Tuesday, 2018-12-04

« Monday, 2018-12-03 Index Wednesday, 2018-12-05 »
*** premsankar has joined #openstack-kuryr01:21
*** takamatsu has quit IRC01:37
*** hongbin has joined #openstack-kuryr01:43
*** hongbin_ has joined #openstack-kuryr03:11
*** hongbin has quit IRC03:13
*** hongbin has joined #openstack-kuryr03:22
*** hongbin_ has quit IRC03:23
*** janki has joined #openstack-kuryr05:22
*** hongbin has quit IRC05:39
*** yboaron_ has joined #openstack-kuryr06:46
*** aojea has joined #openstack-kuryr06:50
*** takamatsu has joined #openstack-kuryr06:53
*** aojea has quit IRC07:04
*** pcaruana has joined #openstack-kuryr07:10
openstackgerritDaniel Mellado proposed openstack/kuryr-kubernetes master: devstack: Update Kubernetes to 1.13.0  https://review.openstack.org/62205607:29
openstackgerritNayan Deshmukh proposed openstack/kuryr-kubernetes master: Add missing $ for bash variable in plugin.sh  https://review.openstack.org/62208107:59
*** takamatsu has quit IRC08:15
*** gkadam has joined #openstack-kuryr08:22
openstackgerritLuis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Add namespaceSelector support for NetworkPolicies  https://review.openstack.org/61963208:27
openstackgerritLuis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Ensure pod relabeling is supported by the Network Policy  https://review.openstack.org/61852108:28
*** ccamposr has joined #openstack-kuryr08:37
openstackgerritDaniel Mellado proposed openstack/kuryr-kubernetes master: devstack: Update Kubernetes to 1.13.0  https://review.openstack.org/62205608:39
*** takamatsu has joined #openstack-kuryr08:46
openstackgerritmelissaml proposed openstack/kuryr master: Change openstack-dev to openstack-discuss  https://review.openstack.org/62211008:57
*** garyloug has joined #openstack-kuryr08:57
openstackgerritmelissaml proposed openstack/kuryr-kubernetes master: Change openstack-dev to openstack-discuss  https://review.openstack.org/62211208:59
openstackgerritmelissaml proposed openstack/kuryr-libnetwork master: Change openstack-dev to openstack-discuss  https://review.openstack.org/62211509:02
openstackgerritDaniel Mellado proposed openstack/kuryr-kubernetes master: devstack: Update Kubernetes to 1.13.0  https://review.openstack.org/62205609:04
openstackgerritDaniel Mellado proposed openstack/kuryr-kubernetes master: devstack: Update Kubernetes to 1.13.0  https://review.openstack.org/62205609:05
*** ccamposr has quit IRC09:06
*** ccamposr has joined #openstack-kuryr09:06
openstackgerritDaniel Mellado proposed openstack/kuryr-kubernetes master: devstack: Update Kubernetes to 1.13.0  https://review.openstack.org/62205609:06
openstackgerritmelissaml proposed openstack/kuryr-tempest-plugin master: Change openstack-dev to openstack-discuss  https://review.openstack.org/62212509:14
*** takamatsu has quit IRC09:14
*** takamatsu has joined #openstack-kuryr09:16
ltomasboirenab, I added some extra info at network_policy.rst as you suggested: https://review.openstack.org/#/c/619632. Let me know if that is enough09:23
irenabltomasbo, thanks. Will check asap09:24
ltomasboirenab, great! thanks! (no hurry)09:24
*** maysams has joined #openstack-kuryr09:42
*** takamatsu has quit IRC09:49
*** takamatsu has joined #openstack-kuryr09:50
irenabltomasbo, done09:50
ltomasboirenab, thanks!09:52
irenablets wait for CI to be happy and merge09:52
ltomasboirenab, yep! it was ok before documentation, so it should be ok too, but better to wait!09:54
openstackgerritGenadi Chereshnya proposed openstack/kuryr-kubernetes master: Debugging extra fips generation for port_pool failures  https://review.openstack.org/62216709:59
openstackgerritGenadi Chereshnya proposed openstack/kuryr-tempest-plugin master: Working with noop for vif_pool_driver  https://review.openstack.org/62141110:04
openstackgerritMerged openstack/kuryr-kubernetes master: devstack: parametrize openshift version fetching  https://review.openstack.org/61901810:09
openstackgerritLuis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Ensure pod relabeling is supported by the Network Policy  https://review.openstack.org/61852110:11
ltomasbodmellado, ^^ done with your remarks!10:11
ltomasboirenab, dulek, maysams, celebdor ^^ please review when you have time!10:11
*** takamatsu has quit IRC10:13
openstackgerritGenadi Chereshnya proposed openstack/kuryr-tempest-plugin master: Working with noop for vif_pool_driver  https://review.openstack.org/62141110:20
openstackgerritLuis Tomas Bolivar proposed openstack/kuryr-tempest-plugin master: Working with noop for vif_pool_driver  https://review.openstack.org/62141110:21
celebdordulek: dmellado: https://review.openstack.org/#/c/618545/10:34
dmelladocelebdor let's see what you have to offer here xD10:37
*** takamatsu has joined #openstack-kuryr10:50
openstackgerritmelissaml proposed openstack/kuryr master: Change openstack-dev to openstack-discuss  https://review.openstack.org/62211010:52
openstackgerritmelissaml proposed openstack/kuryr-kubernetes master: Change openstack-dev to openstack-discuss  https://review.openstack.org/62211210:54
*** takamatsu has quit IRC11:04
*** takamatsu has joined #openstack-kuryr11:05
dulekWhooo, today's a good day, got my NIE card.11:14
*** yboaron_ has quit IRC11:15
dulekSorry for being out so long, wasn't expecting it'll take so much time.11:15
dulekGetting on track with reviews now.11:15
*** yboaron_ has joined #openstack-kuryr11:15
openstackgerritMerged openstack/kuryr master: Change openstack-dev to openstack-discuss  https://review.openstack.org/62211011:19
*** maysams has quit IRC11:29
*** aperevalov has quit IRC11:50
openstackgerritYossi Boaron proposed openstack/kuryr-kubernetes master: Allocate service FIP after LB was provisioned  https://review.openstack.org/62225711:57
*** yboaron_ has quit IRC12:07
openstackgerritMerged openstack/kuryr-kubernetes master: Add namespaceSelector support for NetworkPolicies  https://review.openstack.org/61963212:10
openstackgerritDaniel Mellado proposed openstack/kuryr-kubernetes master: Add KURYR_ENABLED_DRIVERS to devstack sample files  https://review.openstack.org/62227212:10
*** yboaron_ has joined #openstack-kuryr12:12
dmelladodulek: congrats! xD12:17
dulekdmellado: :)12:18
dmelladoyou should assume that these kind of things *always* take a long time in Spain xD12:18
*** maysams has joined #openstack-kuryr12:18
dulekdmellado: It's not that bad, I'm here only since last month. ;)12:18
dulekdmellado: I just don't understand that I'm here for a month only and yet antiimmigration party gets 10% of votes in the election.12:19
dulekLike come one, it's only temporary and I'll be nice, really.12:19
duleks/one/on12:19
openstackgerritDaniel Mellado proposed openstack/kuryr-kubernetes master: Add KURYR_ENABLED_HANDLERS to devstack sample files  https://review.openstack.org/62227212:19
dmelladodulek: LOL12:20
dmelladothat's so sad, dude!12:20
dmelladotbh I was freaking out when I read the results from elections12:20
dmelladoI mean, if you get an overview about what the world's doing is just normal12:20
dmelladobut...12:20
dulekWell, dunno. Whatever, at least I'm "more legal" here now. ;)12:22
dmelladodulek: actually the card is quite handy, you can use to perform most of the legal actions that you'd be interested into using internet12:23
dmelladoas you can sign using it12:23
dulekdmellado: Oh, that's neat.12:23
dmelladoand they have just updated opensc to support it without an issue, unlike before, when it was a nightmare12:24
dmelladoso you got lucky xD12:24
*** mrostecki has joined #openstack-kuryr12:53
openstackgerritMerged openstack/kuryr-kubernetes master: devstack: Wait for the k8s api to be healthy  https://review.openstack.org/61854512:59
*** rh-jelabarre has joined #openstack-kuryr13:01
openstackgerritMerged openstack/kuryr-tempest-plugin master: Add network policy gate to experimental pipeline  https://review.openstack.org/62155013:13
*** aperevalov has joined #openstack-kuryr13:26
*** aojea_ has joined #openstack-kuryr13:28
*** yboaron_ has quit IRC13:29
*** aojea_ has quit IRC13:47
*** janki has quit IRC13:50
openstackgerritGenadi Chereshnya proposed openstack/kuryr-tempest-plugin master: Making sure all the pods reside on the same node for port_pool  https://review.openstack.org/62233114:04
*** aojea_ has joined #openstack-kuryr14:15
*** irclogbot_1 has quit IRC14:15
*** janki has joined #openstack-kuryr14:29
*** aojea_ has quit IRC14:33
*** aojea_ has joined #openstack-kuryr14:40
*** janki has quit IRC14:44
*** irclogbot_1 has joined #openstack-kuryr14:54
openstackgerritAntoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: move master config to separate dir  https://review.openstack.org/62159715:02
*** sean-k-mooney has joined #openstack-kuryr15:37
openstackgerritLuis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Ensure pod relabeling is supported by the Network Policy  https://review.openstack.org/61852115:41
ltomasboirenab, dmellado ^^ if you are still around, I lost your +2s fixing the docs issue (missing a few spaces here and there)15:43
dmelladoltomasbo: there you go15:44
dmelladodamn docs xD15:44
ltomasboxD thanks!15:45
sean-k-mooneyhi folks can people on the kuryr/kuryr kubernetes core team take a look at http://lists.openstack.org/pipermail/openstack-discuss/2018-November/000416.html15:46
openstackgerritMerged openstack/kuryr-kubernetes master: Change openstack-dev to openstack-discuss  https://review.openstack.org/62211215:54
sean-k-mooneyltomasbo: o/16:04
ltomasbosean-k-mooney, o/16:05
sean-k-mooneyltomasbo: i dont know if you saw but could you bring http://lists.openstack.org/pipermail/openstack-discuss/2018-November/000416.html up at the next kuryr-kubernetes team meeting?16:06
sean-k-mooneymy self and the other people working on os-vif would like to know how ye are using it as we are planning to change the datamodel going forward in a way that my break ye16:07
sean-k-mooneyeverything in https://github.com/openstack/kuryr-kubernetes/blob/master/kuryr_kubernetes/objects/vif.py causes us concern as none of it is allowed by what we considerd our public api so we want to get your input and understand if we can change some of your current useage paterens16:09
*** pcaruana has quit IRC16:12
ltomasbocelebdor, dmellado ^^16:13
dmelladohey!16:13
dmelladolet me see16:13
ltomasbosean-k-mooney, why is that not allowed?16:13
sean-k-mooneyos-vif has never allowed inheritance form object under os_vif.objects16:14
dmelladowell 'break everyhing' doesn't sound good xD16:14
sean-k-mooneyand to date we have not supported serialisation of the objects or garenteed compatible betten objects form two different veriosn of the lib16:15
dmelladosean-k-mooney: ack, I'll handle thi16:16
dmellados16:16
sean-k-mooneywe would like to avoid breaking this but we would like to move you vif object ot os-vif itself and then understand if you are serialising them how and are they stored or sent between services16:16
dmelladocould you attend our next Mon meeting?16:16
celebdorsean-k-mooney: thanks for reaching out16:17
celebdorsean-k-mooney: the serializations are stored in the kubernetes API16:17
celebdorwhich further stores them on etcd16:17
celebdorthey are stored in the annotations field16:17
sean-k-mooneycelebdor: ok thats not ideal but it is good to know16:17
celebdorbut we are considering moving that from annotations to CRD16:17
celebdoralso, as part of moving the info to CRD, we may not serialize that anymore16:18
sean-k-mooneydmellado: when is your next meeting16:18
* sean-k-mooney goes to look it up16:18
dmelladosean-k-mooney: next Mon16:18
dmellado15:00 CET16:18
dmelladoat #openstack-meeting-416:19
sean-k-mooneyah ok ill add it to my calander16:19
celebdorand either add the necessary information in a CRD and have the CNI daemon craft the os-vif object16:19
dmelladosean-k-mooney, thanks!16:19
dmelladoI'm in the middle of another meeting16:19
celebdoror just use that info to plug the device directly (since we have old plugging logic (pre os-vif adoption) in openstack/kuryr (kuryr-lib)16:19
dmelladocelebdor: please address sean-k-mooney comments here16:19
dmelladoand we'll discuss later16:19
dmelladosean-k-mooney: thanks for reaching out!16:20
celebdordmellado: where's *here* ?16:20
sean-k-mooneydmellado: no worries16:20
dmelladocelebdor: over the irc channel16:20
dmelladojust like you're doing16:20
celebdoroh16:20
sean-k-mooneycelebdor: well or to the mailing list post16:20
sean-k-mooneyi can chat for a while here but jaypies and other will be interested too16:20
celebdorsorry, I failed to move to the openstack-discuss mailing list16:21
sean-k-mooneycelebdor: e.g. you have not yet our your not planning too ?16:21
celebdorI have not yet16:22
celebdorso I did not receive the email you sent, I just saw it in the link you posted here16:22
sean-k-mooneyah well i can resent the main in a while to let you and other reply16:22
sean-k-mooneythere are really two parts to https://github.com/openstack/kuryr-kubernetes/blob/master/kuryr_kubernetes/objects/vif.py16:23
sean-k-mooneythe tree vif objects at the bottom which may be able to use existing vifs in os-vif or that could be moved16:23
sean-k-mooneythen there is the serialisation of the objects and storage in k8s/etcd16:24
sean-k-mooneyby the way i also notice ye have a noop plugin https://github.com/openstack/kuryr-kubernetes/blob/master/setup.cfg#L27 we added one to os-vif in rocky too so that should not be needed anymore16:25
celebdorsean-k-mooney: ok. I am subscribed now16:25
celebdorthanks16:26
sean-k-mooneyok ill resend in a few minute. i think the old openstack-dev list is retried now also16:27
celebdoryes, I received an email about its retiring16:27
celebdorI only read it today16:27
celebdorI had quite a bit of piled up mail16:27
celebdor~200 emails among all the accounts16:27
duleksean-k-mooney: Okay, read the email, catched up with the scrollback.16:28
*** gkadam has quit IRC16:28
duleksean-k-mooney: So I guess we have two issues here.16:28
dulekOne is that we do unsupported extensions of your o.vo's.16:28
sean-k-mooneyya that the more minor of the two16:28
dulekThis is fairly easy to fix if we provide a rationale to include them into os-vif, right?16:28
sean-k-mooneyyep eitehr we can add them directly as they are or see if we can consolidate with existing vif and port profiles16:29
duleksean-k-mooney: Good point, port profiles may be way to go.16:29
duleksean-k-mooney: The second is that we persist os-vif's o.vos and now you want to drop backward compatibility.16:30
duleksean-k-mooney: Now that one is funny, but not really tragic since we now have mechanism of versioning the annotations.16:31
sean-k-mooneydulek: well we have not decided but we want to change some of the names of classes and may or may not move away form ovo eventrully16:31
sean-k-mooneywe want to rename VIFBase to vif for example becase well its not an abstract class and its confusing to new people16:32
duleksean-k-mooney: Yeah, so in case of the latter we'd need to know at least a release before to be able to force a new format.16:33
dulekOr copy o.vo definitions into our code.16:33
sean-k-mooneyby the way there is a kuryr vif https://github.com/openstack/os-vif/blob/master/os_vif/objects/vif.py#L165 and and port profile for dpdk already so we are happy to add more if needed16:33
dulekOtherwise we won't be able to read annotations on Pods created by older Kuryr-Kubernetes versions.16:33
duleksean-k-mooney: I think I've reviewed the patch adding it. :)16:33
duleksean-k-mooney: Now the renames of the classes are quite fun case…16:34
dulekBecause there's no mechanism in o.vo to do that in backward-compatible way…16:35
sean-k-mooneythe renames we can support initally via the oslo decolector i think16:35
dulekAnd in Kuryr we still want to be able to read older annotations.16:35
duleksean-k-mooney: oslo decolecotr?16:35
sean-k-mooneyhttps://docs.openstack.org/debtcollector/latest/reference/index.html#debtcollector.removals.removed_class16:37
sean-k-mooneyhttps://github.com/openstack/debtcollector16:37
dulekAh, debt…16:37
duleksean-k-mooney: I don't think this will help on o.vo level.16:38
duleksean-k-mooney: We will need functions translating old class to new one and handling that manually. I don't see any other viable approach.16:39
sean-k-mooneyit depends if you are crateing VIFBase objects then no if you are creating things derived form VIFBase it will be fine16:39
sean-k-mooneydulek: the thing is while we used ovo to define the VIF we never considerd calling any of the ovo methodes to be supported yet16:40
duleksean-k-mooney: Not really: https://github.com/openstack/kuryr-kubernetes/blob/8e60dcc4aa4cc41a4c0276c4b8fdcb775d2eb6f1/kuryr_kubernetes/objects/vif.py#L33-L3416:40
sean-k-mooneythat was all predicated on use in neutron api16:40
dulekWe never create VIFBase objects but we assume all VIF's are superclasses of it.16:41
sean-k-mooneyi belive obj_osvif.VIFBase woudl still be a valid symbol16:41
sean-k-mooneye.g. if we did  class VIF: ...16:41
sean-k-mooneyand VIFBase  = VIF16:42
sean-k-mooneythe any use of VIFBase woudl just become an alias of VIF16:42
duleksean-k-mooney: But for o.vo to work correctly it would need to be in o.vo registry and with all the metaclass black magic that o.vo does I'm not sure how that would behave. :D16:42
dulekI mean for o.vo serialization-deserialization.16:43
sean-k-mooneydulek: that certenly a possiblity16:43
duleksean-k-mooney: Okay, from my perspective it looks like this. You don't consider os-vif o.vo's a stable interface. We incorrectly do that.16:44
dulekThe only way out is for kuryr-kubernetes to say "bye os-vif" and create our own objects.16:44
sean-k-mooneydulek: that would be the worst case senario16:44
dulekAnd then do the transition over Stein->Train.16:44
sean-k-mooneywe intend them to become a more stable interface in the futre and that is what is motivatin this conversation16:45
sean-k-mooneythe thing is when reviweing https://review.openstack.org/#/c/607610/4/specs/stein/approved/generic-os-vif-offloads.rst@28916:45
duleksean-k-mooney: Okay, but first you want to do some substantial changes to those interfaces.16:46
duleksean-k-mooney: That will possibly break us.16:46
sean-k-mooneyjan mention that os-vif was only used by nova and i rembere kuryer kuberneted did and extended the vif16:46
duleksean-k-mooney: So either we drop that dependency or you do the changes in the way that will allow us to keep compatibility.16:46
duleksean-k-mooney: And the latter might be a bit hard and limiting for you.16:47
sean-k-mooneydulek: well we did not even know you we may need compatible until i check how ye used os-vif.16:47
sean-k-mooneydulek: so no decision have been made16:48
sean-k-mooneyill resend the mail to the discuss list are you on it16:48
duleksean-k-mooney: Okay, makes sense. I think we should start with some kind of list of what you want to drop/change/rename.16:48
dulekAnd checking how hard would it be to keep that compatible?16:48
sean-k-mooneyyes16:49
sean-k-mooneyif we can keep compatiblity or come up with a way to  upgrade/downgrade the object then we will try16:50
duleksean-k-mooney: Okay. Changes in the objects itself should be more-or-less easy. The broader changes like renaming or deleting of classes are the main pain point.16:50
dulekBut I don't say they're impossible to handle.16:51
sean-k-mooneydulek: celebdor  if i send the mail again if 10 mins woudl ye be able to reply on the list16:51
sean-k-mooneyya well renameing the classes could be avoided until later16:52
duleksean-k-mooney: Sure thing! I do see the initial email BTW.16:52
sean-k-mooneywe might be able to provide a object feild definiton to replace https://github.com/openstack/kuryr-kubernetes/blob/8e60dcc4aa4cc41a4c0276c4b8fdcb775d2eb6f1/kuryr_kubernetes/objects/vif.py#L3316:52
duleksean-k-mooney: Ha, then we fear the "later". :P16:52
sean-k-mooneyso that you dont have to dircetly use obj_osvif.VIFBase.__name__16:53
duleksean-k-mooney: Yeah, that sounds like feasible thing, though remember that we still need to be able to deserialize older objects.16:53
sean-k-mooneyyep we wil have to add some functional test cases16:54
celebdordulek: so you reply16:54
celebdorinline16:54
celebdorand this way I'll see the original too16:54
dulekcelebdor: So you'll read it? Sure. :)16:54
sean-k-mooneyfor now we will likely put the backward imcompatible chagne on hold for a while. we have a kuryr job that will tell us if a patch would break ye16:55
sean-k-mooneyalther it may not catch upgrades.16:55
dulekcelebdor: You can still read it in the archive. ;)16:55
duleksean-k-mooney: Yeah, it won't capture that.16:55
duleksean-k-mooney: So no need to resend that, let me answer to the initial email.16:56
sean-k-mooneywell no that im aware ye are serialising things we can review for change to the object that may impact it16:56
sean-k-mooneyoh your on the list cool16:56
sean-k-mooneyanyway thanks for your time. let contiue the conversation on the ml and we will try not to break ye :)16:57
*** takamatsu has quit IRC17:06
*** pcaruana has joined #openstack-kuryr17:12
openstackgerritMichał Dulko proposed openstack/kuryr-kubernetes master: Add experimental gate with cri-o  https://review.openstack.org/62038817:21
*** aojea_ has quit IRC17:23
*** ccamposr has quit IRC17:52
*** pcaruana has quit IRC17:56
*** maysams has quit IRC18:12
openstackgerritMerged openstack/kuryr-kubernetes master: devstack: Update Kubernetes to 1.13.0  https://review.openstack.org/62205618:36
*** ccamposr has joined #openstack-kuryr18:43
openstackgerritMichał Dulko proposed openstack/kuryr-kubernetes master: Introduce test case document for SRIOV functionality  https://review.openstack.org/60002219:02
*** ccamposr has quit IRC19:24
*** celebdor has quit IRC20:13
*** EricAdamsZNC2 has joined #openstack-kuryr20:13
*** EricAdamsZNC has quit IRC20:14
openstackgerritMerged openstack/kuryr-tempest-plugin master: Change openstack-dev to openstack-discuss  https://review.openstack.org/62212520:28
*** aojea has joined #openstack-kuryr20:39
*** takamatsu has joined #openstack-kuryr20:44
openstackgerritYossi Boaron proposed openstack/kuryr-kubernetes master: Allocate service FIP after LB was provisioned  https://review.openstack.org/62225721:42
*** aojea has quit IRC21:43
*** aojea has joined #openstack-kuryr22:17
*** takamatsu has quit IRC22:29
*** celebdor has joined #openstack-kuryr22:32
*** celebdor has quit IRC22:44
*** aojea has quit IRC22:50
*** aojea has joined #openstack-kuryr23:42
« Monday, 2018-12-03 Index Wednesday, 2018-12-05 »

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