Thursday, 2014-08-14

*** markmcclain has joined #openstack-lbaas00:01
*** markmcclain1 has quit IRC00:02
*** fnaval has quit IRC00:10
*** sbfox has quit IRC00:14
*** crc32 has quit IRC00:23
openstackgerritStephen Balukoff proposed a change to stackforge/octavia: Octavia v0.5 component design  https://review.openstack.org/11345800:24
*** sbalukoff has quit IRC00:28
*** rm_you|wtf is now known as rm_you00:44
*** Krast has quit IRC01:17
*** Krast has joined #openstack-lbaas01:17
*** markmcclain1 has joined #openstack-lbaas01:29
*** markmcclain has quit IRC01:30
*** magesh has quit IRC01:32
*** markmcclain1 has quit IRC01:36
*** s3wong has quit IRC01:38
*** mlavalle has quit IRC01:52
*** VijayB_ has quit IRC02:16
*** fnaval has joined #openstack-lbaas02:32
*** fnaval has quit IRC02:45
*** sbfox has joined #openstack-lbaas02:57
*** fnaval has joined #openstack-lbaas03:14
*** vjay has joined #openstack-lbaas03:24
*** vjay has quit IRC03:46
*** enikanorov__ has joined #openstack-lbaas03:47
*** vjay has joined #openstack-lbaas03:47
*** woodster__ has joined #openstack-lbaas03:49
*** johnsom_ has joined #openstack-lbaas03:49
*** woodster_ has quit IRC03:52
*** enikanorov has quit IRC03:52
*** clev has quit IRC03:52
*** johnsom has quit IRC03:52
*** woodster__ is now known as woodster_03:53
*** clev has joined #openstack-lbaas03:53
*** vjay has quit IRC03:57
*** fnaval has quit IRC04:06
*** fnaval has joined #openstack-lbaas04:12
*** fnaval has quit IRC04:38
*** yfried has quit IRC04:41
*** fnaval has joined #openstack-lbaas04:43
*** fnaval has quit IRC05:03
*** amotoki has joined #openstack-lbaas05:38
*** yfried has joined #openstack-lbaas05:40
*** orion_ has joined #openstack-lbaas05:59
*** orion_ has quit IRC06:04
*** evgenyf has joined #openstack-lbaas06:38
*** jschwarz has joined #openstack-lbaas07:04
*** sbalukoff has joined #openstack-lbaas07:22
*** woodster_ has quit IRC07:25
*** sballe has joined #openstack-lbaas07:39
*** sbfox has quit IRC07:46
*** sballe_ has joined #openstack-lbaas09:04
*** sballe has quit IRC09:05
*** sballe_ has quit IRC09:09
*** sballe has joined #openstack-lbaas09:18
*** sballe has quit IRC09:24
*** Krast has quit IRC11:31
*** amotoki has quit IRC11:48
*** yfried has quit IRC12:05
*** yfried has joined #openstack-lbaas12:06
*** yfried_ has joined #openstack-lbaas12:07
*** yfried has quit IRC12:11
jschwarzHey guys, I just read the LBaaS API and Object Model improvement blueprint and had a question,12:13
jschwarzaround line 196 and line 215, it is said that the 'subnet_id' field is being moved from the Pool DB object to the Member DB object12:14
jschwarzThis implies that the 2 different members which are connected to the same pool can be in different subnets, and specifically from different subnets than the Listener12:15
jschwarzHow is it planned to conserve connectivity between them?12:15
*** yfried__ has joined #openstack-lbaas12:25
*** yfried_ has quit IRC12:28
*** jorgem has joined #openstack-lbaas13:16
*** woodster_ has joined #openstack-lbaas13:22
*** sballe has joined #openstack-lbaas13:24
*** fnaval has joined #openstack-lbaas13:25
*** woodster_ is now known as woodster13:29
*** sballe has quit IRC13:33
*** ptoohill has joined #openstack-lbaas13:44
*** orion__ has joined #openstack-lbaas13:49
*** sballe has joined #openstack-lbaas13:54
*** vjay has joined #openstack-lbaas13:58
dougwigjschwarz: if a subnet id is specified in members, i believe the longer-term plan is to create neutron ports where necessary to connect it all together.13:59
*** xgerman has joined #openstack-lbaas13:59
bloganor l3 connectivity would work14:00
*** tmc3inphilly has joined #openstack-lbaas14:04
*** tmc3inphilly has quit IRC14:04
*** tmc3inphilly has joined #openstack-lbaas14:04
*** tmc3inphilly has joined #openstack-lbaas14:05
jschwarzfuture? so currently not planned for upcoming version?14:07
*** crc32 has joined #openstack-lbaas14:14
bloganjschwarz: any driver is free to create the ports necessary to get connectivity14:14
*** crc32 has quit IRC14:14
bloganjschwarz: so drivers should do that14:14
jschwarzquestion was, is it currently implemented/up for review, or is it something that will come in for the K release?14:15
jschwarzalso, dougwig, nice job with Kevin benton there ;-)14:16
dougwig:)14:17
bloganjschwarz: no its not currently implemented, but it will be in the ref implemenation by K14:17
jschwarzblogan, thanks :)14:17
bloganjschwarz: i'm also debating whether the plugin should do the port creation automatically14:17
jschwarzblogan, it's not only ports, but also being aware of the different namespaces and routes throughout both nets14:19
jschwarzAlso if 2 members are sitting behind different compute nodes it also complicates it a bit more14:19
jschwarzI believe it should be in the service plugin side, but something doesn't feel quite right about it14:20
*** crc32 has joined #openstack-lbaas14:20
jschwarz... why would the lbaas plugin create ports, change routes? Isn't that the L3 plugin's job? Also all the data is already at the L3 plugin14:21
jschwarzdid you consider an IPC between them of some sort? Maybe issue requests from the LBaaS side to the L3 plugin using the notifier or something like that?14:23
jschwarzblogan, &14:23
jschwarz^14:23
*** markmcclain has joined #openstack-lbaas14:25
*** markmcclain has quit IRC14:25
*** markmcclain has joined #openstack-lbaas14:27
*** vivek-ebay has joined #openstack-lbaas14:27
jschwarzblogan, I got to go but will be back in an hour or so if you want to continue this! :)14:37
*** jschwarz has quit IRC14:38
bloganjschwarz: sorry in the lbaas meeting right now14:38
*** yfried__ has quit IRC14:43
*** fnaval has quit IRC14:50
*** fnaval has joined #openstack-lbaas14:50
*** fnaval has quit IRC14:55
bloganbbiab15:00
*** mestery is now known as mestery_afk15:01
*** tmc3inphilly has quit IRC15:01
rm_workactually, european style breakfast of meat and cheese is much better than our american cereal and shit, IMO :P15:01
*** HenryG_ has joined #openstack-lbaas15:01
*** jorgem1 has joined #openstack-lbaas15:01
sbalukoffrm_work: it's definitely tastier.15:02
*** jorgem1 has quit IRC15:02
sballerm_work, black coffee is the best breakfast15:02
*** jorgem1 has joined #openstack-lbaas15:02
rm_workT_T15:02
rm_workblack coffee on an empty stomach wrecks me a bit15:02
sbalukoffbitter disappointment makes for the worst breakfast. :P15:03
sbalukoffAah well.15:03
rm_workthen again, i don't often eat breakfast at all, so it's moot :P15:03
sballe:-)15:03
sbalukoffThat went about as I expected. :P15:03
sbalukoffUnfortunately.15:03
rm_workyeah, ah well15:03
sbalukoffI'm not usually this much of a pessimist.15:03
rm_workme either15:03
sballeAnywho... I think we should think about incubator --> separate incubator openstack pproject15:03
dougwigred bull and tums15:03
*** HenryG has quit IRC15:03
sbalukoffsballe: Yeah, there was discussion of this at yesterday's Octavia meeting too.15:04
sballeI do not believe we can do it without starting under openstack neutron.15:04
sbalukoffI'm looking through Trevor's notes from the meeting, will be helping to flesh those out, and then I expect they'll be sent to the mailing list.15:04
sballeas an incubator project under the networking umbrella15:04
dougwigi think we should sit in the incubator for K, while we focus on octavia, and then either spin out or neutron based on how that experiment has gone (most likely, spin out, for a variety of reasons, neutron politics not included.)15:04
rm_worksballe ++15:04
*** jorgem has quit IRC15:05
rm_workdougwig ++15:05
sbalukoffdougwig: I think that might be the only way forward.15:05
sballewe need to stay under the openstack umbrella. Just spining out into stakforge is not of interest for hP15:05
sballeso we'll have to deal with neutron politics for a little while longer and do it the rigth way15:06
rm_worksballe: no one is saying anything about stackforge15:06
*** HenryG_ has quit IRC15:06
sballerm_work, understood. Just mentioning in case somebody was thiking it15:06
rm_worki would assume we'd go from neutron-incubator to openstack-lbaas15:06
sballerm_work, agree. that is my assumption too15:07
sbalukoffsballe: I suspect that promises may have been implied in the last meeting regarding what it means to be part of "openstack" and "incubator" that may not actually be all that clear in the long run.15:07
rm_workthough i agree this stinks of "stackforge v2"15:07
*** HenryG has joined #openstack-lbaas15:07
dougwigrm_work: that is most likely, unless neutron undergoes some interesting changes in the next six months.15:07
sballedougwig, rm_work sbalukoff I agree with what your are saying. Let's focus on Octavia as part of the incubator project and make a decision based on how things are at the end of Kilo15:08
rm_workkk15:09
sbalukoffSpecifically, I think what Kyle and Mark were saying was meant more to placate and calm us, rather than them actually knowing what it means to be spun out right now.15:09
sbalukoffI can guess the amout of heat they're under right now in this GBP battle... and they don't want us against them, too.15:09
sbalukoffBut yes, all signs point toward concentrating on other areas until there are fewer mines in that particular war zone.15:10
sballeAgreed15:10
dougwigwe should not let politics distract us from the true value, which is load balancing.  what interface gets put on top, and where it lives, will not get the main useful code working.  (and there's just not enough data yet on what that battlefield will look like, so it's speculation.)15:10
dougwigi think that's close enough to call jinx.15:10
sbalukoffdougwig: +115:11
sballeWe migth not be the problem child but they cannot grandfather us without maybe doing hte same for the GBP15:11
sbalukoffsballe: From what I've read, GBP has a better case for being grandfathered in. So if they don't get in, there's no chance we will.15:11
sbalukoffI think their code was essentially ready at the last summit.15:12
dougwigi'm not sure it's apples/apples.  afaik, we don't have any cores actively opposed, we're just not really "done".15:12
sbalukoffOurs... not so much.15:12
*** evgenyf has quit IRC15:13
sbalukoffFor Octavia's part: I see all this being a sign that we should try to be less closely coupled with Neutron.15:13
sballedougwig, I agree with your statement earlier let's not speculate.15:13
sbalukoffHence the reason for the invention of that "networking driver" in the v0.5 component design spec.15:13
dougwigi think it's too early to make any changes; i think your current spec is fine, with perhaps a tad more emphasis on identifying problem neutron touch points in time to submit bp's to fix them in kilo.15:14
sbalukoffdougwig: Sure!15:14
*** vjay has quit IRC15:14
sballesbalukoff, Like I mentioned a few month ago we need an OpenStack LBaaS and not a random LBaaS. So while the integration with Neutron should be clean we shouldn;t re-invent Neutron15:14
*** fnaval_ has joined #openstack-lbaas15:15
sbalukoffsballe: I'm not sure I understood you at the time.15:15
sbalukoffAlso, I'm not sure what you mean by "re-invent Neutron"15:15
dougwigoff to get ready and head to the office.  bbiab.15:15
*** jorgem has joined #openstack-lbaas15:16
sballesbalukoff, I need to revisit the spec. At HP we only use floating IP to access the LB VM so we cannot do anythign with static IPs or virtual IP taht are not managed by Neutron15:16
sbalukoffsballe: Aah. Ok.15:16
*** jorgem1 has quit IRC15:17
sballesbalukoff, By "re-invent Neutron" I just meant we need to access Neutron functionality and not try to re-create it in "networking driver" but I belive we are in agreeemnt15:17
rm_worksballe: i think the "networking driver" is just an abstraction so we can hit Neutron *or* nova-networks, for instance… not rewrite our own networking subsystem :P15:18
sballerm_work, that is kind of what I thought but just wanted to make sure15:19
sbalukoffsballe: Right. Ideally the networking driver would just be a translation layer to Neutron API commands. But it's likely there will be things to do that aren't exposed via the neutron API, and hence we'll have to do some "dirty hack" stuff in the short term anyway (until API can be exposed)15:19
sballerm_work, Thanks for the clarification15:19
rm_workor we can have a rackspace-networks plugin fork of "neutron-network-plugin" and same for HP, to handle nuances in our floating-ip implementations, for example15:19
sbalukoffWhat that might entail is still not well defined yet.15:19
*** mestery_afk is now known as mestery15:19
sbalukoffrm_work: +115:20
sballerm_work, +1 too15:20
sbalukoffYes, it also allows a 3rd party SDN product to have its own networking driver, so Octavia could potentially live on top of that, too.15:20
rm_workyeah, i like it from a flexibility standpoint15:20
rm_workwe definitely won't be writing a full networking stack as a plugin though :P15:20
rm_workso no worries there15:21
sbalukoffAnd I like it from a "loosely coupled" standpoint.15:21
sbalukoffrm_work: +115:21
sballethe SDN plugin would plugin to Neutron as the ML2 driver so I am still asumming we would use the Neutron API15:25
sbalukoffsballe: That'll be the one we develop first.15:27
sballeperfect15:27
sbalukoffIf Octavia comes along nicely, I wouldn't be surprised to hear from, say, the OpenDaylight people wanting to develop their own networking driver.15:27
sbalukoffJust sayin'  ;)15:27
*** yfried__ has joined #openstack-lbaas15:30
*** dkehn_ has joined #openstack-lbaas15:30
*** fnaval has joined #openstack-lbaas15:32
*** dkehn_ is now known as dkehnx15:32
*** jorgem1 has joined #openstack-lbaas15:32
*** sballe_ has joined #openstack-lbaas15:33
*** sballe has quit IRC15:36
*** HenryG has quit IRC15:37
*** fnaval_ has quit IRC15:38
*** ptoohill has quit IRC15:38
*** jorgem has quit IRC15:40
*** ptoohill has joined #openstack-lbaas15:41
*** HenryG has joined #openstack-lbaas15:43
*** dkehnx has quit IRC15:46
rm_worksbalukoff: yeah, it definitely leaves us open to be flexible like that, which i think is good15:47
sbalukoffrm_work: And it actually makes dealing with upstream changes easier for us. Total win-win.15:49
*** jschwarz has joined #openstack-lbaas15:50
jschwarzblogan, ping15:52
bloganjscwarz: sorry was in a meeting earlier15:52
blogangot another soon as well15:52
jschwarzthanks alright15:52
jschwarzhow about an hour from now?15:52
*** VijayB has joined #openstack-lbaas15:53
blogangot lunch at that time15:53
bloganbut I do agree there are things that we need to solve to get the member subnets in a good state15:54
bloganoriginally we did say that if the subnet_id is not provided by the member hten its either already accessible by routing or by a port that already exists15:55
bloganthat doesn't solve for the compute node though that already has a port created15:55
bloganwell maybe it does15:55
dougwig"we definitely won't be writing a full networking stack as a plugin though :P" <-- i've heard neutron cores say this. :)15:56
rm_workhahaha15:56
jschwarzlol15:56
jschwarzhow did that turn out for them?15:56
rm_worksometimes I feel sorry even for the neutron devs15:56
sbalukoffHAha!15:56
* jschwarz is a neutron dev15:57
dougwigit's the "must have scalable open source".  pretty much means they signed up to write a full software SDN controller.15:57
* blogan pities jschwarz15:57
* jschwarz sits in a corner, cries alone in the dark15:57
*** HenryG has quit IRC15:57
* blogan collects his tears15:57
sbalukoffMmmm.... delicious!15:57
*** vjay has joined #openstack-lbaas15:59
jschwarzif you have 3 private networks with vas on each one, and across 3 compute nodes.. that's a lot of ports to create15:59
bloganjschwarz: it's late for you right now right?15:59
dougwigor they all have routes, and you don't need ports.15:59
jschwarzalso problematic because they are on a separate network for a reason15:59
jschwarzblogan, 19:00 right now, but I can meet in like 2 hours or so16:00
bloganjschwarz: if they are on a separate network for a reason then the user shouldn't create a load balancer to balance between the two16:00
jschwarzthen why even allow the option?16:00
bloganso that the users that want to do it can16:01
jschwarzthat's contradictory :P16:01
sbalukoffjsxhwarz: We do occasionally see strange requirements like this.16:01
sbalukoffjschwarz: There's often business reasons for keeping them on separate networks... and still other business reasons for needing them accessible by the same loadbalancer.16:02
jschwarzWell, the architecture needs some work (because we didn't start talking about it)16:04
sbalukoffExamples include:  Backend A is handled by organizational unit A, and backend B is handled by organizational unit B. And upper management has declared both shall be accessible via different URIs on the same website.16:04
jschwarzI do think that responsibility for this should reside in the L3 plugin, and not spread out across different plugins16:05
*** markmcclain has quit IRC16:05
jschwarzDo you guys have weekly meetings I can attend? :)16:05
sbalukoffAlso, we're used to dealing with "I don't care if it's not advisable, is it possible?"16:05
*** sballe has joined #openstack-lbaas16:06
*** vjay has quit IRC16:06
*** sbfox has joined #openstack-lbaas16:06
sbalukoffjschwarz: Yes! Wednesdays at 20:00 UTC16:06
jschwarzsbalukoff, agreed, but if we have the chance to make this right, why not try?16:06
sbalukoffShoot me an e-mail and I'll get you an invite (we're doing them over Webex for now):16:06
sbalukoffsbalukoff@bluebox.net16:06
*** VijayB has quit IRC16:07
sbalukoffWell, that's for Octavia16:07
jschwarzwill do in a sec16:07
sbalukoffNeutron LBaaS is Thursdays at 14:00 UTC16:07
sbalukoffIn #openstack-meeting16:07
*** sballe__ has joined #openstack-lbaas16:07
*** sballe_ has quit IRC16:08
jschwarzsbalukoff, sent :)16:09
sbalukoffjschwarz: Cool! I'm in a meeting at the moment, too.. will take care of this afterward.16:10
jschwarzroger roger16:10
*** sballe has quit IRC16:11
*** VijayB has joined #openstack-lbaas16:13
*** vivek-ebay has quit IRC16:16
*** yfried__ has quit IRC16:18
*** mlavalle has joined #openstack-lbaas16:19
*** samuelbercovici has joined #openstack-lbaas16:20
*** sballe__ has quit IRC16:21
*** samuelbercovici has quit IRC16:28
*** samuelbercovici has joined #openstack-lbaas16:29
*** jroyall has joined #openstack-lbaas16:38
*** samuelbercovici has quit IRC16:40
*** sballe has joined #openstack-lbaas16:55
*** sballe has quit IRC16:56
*** sballe has joined #openstack-lbaas16:56
*** sballe has quit IRC16:56
*** jroyall has left #openstack-lbaas16:57
*** sballe has joined #openstack-lbaas16:57
*** markmcclain has joined #openstack-lbaas17:02
*** jschwarz has quit IRC17:12
*** vivek-ebay has joined #openstack-lbaas17:14
*** jschwarz has joined #openstack-lbaas17:15
*** jschwarz has quit IRC17:25
*** jschwarz has joined #openstack-lbaas17:26
*** HenryG has joined #openstack-lbaas17:29
*** jschwarz has quit IRC17:31
*** jschwarz has joined #openstack-lbaas17:33
*** vivek-eb_ has joined #openstack-lbaas17:42
*** vivek-ebay has quit IRC17:43
*** fnaval has quit IRC17:51
*** vivek-eb_ has quit IRC18:01
*** vivek-ebay has joined #openstack-lbaas18:01
*** sballe has quit IRC18:08
*** Youcef has joined #openstack-lbaas18:09
*** sbfox has quit IRC18:15
*** yfried__ has joined #openstack-lbaas18:15
*** magesh has joined #openstack-lbaas18:21
*** jschwarz has quit IRC18:30
*** sbfox has joined #openstack-lbaas18:33
*** fnaval has joined #openstack-lbaas18:48
*** ptoohill has quit IRC18:57
*** ptoohill has joined #openstack-lbaas18:58
*** sbfox has quit IRC19:07
*** VijayB has quit IRC19:11
*** vivek-ebay has quit IRC19:22
*** vivek-ebay has joined #openstack-lbaas19:23
*** sbfox has joined #openstack-lbaas19:27
*** sbfox1 has joined #openstack-lbaas19:30
*** sbfox has quit IRC19:32
*** vivek-ebay has quit IRC19:50
*** vivek-ebay has joined #openstack-lbaas20:08
*** VijayB has joined #openstack-lbaas20:36
*** orion_ has joined #openstack-lbaas21:00
*** orion__ has quit IRC21:04
*** orion_ has quit IRC21:09
*** orion_ has joined #openstack-lbaas21:10
*** orion_ has quit IRC21:14
*** rm_you| has joined #openstack-lbaas21:24
*** rm_you has quit IRC21:26
*** openstackgerrit has quit IRC21:46
*** openstackgerrit has joined #openstack-lbaas21:48
*** markmcclain has quit IRC21:57
*** fnaval has quit IRC21:59
*** fnaval has joined #openstack-lbaas22:15
*** vivek-ebay has quit IRC22:16
*** vivek-ebay has joined #openstack-lbaas22:16
*** mlavalle has quit IRC22:26
*** vivek-ebay has quit IRC22:55
*** vivek-ebay has joined #openstack-lbaas22:56
*** sbfox1 has quit IRC22:56
*** sbfox has joined #openstack-lbaas22:58
*** jorgem1 has quit IRC23:01
*** sbfox has quit IRC23:17
*** sbfox has joined #openstack-lbaas23:22
*** sbfox has quit IRC23:24
*** vivek-ebay has quit IRC23:37
*** vivek-ebay has joined #openstack-lbaas23:37
*** xgerman has quit IRC23:40

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