Tuesday, 2015-02-24

*** sbalukoff has joined #openstack-lbaas00:07
openstackgerritMichael Johnson proposed stackforge/octavia: Implements Octavia Controller Worker  https://review.openstack.org/15149600:29
*** ajmiller has quit IRC00:53
*** devlaps has quit IRC01:11
*** mlavalle has quit IRC01:12
*** sbfox has quit IRC01:17
*** xgerman has quit IRC01:26
*** Madhu has quit IRC01:47
*** SumitNaiksatam has quit IRC02:22
*** vivek-ebay has quit IRC02:22
*** mestery has joined #openstack-lbaas02:48
*** sbalukoff has quit IRC03:05
*** vivek-ebay has joined #openstack-lbaas03:38
*** fnaval has quit IRC03:52
*** amotoki has joined #openstack-lbaas03:55
*** fnaval has joined #openstack-lbaas04:02
*** devlaps has joined #openstack-lbaas04:17
*** sbalukoff has joined #openstack-lbaas04:24
*** blogan_ has joined #openstack-lbaas04:32
*** fnaval has quit IRC04:33
*** blogan_ has quit IRC04:38
*** xgerman has joined #openstack-lbaas04:42
*** vivek-ebay has quit IRC04:58
*** SumitNaiksatam has joined #openstack-lbaas05:09
*** SumitNaiksatam has quit IRC05:10
*** xgerman has quit IRC05:10
*** devlaps has quit IRC05:17
*** cipcosma has joined #openstack-lbaas05:28
*** cipcosma has quit IRC06:36
*** kobis has joined #openstack-lbaas07:11
*** Miouge has joined #openstack-lbaas07:27
*** chlong has quit IRC07:34
*** evgenyf has joined #openstack-lbaas07:53
*** kobis has quit IRC07:54
*** kobis has joined #openstack-lbaas07:55
*** vivek-ebay has joined #openstack-lbaas07:59
*** woodster_ has quit IRC08:00
*** vivek-ebay has quit IRC08:03
*** cipcosma has joined #openstack-lbaas08:24
*** cipcosma has quit IRC08:24
openstackgerritEvgeny Fedoruk proposed openstack/neutron-lbaas: Neutron LBaaS v2 Radware driver implementation WIP until https://bugs.launchpad.net/neutron/+bug/1420960 will be fixed and CI will be up and running  https://review.openstack.org/15504209:10
openstackLaunchpad bug 1420960 in neutron "Pool's session persistence is deleted when HM added" [Undecided,In progress] - Assigned to Brandon Logan (brandon-logan)09:10
*** apuimedo has joined #openstack-lbaas09:14
*** kobis has quit IRC09:24
*** kobis has joined #openstack-lbaas09:25
*** openstackgerrit has quit IRC09:36
*** openstackgerrit has joined #openstack-lbaas09:36
openstackgerritEvgeny Fedoruk proposed openstack/neutron-lbaas: Neutron LBaaS v2 Radware driver implementation WIP until https://bugs.launchpad.net/neutron/+bug/1420960 will be fixed and CI will be up and running  https://review.openstack.org/15504209:49
openstackLaunchpad bug 1420960 in neutron "Pool's session persistence is deleted when HM added" [Undecided,In progress] - Assigned to Brandon Logan (brandon-logan)09:49
*** kobis has quit IRC10:16
*** sbfox has joined #openstack-lbaas10:17
*** kobis has joined #openstack-lbaas10:17
*** sbfox has quit IRC11:43
*** chlong has joined #openstack-lbaas11:56
*** chlong has quit IRC12:44
*** Miouge has quit IRC12:53
*** Miouge has joined #openstack-lbaas12:54
*** amotoki has quit IRC13:21
openstackgerritEvgeny Fedoruk proposed openstack/neutron-lbaas: Moving base driver code to a new location  https://review.openstack.org/15826513:28
*** woodster_ has joined #openstack-lbaas13:44
*** sbfox has joined #openstack-lbaas13:47
*** sbfox has quit IRC13:47
*** Miouge_ has joined #openstack-lbaas13:57
*** Miouge has quit IRC13:58
*** Miouge_ is now known as Miouge13:58
*** Miouge has quit IRC14:06
*** Miouge has joined #openstack-lbaas14:06
*** localloop127 has joined #openstack-lbaas14:15
*** amotoki has joined #openstack-lbaas14:24
*** apuimedo has quit IRC15:03
*** ajmiller has joined #openstack-lbaas15:10
*** localloop127 has quit IRC15:10
*** devlaps has joined #openstack-lbaas15:24
*** devlaps has quit IRC15:24
*** xgerman has joined #openstack-lbaas15:54
*** Aish has joined #openstack-lbaas15:59
*** sbfox has joined #openstack-lbaas16:04
*** fnaval has joined #openstack-lbaas16:08
*** sbfox has quit IRC16:08
*** Aish has left #openstack-lbaas16:26
*** TrevorV_ has joined #openstack-lbaas16:28
*** TrevorV_ has quit IRC16:29
TrevorVSo was anyone around from rax for the meeting?16:30
dougwigptoohill was there16:35
dougwigwe voted everyone else from rax off the island.16:35
xgermanand we voted to rename blogans handle into towgan16:37
*** mlavalle has joined #openstack-lbaas16:44
TrevorVThat's fine. Enjoy the Octavia direction!  Lata17:06
johnsomMaybe blogan found an immunity idol in his truck when he picked it up.....  The plot thickens17:07
*** amotoki has quit IRC17:12
*** evgenyf has quit IRC17:13
*** vivek-ebay has joined #openstack-lbaas17:19
*** TrevorV_ has joined #openstack-lbaas17:20
*** Aish has joined #openstack-lbaas17:24
*** sbfox has joined #openstack-lbaas17:24
*** kobis has quit IRC17:26
*** sbfox has quit IRC17:41
*** banix has joined #openstack-lbaas17:44
*** mwang2 has quit IRC17:59
*** sbfox has joined #openstack-lbaas18:02
*** sbalukoff has quit IRC18:03
*** jschwarz has joined #openstack-lbaas18:06
*** devlaps has joined #openstack-lbaas18:11
*** madhu has joined #openstack-lbaas18:14
*** jschwarz has quit IRC18:42
*** SumitNaiksatam has joined #openstack-lbaas18:51
openstackgerritAl Miller proposed openstack/neutron-lbaas: Add bytes_in and bytes_out statistics to the listener  https://review.openstack.org/15882318:59
*** sbalukoff has joined #openstack-lbaas19:09
*** crc32 has joined #openstack-lbaas19:13
*** Aish has quit IRC19:28
*** devlaps has quit IRC19:33
*** Miouge has quit IRC19:44
*** TrevorV_ has quit IRC19:50
*** Varun_Lodaya has joined #openstack-lbaas19:52
Varun_LodayaHey Guys, I am kinda confused about what source-IP based LB method implies? we have a separate source-IP persistence under VIP19:53
Varun_LodayaCan anybody help me understand what does that LB method imply?19:53
johnsomsource IP load balancing has to do with picking the initial backend server for the flow.  source-ip persistence has to do with sticking all subsequent connections from that source IP to the same backend.19:58
dougwigthere is a very subtle difference, that only a few in this group will insist is huge enough to warrant the confusion it causes.19:59
*** rm_work is now known as rm_work|away19:59
bloganim not one of those!20:00
johnsombedis probably has the best description20:00
Varun_LodayaBut if we select source-IP based load-balancing, we don’t really mention any particular backend along with it, so how does it pick one? and unless we have session persistence, would it still stick to that?20:00
dougwigVarun_Lodaya: it hashes the available backends.20:01
dougwigand yes, it sticks until a backend member is removed from the pool (or added)20:01
Varun_LodayaI see, It’s more of hashing the source-IP and picking based on hash-index20:01
johnsomAs the backend pool grows or shrinks it may not go to the same backend server outside of persistence20:01
bloganall i can say is that it wasn't compelling enough for me to remember the differences20:02
Varun_Lodayaas long as number of instances remain same, hashing will always point to the same real server20:02
dougwigVarun_Lodaya: the persistence is more of a "hash once, remember", the LB method is more of a "hash each time, which will usually be the same".20:02
dougwigVarun_Lodaya: right20:02
*** rm_work|away is now known as rm_work20:02
Varun_LodayaOk I get it now, the name is confusing, it should be called more of hash-based or something20:03
Varun_LodayaBut thanks for the clarification20:03
dougwigwe talked about re-labing it source ip hash, though both are source ip hash's, just applied at different places.  :)20:04
bloganid rather just get rid of it ;)20:04
bloganand get rid of round_robin and least_connections20:05
bloganand just have the LB send all traffic to every member20:05
Varun_Lodayathat would be round-robin right? ;)20:05
Varun_Lodayaor ECMP20:05
dougwigunicast -> broadcast gateway.20:06
blogansounds like a great load balancer20:06
bloganminus the balancing20:06
dougwigwe need to make sure to merge the replies back into unicast.20:07
bloganbut seriously though, source_ip algorithm is confusing for everyone20:07
Varun_Lodayawe used to call it source-IP-hash algorithm in our load-balancers back then20:08
dougwigwe could yank it.20:08
dougwigi map it to weighted-least-connection20:08
bloganajmiller: if the neutron-lbaas in tree devstack plugin is loaded with the one that already exists in devstack for v1 lbaas, would that cause a problem?20:10
blogani know i can test this out but wondering if its been thought about20:10
blogantesting now20:11
dougwiggood point.  if not, we'll likely need to edit to make it work, else we'll be in chicken/egg merge land.20:12
bloganwell why can't the one in devstack just depend on the one in devstack that removes it?20:13
blogansorry the one in neutron-lbaas20:13
dougwigthat should work, since no one has enable_plugin yet.  i'm most concerned about the q-lbaas token, as that will break all CIs everywhere without some magic in devstack itself.20:14
blogani agree we shouldn't break that, but deprecate it in whatever ways there are to do that20:15
bloganwhich im not sure that process even exists for something like a enabled_services token20:15
dougwigan alias file in devstack to map tokens to external plugins, and eventually phase them out over a few cycles, yeah.20:15
dougwigi know infra/qa is ready to drop us instantly, so we'll have to push to not break the world.20:15
blogani can't wait to get everything in tree20:17
blogani know we're probably the canary in the coal mine for this but it just seems like the better solution20:17
*** LinuxJedi has left #openstack-lbaas20:48
*** vivek-ebay has quit IRC21:02
ajmillerblogan Just back from errands and lunch21:08
ajmillerThe neutron-lbaas patch should just sit silently until it is enabled in localrc21:09
ajmillerSo https://review.openstack.org/#/c/155540 can go in without any side-effects to existing configurations.21:10
ajmillerHowever, as dougwig points out, the devstack patch https://review.openstack.org/#/c/153079/ needs a tweak to not break existing configs21:11
*** vivek-ebay has joined #openstack-lbaas21:19
bloganajmiller: im just wondering if the neutron-lbaas devstack review needs to be dependent on the one in devstack that removes the q-lbaas token, so that they both can't be enabled at the same time21:26
bloganor if them being enabled at the same time is even an issue21:26
*** sbfox has quit IRC21:26
ajmillerblogan.  The devstack patch depends on the neutron-lbaas patch.  I haven't made that explict in gerrit, do you think I should do that?21:27
blogani dont know yet, i started testing to see if it was an issue but got sidetrack and now im in a meeting21:28
*** sbfox has joined #openstack-lbaas21:28
*** vivek-ebay has quit IRC21:33
openstackgerritMerged openstack/neutron-lbaas: Updated from global requirements  https://review.openstack.org/15777221:40
TrevorVHey dougwig you around?21:47
openstackgerritGerman Eichberger proposed stackforge/octavia: Implements Octavia Controller Worker  https://review.openstack.org/15149621:57
*** banix has quit IRC22:03
*** apuimedo has joined #openstack-lbaas22:10
*** rm_work is now known as rm_work|away22:16
*** apuimedo has quit IRC22:26
*** madhu has quit IRC22:28
sballe__ptoohill: Are you around?22:28
*** openstackgerrit has quit IRC22:37
*** openstackgerrit has joined #openstack-lbaas22:37
dougwigTrevorV: ack22:38
TrevorVdougwig its all good, I think we've figured out what I was looking for22:39
dougwigsweet, i should lag in my replies more often22:40
*** vivek-ebay has joined #openstack-lbaas22:40
TrevorVHa ha nah, we just dug enough to find an answer (see what I did there?)22:40
ptoohillsballe__ im around22:42
sballe__oh great!22:44
sballe__I am looking at the gerrit reviews around the ha-proxy reference implementation and I can see you guys have done a lot of work :-)22:45
sballe__I am looking at https://review.openstack.org/#/c/144348/ and https://review.openstack.org/#/c/126801/. Is there another review that I am missing?22:46
ptoohillOne sec22:46
sballe__I meant https://review.openstack.org/#/c/150547/ not 1268122:47
sballe__https://review.openstack.org/#/c/150547/ is on the Amphorae and 144348 in the Controller.22:48
ptoohillWell, this was the amphora api22:48
sballe__yes agreed22:48
ptoohillThere was this that i had moved around if you guys were to going to use it, you would need to rebase where appropriate22:49
ptoohillbesides the deps there shouldnt be other reviews. unless im not understanding question22:50
openstackgerritFranklin Naval proposed openstack/neutron-lbaas: WIP: Neutron LBaaS: Basic Scenario Test  https://review.openstack.org/15890422:50
ptoohilloh, well there is the jinja templates22:50
ptoohilland the amphora api would also need to be based off of that22:51
ptoohillif you want to use those also22:51
sballe__ok. I am just trying to understand what yu guys have already done.22:51
ptoohillThe api itself was minimal, really needs to be thought from scratch honestly. Trevor was working on it but think the focus switched and not sure what actually got done22:52
ptoohillOn my side,22:52
ptoohilli did the templates and updated the driver to utilize the api methods22:52
ptoohilland simple tests for the driver. Though, that may end up being completely written22:53
sballe__ok cool! this will be a good starting point for me to start looking at it22:53
ptoohill:) hope it helps, let me know if theres anything i may be able to help with.22:54
ptoohillTrevor was more familiar with the api, but i can help with templates and that driver22:54
sballe__ptoohill: thx :-)22:54
ptoohillnp, im heading out of office, but should be on in 30 mins ;)22:54
*** rm_work|away is now known as rm_work23:00
*** apuimedo has joined #openstack-lbaas23:04
*** mestery has quit IRC23:07
*** vivek-ebay has quit IRC23:17
bloganping dougwig23:18
dougwigack towgan23:18
bloganagents is a neutron table, not a v1 table23:18
dougwigahh, ok.23:18
bloganand it is necessary unless we rewrite all the common agent code23:19
bloganwhich i would advise against23:19
*** chlong has joined #openstack-lbaas23:19
dougwigthe rest was good.23:19
bloganajmiller: seing an issue with the devstack plugin23:24
ajmillerblogan -- OK, what's up?23:24
bloganso devstack runs fine, but that is the message when it attempts to start the neutron-lbaas-agent23:25
bloganusing q-lbaas23:26
bloganin localrc23:26
bloganNEUTRON_AGENT_CONF isn't defined in your review so its gotta be in devstack core23:26
ajmillerAnd this is with merged devstack (not my patch)?23:27
ajmillerI'll look into that23:27
bloganok ill -1 for now, always possible its me that did somethign wrong, but i dont think so in this case because NEUTRON_AGENT_CONF doesn't exist in the new devstack plugin code23:28
ajmillerI will let you know when I have an update.23:29
*** chlong_ has joined #openstack-lbaas23:50

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