack.org/7242517:12
*** rossella_s has quit IRC17:13
*** dave_tucker is now known as dave_tucker_zzz17:16
*** spandhe has joined #openstack-neutron17:16
spandheHi salv-orlando yt?17:17
*** itzikb has quit IRC17:17
*** jroovers has joined #openstack-neutron17:17
salv-orlandoih spandhe but I need to reboot my laptop in 10 minutes as it's acting crazy17:17
openstackgerritPaul Ward proposed a change to openstack/neutron: Expose ML2 mechanism driver exceptions to the user  https://review.openstack.org/6974817:18
spandhesalv-orlando: ok.. I have a question regarding http://lists.openstack.org/pipermail/openstack-dev/2013-June/010805.html.. we can talk once your laptop problem is resolved..17:19
*** _gk_ has joined #openstack-neutron17:19
*** saju_m has quit IRC17:19
*** yfried has quit IRC17:21
*** dave_tucker_zzz is now known as dave_tucker17:21
openstackgerritEdgar Magana proposed a change to openstack/neutron: Implements provider network support in PLUMgrid plugin  https://review.openstack.org/6038317:23
*** bada_ has joined #openstack-neutron17:24
*** bada has quit IRC17:25
*** sweston_ has quit IRC17:25
openstackgerritPaul Ward proposed a change to openstack/neutron: Expose ML2 mechanism driver exceptions to the user  https://review.openstack.org/6974817:25
*** dave_tucker is now known as dave_tucker_zzz17:26
*** jpich has quit IRC17:26
*** jroovers has quit IRC17:27
*** salv-orlando has quit IRC17:27
openstackgerritSean M. Collins proposed a change to openstack/neutron: Developer documentation  https://review.openstack.org/7242817:29
*** salv-orlando has joined #openstack-neutron17:30
*** sn6i23a has joined #openstack-neutron17:31
*** sn6i23a has quit IRC17:31
*** yfried has joined #openstack-neutron17:31
*** sn6i23a has joined #openstack-neutron17:32
*** saju_m has joined #openstack-neutron17:32
salv-orlandospandhe: I resurrected17:32
spandhesalv-orlando: great! :)17:33
salv-orlandoso… about your question?17:33
spandheIn my setup. I faced the same issue as mentioned in this thread: http://lists.openstack.org/pipermail/openstack-dev/2013-June/010799.html17:33
spandheI used the workaround as suggested, added sql_dbpool_enable = True in the ini file.17:34
*** jistr has quit IRC17:34
spandhesalv-orlando:  What does that setting exactly achieve?17:34
spandheand after enabling that setting, that issue got resolved, but some other problems started to appear..17:35
*** jistr has joined #openstack-neutron17:35
salv-orlandospandhe: That setting enables db connection pooling in the sqlalchemy backend. Depending on your plugin a few things might be sorted, but other might screw up17:35
salv-orlandowhat plugin are you using? If using ml2, what drivers are you using?17:36
spandhei am using nvp plugin17:36
spandhethe other issue that we observed after adding the setting was: if the system was kept idle for about 5 minutes, quantum lost connectivity to the database. OperationalError: (2006, 'MySQL server has gone away')17:38
spandhesalv-orlando: ^17:38
salv-orlandoWe've never seen the latter issue; and no, don't use db pooling with the nvp plugin.17:39
salv-orlandoBecause it will trigger out of order operations on the NVP backend17:39
*** AlexF has joined #openstack-neutron17:40
spandhesalv-orlando: ok.. then whats the solution for allowing concurrent create_ports?17:40
salv-orlandoThe NVP (now NSX) plugin is being progressively improved to support concurrent operations as well asynchronous backend17:40
salv-orlandoone common way however of deploying it so far is to replicate the server process and put it behind a load balancer17:41
*** dave_tucker_zzz is now known as dave_tucker17:42
*** markmcclain has quit IRC17:42
spandhesalv-orlando: thanks a lot! Is that the only solution? I will look into it.17:42
salv-orlandoup to havana yes17:42
salv-orlandoWe've been trying db pooling internally, but created a lot of problems with the dhcp agent17:43
carl_baldwinotherwiseguy: ping17:43
otherwiseguycarl_baldwin: pong17:43
salv-orlandospandhe: I'm not sure if that the same problem you're seeing, but we'll be happy to improve concurrency also for pre-icehouse releases17:44
carl_baldwinI just noticed your review to add multiple process rpc.  I just started hacking one out myself last week.17:44
carl_baldwinI did not post anything for review yet, though.17:45
otherwiseguycarl_baldwin: Mine is really simple, I've tested it and it appears to work, but there may be something missing that I didn't think of.17:45
spandhesalv-orlando: we are not using dhcp, as far as I know.. and the "SQL gone" error happens on any simple command such as "nova list".. but thats if the system is idle for about 5 minutes.17:46
otherwiseguyIt's 10 lines of code, so  ...17:46
otherwiseguyI was trying to just get up to speed on how it all fit together for most of the time.17:46
*** ygbo has quit IRC17:46
otherwiseguyIs your implementation appreciably different?17:47
carl_baldwinotherwiseguy: I'll put some comments in to the review.  Mainly, I was struggling with the fact that each plugin starts RPC as part of its own init.17:47
carl_baldwinotherwiseguy: I have not finished looking over yours.  I just didn't want to wait to get in touch with you on it.17:47
otherwiseguycarl_baldwin: Yeah. I basically just swoop in and set up the RPC workers first. But because of how the plugin is instantiated, unlike the api workers patch, the parent process will end up with an rpc worker as well with my patch.17:48
*** _gk_ has quit IRC17:48
otherwiseguycarl_baldwin: blueprint spec here: https://docs.google.com/document/d/1LYgoARbgFrdMNte1ki1aC-pXRk_r0ZsqnEop9BedM10/edit#17:49
carl_baldwinMine followed the pattern of the multiple api workers patch.  There is a little more complexity in it.  I was taking care of a few more of those loose ends.17:49
*** jistr has quit IRC17:50
openstackgerritSean M. Collins proposed a change to openstack/neutron: Developer documentation  https://review.openstack.org/7242817:50
*** safchain has quit IRC17:50
otherwiseguyI figured it didn't really matter if the RPC worker ran in the main process or not. Thought about just passing in rpc_workers - 1 to the ProcessLauncher stuff so the number came out right. Figured it would be good to just post something up for discussion.17:51
carl_baldwinotherwiseguy: I'll have a more thorough look at yours.  I think I can help you fix some of those loose ends.17:51
*** dave_tucker is now known as dave_tucker_zzz17:51
otherwiseguycarl_baldwin: Cool, thanks. I was pretty sure there had to be several lurking in there to tie up. :)17:52
*** alagalah has quit IRC17:52
salv-orlandospandhe: does the sql gone away error occur only with sql_db_pool enabled?17:53
salv-orlandoon the other hand, another big difference is: are you running grizzly or havana17:53
carl_baldwinotherwiseguy: I can put what I have up as a draft for you to look at for comparison.  I'll find some time to clean it up and put it up today.17:53
spandhesalv-orlando: yes.17:55
salv-orlandospandhe: thanks. Grizzly or havana?17:55
*** bjornar has joined #openstack-neutron17:55
otherwiseguycarl_baldwin: thanks!17:56
carl_baldwinotherwiseguy: thanks, I look forward to working with you.17:56
spandhesalv-orlando: Grizzly..17:56
otherwiseguycarl_baldwin: same here. I'm sure I will reap the most benefit out of that. ;)17:57
*** igor_ has joined #openstack-neutron17:58
*** harlowja_away is now known as harlowja17:58
salv-orlandospandhe: cool. So what might be causing your mysql gone away problem is a pool exhaustion with no connection recycling.17:59
salv-orlandoThis is actually caused by a nasty behaviour on the neutron authZ checks which trigger a DB operation for each item in a list operation17:59
salv-orlandowhen you get to some scale (say 1,000 ports)17:59
salv-orlandoand you run commands such as nova-list, you start having problems.18:00
spandhesalv-orlando: my scale is actually way low.. 5 ports or so..18:00
salv-orlandoThis issue was fixed in havana, but the fix was too large for being considered for backport18:00
salv-orlandospandhe: well, that's interesting. You should hardly exhaust the pool with so little connections.18:01
*** jroovers has joined #openstack-neutron18:01
*** carl_baldwin has quit IRC18:01
spandhesalv-orlando: I have default pool size,, I think the default is 5..18:01
*** igor_ has quit IRC18:02
salv-orlandospandhe: yup but then again, sqlalchemy also adds overflow connections in the pool. Also, as you witness this only when the neutron server has been idle, this makes the problem even stranger.18:02
*** carl_baldwin has joined #openstack-neutron18:04
salv-orlandoI think there was some history around the db pool mechanisms used in Grizzly. Basically we ended up moving to something else for Havana, as stated in the discussion thread you were reading.18:04
*** markmcclain has joined #openstack-neutron18:04
*** clev has quit IRC18:04
spandhesalv-orlando: yes.. should I try increasing sqlalchemy_pool_size?18:04
salv-orlandospandhe: that helps but does not solve the issue.18:04
spandheOr you think the issue probably wont be seen in Havana?18:04
salv-orlandospandhe: in my opinion this was a problem with Grizzly. But I am unable to confirm it now.18:05
*** ijw has quit IRC18:05
spandhesalv-orlando: ok.. thanks!18:07
salv-orlandospandhe: np, I'm here if you need anything else.18:08
*** ijw has joined #openstack-neutron18:10
*** saju_m has quit IRC18:12
spandhesalv-orlando: but in general, you wouldnt recommend using dbpool for NSX plugin, right?18:13
marunanteaya: I'm not sure I claimed the havana jobs were noop.18:14
salv-orlandospandhe: In general not. If you are hitting performance problems without it using grizzly18:14
salv-orlandospandhe: ping me and we'll see how that can be sorted together18:14
spandhesalv-orlando: sure.. thanks!18:15
*** jp_at_hp has quit IRC18:16
*** suresh12 has joined #openstack-neutron18:26
markmcclainbeagles, carl_baldwin, sc68cal, mlavalle, mestery, rkukura: don't forget to update the agenda18:28
rkukuraok18:28
*** saju_m has joined #openstack-neutron18:29
carl_baldwinok18:29
*** zzelle has joined #openstack-neutron18:30
beaglesack18:30
*** alexpilotti has joined #openstack-neutron18:31
*** sweston has joined #openstack-neutron18:35
*** rwsu has quit IRC18:36
openstackgerritTerry Wilson proposed a change to openstack/neutron: Remove psutil dependency  https://review.openstack.org/6899318:40
mlavallemarkmcclain: I'll do it over the next 90 minutes or so18:41
mlavalle:-)18:41
markmcclainrkukura, carl_baldwin, beagles, mlavalle: thanks18:41
*** ijw_ has joined #openstack-neutron18:41
*** Sukhdev has joined #openstack-neutron18:43
*** ijw has quit IRC18:44
*** zigo has quit IRC18:45
mesterymarkmcclain: Ack on that18:46
*** haleyb has quit IRC18:46
*** zigo has joined #openstack-neutron18:49
*** rwsu has joined #openstack-neutron18:51
*** ijw_ has quit IRC18:51
*** saju_m has quit IRC18:51
*** saju_m has joined #openstack-neutron18:52
*** morganfainberg_Z is now known as morganfainberg18:53
*** sweston has quit IRC18:56
openstackgerritKevin Benton proposed a change to openstack/neutron: BigSwitch: Fix tenant_id for shared net requests  https://review.openstack.org/7245018:57
*** sbalukoff has joined #openstack-neutron18:57
*** saju_m has quit IRC18:58
*** otherwiseguy has quit IRC18:59
*** markmcclain has quit IRC19:01
*** markmcclain has joined #openstack-neutron19:02
*** haleyb has joined #openstack-neutron19:02
*** clev has joined #openstack-neutron19:03
*** ijw has joined #openstack-neutron19:03
*** AlexF has quit IRC19:04
*** carl_baldwin has quit IRC19:11
*** ramishra has quit IRC19:14
openstackgerritRobert Kukura proposed a change to openstack/neutron: Replace binding:capabilities with binding:vif_details  https://review.openstack.org/7245219:14
*** ramishra has joined #openstack-neutron19:14
*** beagles is now known as beagles_brb19:26
openstackgerritLuis A. Garcia proposed a change to openstack/neutron: Re-enable lazy translation  https://review.openstack.org/6697719:28
openstackgerritLuis A. Garcia proposed a change to openstack/neutron: Do not append to messages with +  https://review.openstack.org/6698019:28
*** djoreilly has quit IRC19:30
*** thedodd has quit IRC19:30
*** markmcclain has quit IRC19:38
*** saju_m has joined #openstack-neutron19:38
*** sweston has joined #openstack-neutron19:40
*** ijw has quit IRC19:41
*** godara has joined #openstack-neutron19:41
*** ijw has joined #openstack-neutron19:41
*** otherwiseguy has joined #openstack-neutron19:42
*** _gk_ has joined #openstack-neutron19:45
*** ijw has quit IRC19:46
*** thuc has quit IRC19:46
*** ramishra has quit IRC19:46
*** thuc has joined #openstack-neutron19:46
anteayamarun: it is entirely possible I mis-understood19:47
*** zigo has quit IRC19:48
*** _gk_ has quit IRC19:50
*** thuc has quit IRC19:50
openstackgerritArmando Migliaccio proposed a change to openstack/neutron: Add migration support from agent to NSX dhcp/metadata services  https://review.openstack.org/5132519:54
*** markmcclain has joined #openstack-neutron19:58
*** saju_m has quit IRC20:01
*** openstack has joined #openstack-neutron21:33
-dickson.freenode.net- [freenode-info] channel trolls and no channel staff around to help? please check with freenode support: http://freenode.net/faq.shtml#gettinghelp21:33
openstackgerritA change was merged to openstack/neutron: Improve handling of security group updates  https://review.openstack.org/6310021:38
*** alagalah has quit IRC21:38
*** ramishra has joined #openstack-neutron21:48
*** baoli has quit IRC21:51
*** ramishra has quit IRC21:52
openstackgerritOleg Bondarev proposed a change to openstack/neutron: LBaaS: apply agent scheduling for loadbalancer instance  https://review.openstack.org/7147221:53
*** aymenfrikha1 has joined #openstack-neutron21:55
*** aymenfrikha has quit IRC21:55
*** RajeshMohan has quit IRC22:02
*** carl_baldwin has quit IRC22:02
enikanorovmarkmcclain: hey, got some time to discuss lb instance?22:02
markmcclainright now.. I've got to return to meetings22:03
markmcclainI figured we can discuss on Thursday22:03
enikanorovmarkmcclain: ok, i remember you've sent me an email on that22:03
marunarosen: ping22:03
enikanorovmarkmcclain: and i've replied with quite detailed explanation. if you could look at it, we could discuss based on that22:04
beaglessweston: ping22:04
*** RajeshMohan has joined #openstack-neutron22:04
markmcclainok22:04
swestonbeagles: pong22:04
enikanorovthanks22:04
*** carl_baldwin has joined #openstack-neutron22:04
swestonbeagles: I have to prepare for another meeting in an hour, what is your availability later today, or tomorrow?22:05
beaglessweston: so the things I was referring to in the meeting are referred to here https://wiki.openstack.org/wiki/NovaNeutronGapHighlights22:05
yamahataenikanorov, ping22:05
enikanorovyamahata: pong22:05
yamahataenikanorov, can we chat a bit on servicevm?22:05
arosenmarun:  pong22:05
enikanorovyamahata: yeah, but i'll be halp asleep :)22:05
enikanorov*half22:05
marunarosen: I have a question regarding nova/neutron interaction...22:06
yamahataenikanorov, sorry. I'll be quick.22:06
*** jecarey_ has quit IRC22:06
marunarosen: I noticed recently that nova network does synchronous network provisioning.22:06
*** jecarey_ has joined #openstack-neutron22:06
yamahataenikanorov,  per your review, you're claiming to change both REST API and data base model?22:06
swestonbeagles:  ok, I will take a look at it, but don't have time now :-(.  May we chat later?22:07
beaglessweston, I'm pretty much done for today here (TZ is -330 GMT). Tuesdays are pretty rough sometimes.. I've back to back meetings from 9:00 till 12:30 and then another at 14:30 (my TZ) I'll ping you iin there somewhere22:07
yamahataenikanorov, hosting_device -> device, logical_service_instance -> service_instance22:07
enikanorovyamahata:  are you talking about 'servicetypes'?22:07
marunarosen: I found myself wondering if nova shouldn't be checking in with neutron to ensure active port status before booting a vm22:07
enikanorovyamahata: i think some of the names may be shortened, expecially the attribute names22:07
*** ywu has joined #openstack-neutron22:07
enikanorov*especially22:08
marunarosen: to maximize the chances of cloud-init/dhcp having connectivity to succeed22:08
yamahataenikanorov, Ok. I'll rename it. And servicetypes.22:08
*** ramishra has joined #openstack-neutron22:08
enikanorovso do you think storing service types as a list of strings makes sense?22:08
yamahataenikanorov, yes.22:09
marunarosen: this could avoid the problem of instance boot racing with network configuration when neutron is used22:09
enikanorovok, good that we agree on that22:09
swestonbeagles: ok, I will be available.  that will give me some time to read over what you want in the document and think about it before tomorrow.22:09
marunarosen: what do you think?22:09
yamahataenikanorov, thanks.22:09
*** WackoRobie has quit IRC22:09
*** salv-orlando_ has joined #openstack-neutron22:10
arosenmarun:  it sounds logical though i think it would probably make things more complex. The cloud-init stuff does a ton of retry so I'm not sure if it would really help22:11
openstackgerritSylvain Afchain proposed a change to openstack/neutron: Add L3 VRRP HA base classes  https://review.openstack.org/6455322:11
marunarosen: we have persistent problems with dhcp, though22:11
marunarosen: and consider the case of containers22:11
marunarosen: they will always win a race with neutron22:11
marunarosen: I would suggest that the lack of nova coordinating directly with neutron is why we see so many non-deterministic failures.22:12
arosenmarun:  i like the idea. It would be interesting to see if it would help. What do we do if the port never becomes active?22:12
marunarosen: if nova were checking with neutron, we would at least have a clear indication of failure.22:12
*** _gk_ has quit IRC22:13
*** ramishra has quit IRC22:13
marunarosen: there would have to be a timeout22:13
arosenthen i guess put the vm in error state22:13
marunarosen: and then vm boot could be failed22:13
arosenyea, i think it makes sense.22:13
marunarosen: yes, exactly22:14
*** salv-orlando has quit IRC22:14
*** salv-orlando_ is now known as salv-orlando22:14
*** _gk_ has joined #openstack-neutron22:14
marunarosen: the port being active won't guarantee success, of course, since it provides no guarantees of dhcp being configured22:15
marunarosen: but I think it's a good starting point.22:15
*** _gk_ has quit IRC22:15
marunarosen: Do you think it would be feasible to submit something by next week?  I'm happy to do the work if you don't have time, though my ignorance of nova will mean I will take longer.22:16
marunarosen: I don't think it should be the default, but I think it would be valuable to get it out there for people to test with this cycle.22:17
lifelessmarun: dhcp problems ?22:19
marunlifeless: :(22:19
arosenmarun:  give me a few sec i want to check where the call is that puts the external_id in ovsdb22:19
marunlifeless: regarding eventual consistency I presume?22:20
arosendepending on where that is this sounds pretty straightforward to add.22:20
lifelessmarun: you said above we have lots of dhcp problems22:20
lifelessmarun: I'm wondering how many are really just https://bugs.launchpad.net/neutron/+bug/127134422:20
marunlifeless: ah, sorry.22:20
marunlifeless: i think that's a different problem22:20
lifelessmarun: I've seen two DHCP problems recently. That one, and the br-int ofctl flows being bong for some unknown reason22:21
lifelesswhich is fixed by neutron-openvswitch-agent restarts.22:21
marunlifeless: better nova/neutron coordination is meant to solve the problem of l2 and dhcp agent racing nova boot and losing when a system is overloaded22:21
lifelessmarun: oh right, sure, I know the one.22:22
*** jlibosva has quit IRC22:23
*** peristeri has quit IRC22:25
enikanorovrkukura: hi, a quick question, is it possible to use gre/vxlan networks and vlan networks simultaneously with ml2 plugin?22:29
openstackgerritA change was merged to openstack/neutron: Fix race condition in network scheduling to dhcp agent  https://review.openstack.org/7139322:29
*** tongli has quit IRC22:34
rkukuraenikanorov: yes22:34
*** arosen has quit IRC22:35
rkukuraenikanorov: Combining tunnel-based tenant networks with vlan and flat provider networks is a common approach22:35
*** arosen has joined #openstack-neutron22:35
rkukuraWith ml2, you can also list multiple tenant_network_types, so you could have a set of available VLAN tenant networks, and if these are exhausted, allocate GRE or VXLAN tenant networks instead.22:36
rkukuraAlso, if you change tenant_network_types (i.e. to switch from GRE to VXLAN), existing tunnel networks of the old type should continue to work.22:37
*** bjornar has quit IRC22:39
*** salv-orlando_ has joined #openstack-neutron22:39
arosenlifeless:  i think i  know why you are hitting that dhcp issue22:40
arosenI commented on the launchpad issue.22:40
*** gdubreui has joined #openstack-neutron22:42
lifelessarosen: dhcp_release is meant to be called ?22:42
arosenlifeless:  it's getting called.22:42
arosenI think the packet is getting blocked.22:43
lifelessarosen: the packet from the neutron node to the dnsmasq on the neutron node?22:43
arosenThe gate was running for like a month without security groups working which is why it was working.22:43
arosenlifeless:  i believe so. Should be sure in a sec.22:43
*** salv-orlando has quit IRC22:43
*** salv-orlando_ is now known as salv-orlando22:43
lifelessarosen: I'm clearly confused. I didn't realise the netns for dhcp would have security groups applied to it.22:44
lifelessarosen: since its not a VM22:44
arosenlifeless: there are iptable rules there though saying a vm can only receive traffic from other vms and dhcp_offers22:44
arosenI think the release might be getting blocked there.22:45
lifelessarosen: but those rules are on an entirely different machine22:45
lifelesshttp://paste.ubuntu.com/6911812/22:45
arosenthe l2-agent runs on the same machine as the dhcp-agent though which is what applies those iptable rules.22:45
lifelessiptables-save from the undercloud machine where clint just saw this22:45
lifelessarosen: Except it has no vifs to apply them too22:47
*** armax has quit IRC22:47
lifelessarosen: see the pastebins[6~22:47
lifelessoh, ip netns exec qdhcp-e85b44c7-1136-4217-954e-cdf0acdddfe1 iptables-save | pastebinit22:47
lifelesshttp://paste.ubuntu.com/6911818/22:47
lifelessis the second one22:47
lifelessfor the netns that the dhcp_release call will be made from22:47
*** clev has quit IRC22:47
arosenyup22:48
*** jecarey_ has quit IRC22:50
zzellehi, a small question, why ip veth are used to connect br-int to physnets for vlans and ovs patchs for tunnels ?22:51
openstackgerritArmando Migliaccio proposed a change to openstack/neutron: Add migration support from agent to NSX dhcp/metadata services  https://review.openstack.org/5132522:51
lifelessarosen: I just ran it by hand getting22:51
lifelessFeb 10 22:51:10 undercloud-notcompute-jws3awlsb2kh dnsmasq-dhcp[26029]: DHCPRELEASE(tap56efc4c6-55) 10.10.16.141 78:e7:d1:23:97:cd22:51
lifelessin syslog22:51
arosenlifeless:  I ran it by hand and it didn't seem to fix the issue for me.22:52
arosenDid it fix it for you?22:52
lifelessarosen: we have restarted dnsmasq already so I can't confirm or deny22:53
lifelessarosen: but it logged that it saw it22:53
lifelessarosen: aee though that the earlier releases say 'lease unknown' which is ... interesting22:53
*** Sukhdev has quit IRC22:54
*** armax has joined #openstack-neutron22:54
lifelessarosen: btw that server is running 7918ff71079239ffab9c16a6c65be3a876cbbdb022:55
arosenlifeless: hrm  I was able to reproduce this a few min ago but now i'm having trouble getting this to occur again22:55
lifelessarosen: because I had to monkey that on22:55
lifelessarosen: first mainline commit is 44649bd7c9bbf5ce27ffc4162980afb02ae5bdf5 which is feb 6th, so quite recent22:55
*** zzelle has quit IRC22:57
openstackgerritKevin Benton proposed a change to openstack/neutron: BigSwitch: Fix tenant_id for shared net requests  https://review.openstack.org/7245022:57
*** nati_uen_ has quit IRC23:01
*** sweston has quit IRC23:02
*** packet has quit IRC23:02
*** banix has quit IRC23:04
*** alagalah has joined #openstack-neutron23:05
*** ramishra has joined #openstack-neutron23:09
*** dave_tucker_zzz is now known as dave_tucker23:10
*** jecarey has joined #openstack-neutron23:11
openstackgerritArmando Migliaccio proposed a change to openstack/neutron: Rename L2 Switch/Gateway related methods for VMware NSX plugin  https://review.openstack.org/6735323:11
openstackgerritAmir Sadoughi proposed a change to openstack/neutron: Assign local VLANs before port filters  https://review.openstack.org/6555723:12
*** thedodd has quit IRC23:13
*** ramishra has quit IRC23:13
*** alexpilotti has quit IRC23:13
*** nati_ueno has joined #openstack-neutron23:17
*** jgrimm has quit IRC23:19
*** AramakiHiroki has joined #openstack-neutron23:19
*** nati_ueno has quit IRC23:23
arosenmarun:  One change that will need to be made for what you proposed is there is a call to:             self.driver.plug_vifs(instance, net_info) in compute/manager.py that puts the neutron port-id in ovsdb that the l2-agent looks for needs to be moved into neutronv2/api.py i think23:28
marunarosen: I'm not sure I follow23:28
*** armax has quit IRC23:29
arosenso the l2-agent looks in ovsdb for the neutron port-id23:29
marunarosen: right23:29
arosenthat doesn't get inserted into ovsdb until after we've exited out of the neutronv2/api.py code.23:30
*** ramishra has joined #openstack-neutron23:30
arosenI was thinking that the sleep/retry logic on port status would be done in the neutronv2/api.py code. What were you thinking?23:30
*** rms_13 has joined #openstack-neutron23:30
marunarosen: It sounds like a chicken/egg issue though.23:31
arosenI think the best place to do the sleep/retry logic would be in that file. I was just saying we'd need to move that line so that we insert the id in ovsdb  before we start polling.23:32
marunarosen: How can we insert that id until vif plug?23:32
marunarosen: I mean, how do we even know which host to target until vif plug?23:32
marunarosen: My thought was the sync point has to be post-vif plug.23:32
arosenThe vif.plug() is run on the compute host.23:32
arosenmarun:  k, that works too. Though we'd need to add a call to neutronv2/api.py to do the polling somewhere in the compute manager.23:33
marunarosen: yeah, that's where I'm a bit fuzzy.23:33
arosenI think that's kind of odd to do because nova-network doesn't need this call that's why we should just sleep retry in allocate_for_instance in neutronv2/api.py23:33
marunarosen: well, nova network does thing synchronously though23:34
*** mtreinish has quit IRC23:34
marunarosen: with neutron, nova calls to provision port (trigger async provisioning0, then schedules and boots vm23:35
arosenmarun:  what do you mean?23:35
*** jroovers has quit IRC23:35
arosenwhy do you say that's synchronous?23:35
marunarosen: I mean, nova network performs network provisioning via a call(), and does not proceed on error23:35
marunarosen: I should dig that up23:35
*** mtreinish has joined #openstack-neutron23:39
marunarosen: While I'm looking, I'm still confused as to where else you could poll for port status other than after vif plug23:39
marunarosen: what am I missing?23:40
arosenmarun:  right I think we'd need to either add a call to the compute manager to neutronv2/api.py to do the polling23:41
arosenor i think move the vif.plug() call23:41
*** mlavalle has quit IRC23:46
marunarosen: where would you move the vif plug call to?23:47
marunarosen: as to nova network being synchronous...23:47
marunhttps://github.com/openstack/nova/blob/master/nova/compute/manager.py#L104623:48
marunThe underlying call is 'async' only in that it doesn't block until access to the result is attempted23:48
marunBut this access will be performed at some point during the booting of an instance23:48
marunso there is essentially a sync point that happens before instance boot that ensures that, say, dhcp will be available for the instance's ip23:49
*** ijw_ has joined #openstack-neutron23:50
*** markwash has quit IRC23:51
marunarosen: As to the options you suggest, I understand the need to add a call to the neutron api to poll for port status23:53
*** markmcclain has quit IRC23:53
*** ijw has quit IRC23:53
marunarosen: I'm still unclear on how to move the vif plug call.23:53
marunarosen: I apologize if I'm missing something obvious.23:54
*** AramakiHiroki has quit IRC23:54
*** arosen has quit IRC23:54
*** sweston has joined #openstack-neutron23:57

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