Thursday, 2016-11-24

*** huikang has joined #openstack-kuryr00:15
*** hongbin has quit IRC00:27
*** huikang has quit IRC00:27
*** limao has joined #openstack-kuryr00:28
*** shashank_hegde has quit IRC00:39
*** huikang has joined #openstack-kuryr00:51
*** diogogmt has quit IRC01:09
*** tonanhngo has joined #openstack-kuryr01:14
*** tonanhngo has quit IRC01:16
*** tonanhngo has joined #openstack-kuryr01:34
*** tonanhngo has quit IRC01:35
*** shashank_hegde has joined #openstack-kuryr01:38
*** hongbin has joined #openstack-kuryr02:13
openstackgerritDongcan Ye proposed openstack/kuryr: Fix wrong exception catch in port_unbind  https://review.openstack.org/40114802:14
*** diogogmt has joined #openstack-kuryr02:15
*** yedongcan has joined #openstack-kuryr02:15
*** tonanhngo has joined #openstack-kuryr02:44
*** tonanhngo has quit IRC02:47
*** vikasc has quit IRC03:04
*** tonanhngo has joined #openstack-kuryr03:05
*** tonanhngo has quit IRC03:08
*** vikasc has joined #openstack-kuryr03:21
*** yamamoto_ has joined #openstack-kuryr03:22
*** tonanhngo has joined #openstack-kuryr03:24
*** tonanhngo has quit IRC03:27
*** tonanhngo has joined #openstack-kuryr03:44
*** tonanhngo has quit IRC03:47
*** huikang has quit IRC04:02
*** tonanhngo has joined #openstack-kuryr04:04
*** tonanhngo has quit IRC04:05
*** tonanhngo has joined #openstack-kuryr04:24
*** tonanhngo has quit IRC04:26
*** tonanhngo has joined #openstack-kuryr04:44
*** tonanhngo has quit IRC04:45
*** irenab has quit IRC05:16
*** tonanhngo has joined #openstack-kuryr05:44
*** tonanhngo has quit IRC05:47
*** tonanhngo has joined #openstack-kuryr06:10
*** tonanhngo has quit IRC06:11
*** hongbin has quit IRC06:12
*** irenab has joined #openstack-kuryr06:13
*** yedongcan has quit IRC06:18
*** yedongcan has joined #openstack-kuryr06:19
*** yedongcan1 has joined #openstack-kuryr06:21
*** yedongcan has quit IRC06:24
openstackgerritvikas choudhary proposed openstack/kuryr: Nested-Containers: vlan driver  https://review.openstack.org/36199306:27
*** tonanhngo has joined #openstack-kuryr06:32
*** tonanhngo has quit IRC06:34
openstackgerritDongcan Ye proposed openstack/kuryr-libnetwork: Fix bug in ipam_release_address  https://review.openstack.org/40187406:36
*** yedongcan has joined #openstack-kuryr06:39
*** yedongcan1 has quit IRC06:39
*** yedongcan has quit IRC06:46
*** yedongcan has joined #openstack-kuryr06:52
*** tonanhngo has joined #openstack-kuryr06:54
*** tonanhngo has quit IRC06:56
*** oanson has joined #openstack-kuryr06:57
*** tonanhngo has joined #openstack-kuryr07:14
*** tonanhngo has quit IRC07:15
*** yamamoto_ has quit IRC07:18
*** yedongcan1 has joined #openstack-kuryr07:32
*** dimak has joined #openstack-kuryr07:33
*** yedongcan has quit IRC07:33
*** tonanhngo has joined #openstack-kuryr07:39
*** tonanhngo has quit IRC07:42
*** tonanhngo has joined #openstack-kuryr07:56
*** shashank_hegde has quit IRC07:56
*** tonanhngo has quit IRC07:56
*** yamamoto_ has joined #openstack-kuryr07:58
irenabapuimedo, ping08:02
apuimedoirenab: pong08:06
apuimedojoining08:06
apuimedosorry for the delay08:06
irenabnp08:06
*** tonanhngo has joined #openstack-kuryr08:18
*** tonanhngo has quit IRC08:19
*** dimak has quit IRC08:24
openstackgerritMerged openstack/kuryr: Fix wrong exception catch in port_unbind  https://review.openstack.org/40114808:27
apuimedovikasc: https://review.openstack.org/#/c/401338/08:28
vikascapuimedo, done..08:31
vikascapuimedo, i am facing an unexpected problem while trying devstack..kubelet start is failing with "no option --cni-bin-dir". earlier it was working. Though cni is not there anyways, i got it working by commenting out this option from plugin.sh08:34
apuimedovikasc: try with the latest master after my patches get merged08:34
apuimedoI fixed it yesterday08:34
irenabvikasc, check what version of hyperkube you runs08:34
vikascirenab, 1.3.708:34
apuimedoif it continues failing remove the hyperkube image08:34
apuimedoand devstack will pull a good new onw08:35
apuimedo*one08:35
irenabapuimedo, had to bump the version to 1.4.608:35
vikascapuimedo, i already tried removing hyperkube image08:35
apuimedoso git pull08:35
apuimedoto get my latest changes08:35
openstackgerritMerged openstack/kuryr-kubernetes: devstack: add convenience kubectl executable  https://review.openstack.org/40133808:36
vikascapuimedo, ok, will try out08:36
apuimedothanks08:36
apuimedovikasc: let me know how it goes08:36
vikascapuimedo, also i need to manually install requirements.. otherwise kuryr-lib does not get install and kuryr-k8s service start fails. am i missing some step08:39
*** tonanhngo has joined #openstack-kuryr08:39
apuimedovikasc: that didn't happen to me somehow08:40
*** tonanhngo has quit IRC08:40
apuimedovikasc: can you try reproducing in a clean VM?08:40
irenabvikasc, same was for me yesterday08:40
irenabivc_, can confirm too08:42
vikascapuimedo, irenab restacking with 1.4.6 in localrc and master kuryr-k8s, deleted older hyperkube08:46
vikascirenab, today i also tried out almost all the latest ivc's changes. I think now i got a better overall picture. i am not able to add 'suggestions' on dvref doc you working on otherwise i could have added some points.08:49
irenabvikasc, let me change the permissions, so you can add08:49
vikascirenab, thanks08:50
irenabvikasc, https://docs.google.com/document/d/1hExm0TNp_OMWY_XMVRYpSvg5G8kFrL5hcrT9M_NDCwo/edit?usp=sharing08:50
irenabvikasc, let me know if there is still problem08:50
vikascirenab, 'suggestions' mode working now08:51
vikascirenab, thanks08:51
irenabyou welcome08:51
*** janki has joined #openstack-kuryr08:54
*** tonanhngo has joined #openstack-kuryr09:04
*** tonanhngo has quit IRC09:06
vikascirenab, apuimedo , kubelet worked, but had to install requirements manually to get kuryr-k8s started09:07
*** tonanhngo has joined #openstack-kuryr09:15
*** tonanhngo has quit IRC09:16
apuimedothat's odd09:30
*** garyloug has joined #openstack-kuryr09:43
*** lmdaly has joined #openstack-kuryr09:50
*** yedongcan1 has left #openstack-kuryr09:55
*** limao has quit IRC10:34
garylougHi all, I've a small commit that's failing Jenkins unit tests. However, when I run unit tests locally I can't get them to fail. Any ideas? https://review.openstack.org/#/c/397640/10:41
apuimedogaryloug: probably depends on newer kuryr-lib. Unit tests run with release, not with master10:51
apuimedowe need to release a new kuryr-lib10:51
apuimedoltomasbo: take a look at https://review.openstack.org/#/c/361993/ when you can10:55
ltomasbohi apuimedo10:56
ltomasboactually testing it (previous version) right now10:56
apuimedoltomasbo: hi10:56
apuimedoah. That's great!10:56
apuimedothanks a lot10:56
ltomasbowith some modifications to the kuryr-libnetwork for the subport creation10:57
garylougapuimedo: Thanks11:02
apuimedogaryloug: thank me after I make the release :P11:02
apuimedoltomasbo: so you are trying the mode where one vlan is used for several containers, right?11:02
ltomasbono, actually I'm trying the one vlan per subport11:03
ltomasbovikasc is working on the binding and I'm working on the kuryr-libnetwork11:04
apuimedook11:04
ltomasbowill upload some code as soon as I have it ready11:04
ltomasbogetting the container connected properly, but the traffic flow is not commit back...11:04
vikascltomasbo, i was expecting yr comments as you said11:05
ltomasboand there is a problem when trying to attach the port as created right now to the trunk11:05
vikascapuimedo, i thought i would cover cni11:05
apuimedoltomasbo: in that case, please check the comments I made to https://review.openstack.org/#/c/361993/17/kuryr/lib/segmentation_type_drivers/vlan.py in regards to parameter modification. If it is something you do not need, then let's drop that11:05
apuimedovikasc: you mean passing ivc_ cni prototype to kuryr-lib after the release?11:06
vikascapuimedo, i meant updating cni with neutron api calls needed to complete vlan driver support for kuryr-k8s11:07
vikascapuimedo, or kurys-k8s controller11:07
apuimedothat's gonna be writing a driver for kuryr-kubernetes controller11:08
vikascapuimedo, yes11:08
apuimedoyou can check how the drivers work and make a new one that instead of getting a port, it sets up the nested stuff11:08
apuimedoit would be really nice11:08
apuimedoI'm working on functional and fullstack testing now11:08
vikascapuimedo, sounds good11:08
ltomasboapuimedo: didn't we say we will target first the one subport per vlan, and then the other use case (1 vlan per subnet)11:09
ltomasbo?11:09
vikascapuimedo, are you also planning to uploading cni patch to kuryr-k8s?11:09
apuimedoltomasbo: yes. That's exactly right11:09
apuimedo:-)11:09
apuimedoI dindn't say otherwise, I just was confused by 'with some modifications to the kuryr-libnetwork for the subport creation'11:10
apuimedobut then you explained the scope of the modifications11:10
apuimedovikasc: no. ivc_ is posting one based on os-vif11:10
apuimedothen we'll move it to kuryr-lib11:10
vikascapuimedo, ok, cool11:10
ltomasbosorry, was writing an email at the same time...11:10
apuimedos/move/adapt/11:10
vikascapuimedo, so we changed decision to put cni to kuryr-k8s and bubbled later to kuryr-lib11:11
ivc_apuimedo, vikasc, i'll try to post a wip/poc cni using os-vif (and if possible cni binary 'bridge' and 'dhcp' drivers) this week or early next week11:12
apuimedovikasc: yes. That was the decision we took in the weekly irc meeting11:12
vikascthanks ivc_ and apuimedo11:13
ivc_vikasc, apuimedo, to clarify, we start with cni in kuryr-k8s repo and move it to kuryr-lib as it matures11:13
irenabvikasc, its just to move faster with end to end flow, and to extract later to kuryr-lib11:13
vikascconfusion11:13
vikascapuimedo, just said to kuryr-lib...11:13
vikascsorry, got it11:14
irenabapuimedo, ?11:14
vikasc:)11:14
vikascall fine11:14
apuimedothat's right. I think it is the natural way of working and I was wrong11:14
apuimedopushing lib -> usages ends usually with unnatural code. Usage -> lib is better11:15
apuimedovikasc: irenab: we can take https://review.openstack.org/#/c/384349/11:24
irenabsure11:25
openstackgerritMerged openstack/kuryr: Add cover produced after coverage run in .gitignore  https://review.openstack.org/38434911:30
*** garyloug has quit IRC11:39
openstackgerritvikas choudhary proposed openstack/kuryr: Nested-Containers: vlan driver  https://review.openstack.org/36199311:45
vikascapuimedo, https://review.openstack.org/#/c/361993/17/kuryr/lib/segmentation_type_drivers/vlan.py "modifying the parameter is often a bit sketchy. What is the purpose of that? Is it used after calling this method?"11:47
vikascapuimedo, 'allocated' is local cache of allocated segmentation ids11:48
vikascapuimedo, if we dont update, 'allocated' will have incorrect info11:49
vikascapuimedo, and we need allocate in case of coe restart11:51
apuimedovikasc: my point is, who is using this cache?11:53
apuimedoif ltomasbo kuryr-libnetwork's code does not use it11:53
apuimedolet's not have it11:53
vikascapuimedo, his code should populate this11:53
vikascapuimedo, and it will be used to determine what all ids were already reserved before kuryr-libnetwork reboo11:54
vikascs/reboot11:54
vikascs/reboot/restart11:54
ltomasboI'm not using it (right now) but I'll need to use it (or similar) to get the vlan when creating the subport11:54
ltomasbobut that is needed at RequestAddress step11:55
apuimedothe thing is. The cache in any case lives in kuryr-libnetwork, so I prefer if kuryr-libnetwork manages it. You should tread the parameter as read only. Kuryr-libnetwork then, if it wants, can use the return allocated to update its cache11:55
apuimedootherwise we are just coupling it too much11:55
apuimedovikasc: ltomasbo: agreed?11:56
vikascapuimedo, then in that case vlan driver will have to query neutron to collect previously allocated ids11:57
vikascapuimedo, but i think neutron apis should be invoked from coe's11:57
ltomasbogiving it a second thought11:58
vikascapuimedo, let say there are some containers who are using some vlan ids and somehow kuryr gets restarted11:58
ltomasbobut agree with apuimedo, kuryr-libnetwork is the one needing that information11:58
ltomasboto create the subport11:59
ltomasbothen, it can be sent to the binding driver11:59
vikascltomasbo, 'allocated' is the id requested for the subport. 'allocated_ids' is another set which has all the allocated vlan ids at any moment.12:02
vikascltomasbo, apuimedo : sorry, many be i dint get you point. Can you guys please ask reword the question?12:03
vikascs/many/may12:03
vikasci think i understood yr point apuimedo :). you are asking that kuryr-libnetwork should update 'allocated_ids'(cache) with 'allocated'(received from kuryr-lib).12:06
openstackgerritvikas choudhary proposed openstack/kuryr: Nested-Containers: vlan driver  https://review.openstack.org/36199312:08
vikascltomasbo, apuimedo PTAL12:08
openstackgerritvikas choudhary proposed openstack/kuryr: Nested-Containers: vlan driver  https://review.openstack.org/36199312:16
openstackgerritvikas choudhary proposed openstack/kuryr: Nested-Containers: vlan driver  https://review.openstack.org/36199312:19
vikascirenab, ^12:28
*** lmdaly has quit IRC12:33
irenabvikasc, asap12:34
*** janki has quit IRC13:07
*** dimak has joined #openstack-kuryr13:09
apuimedovikasc: yup, you got it13:13
vikascapuimedo, yay XD13:14
irenabvikasc, done13:26
vikascirenab, i dint get your concern over vlan availability?13:27
irenabvikasc, there is an assumption that all vlans are available13:31
irenabbut what if some are used already?13:31
vikascirenab, those will be communicated by coe in 'allocated_ids' , please take a look at my reply on comment13:32
irenabat what phase will this call sent by COE?13:33
vikascfor requesting a vlan-id at port creation for container13:34
vikascchild port under vm (parent) port13:35
irenablet me try to explain what I was trying to imply13:35
*** garyloug has joined #openstack-kuryr13:35
vikascltomasbo is going to add that part in kuryr-libnetwork and i will be working on kuryr-k8s side13:35
vikascsure13:35
irenabCannot there be the case where some VLANs are used for some other purposes, not to be assigned for containers?13:35
irenabSo maybe there should be some exclude_vlans at the intialization phase?13:36
irenabOr you assume that COE is the only authority to consume vlans?13:36
vikascwe can add a exclude_vlans config option and coe can add these vlans to allocated_ids and then those will be taken care of while allocating new id13:37
vikascall this has to be done on coe side i think13:37
*** yamamoto_ has quit IRC13:38
irenabthe 'client' of the allocate_segmentation_id is not in this patch, right?13:38
vikascno, that will be in coes...13:38
vikasci have started working on writing a driver for kuryr-k8s13:39
irenabDo you mean like CNI Driver?13:39
vikascltomasbo working for kuryr-libnetwork13:39
vikascnot cni driver,13:40
vikasclogic for making corresponding neutron api calls, like creating trunk port, child port etc13:40
vikascthat would be done in controller side drivers13:40
irenabgot it, so the Controller part for the k8s, still kuryr code13:41
vikascyes13:41
vikascthis patch is binding part only13:41
irenabso just missing the update for the condig to add SegmentationDriver and unit tests13:41
vikasccni driver will use this13:41
*** neiljerram has quit IRC13:42
vikascunit tests are not there for any of the existing binding drivers also.. so would you mind if i add for this driver also in later patches13:42
vikascmay be i will open a bug ticket on my name13:43
vikascfor unit tests for the sake of tracking13:43
vikascirenab, ^13:43
vikascirenab, about docstring, you want me to remove just first line or complete doc string including parameter descriptions also?13:48
vikascirenab, i would prefer to push it as a seperate patch, some people might have opinion to keep it as it is.13:50
openstackgerritLouise Daly proposed openstack/kuryr-libnetwork: [WIP]Driver based model for kuryr-libnetwork  https://review.openstack.org/40036513:50
irenabok, will remove -1, will let you decide :-)13:50
openstackgerritvikas choudhary proposed openstack/kuryr: Nested-Containers: vlan driver  https://review.openstack.org/36199313:54
vikascirenab, updated segmentation id :)13:54
vikascapuimedo, irenab https://review.openstack.org/36199313:54
*** irenab has quit IRC13:57
*** neiljerram has joined #openstack-kuryr13:59
*** oanson has quit IRC14:03
*** lmdaly has joined #openstack-kuryr14:05
*** dimak has quit IRC14:05
lmdalyapologies for being slow replying to comments on the libnetwork driver - have pushed a patch and replied now14:06
lmdalyirenab suggested a 'base driver class for all methods to be overridden by derived drivers' and I like this idea so I was going to implement this if people agree?14:08
lmdalyI am also working on whitelisting drivers for the different implementations (ie. veth will only use veth binding driver and nested can only use ipvlan/macvlan unless a full patch is specified)14:10
lmdalyoh and the unit tests of course :D14:10
*** limao has joined #openstack-kuryr14:16
*** limao_ has joined #openstack-kuryr14:19
*** yamamoto has joined #openstack-kuryr14:21
*** hongbin has joined #openstack-kuryr14:21
*** yamamoto_ has joined #openstack-kuryr14:22
*** limao has quit IRC14:22
*** yamamoto has quit IRC14:25
*** yamamoto_ has quit IRC14:37
*** yamamoto has joined #openstack-kuryr14:42
*** yamamoto has quit IRC14:54
*** hongbin has quit IRC15:00
vikascltomasbo, ping15:01
*** openstackgerrit has quit IRC15:03
*** yamamoto has joined #openstack-kuryr15:03
*** openstackgerrit has joined #openstack-kuryr15:04
*** yamamoto has quit IRC15:05
*** lmdaly has quit IRC15:07
*** lezbar has quit IRC15:07
*** lezbar has joined #openstack-kuryr15:10
*** yamamoto has joined #openstack-kuryr15:10
*** garyloug has quit IRC15:23
*** diogogmt has quit IRC15:28
*** yamamoto has quit IRC15:31
*** irenab has joined #openstack-kuryr15:37
*** hongbin has joined #openstack-kuryr15:40
*** salv-orlando has joined #openstack-kuryr15:58
openstackgerritLiping Mao proposed openstack/kuryr-libnetwork: Build docker busybox image locally  https://review.openstack.org/39995816:03
*** oanson has joined #openstack-kuryr16:04
*** yamamoto has joined #openstack-kuryr16:12
*** diogogmt has joined #openstack-kuryr16:12
*** yamamoto has quit IRC16:13
*** yamamoto has joined #openstack-kuryr16:21
*** yamamoto has quit IRC16:22
openstackgerritLiping Mao proposed openstack/kuryr-libnetwork: [WIP]kuryr-libnetwork rally test job error log  https://review.openstack.org/39995016:27
*** limao_ has quit IRC16:28
*** limao has joined #openstack-kuryr16:28
*** limao has quit IRC16:33
*** yamamoto has joined #openstack-kuryr16:36
*** oanson has quit IRC16:43
*** lmdaly has joined #openstack-kuryr17:04
*** garyloug has joined #openstack-kuryr17:20
*** oanson has joined #openstack-kuryr17:27
*** oanson has quit IRC17:32
*** yamamoto has quit IRC17:38
*** lmdaly has quit IRC17:51
*** garyloug has quit IRC18:04
*** salv-orl_ has joined #openstack-kuryr18:30
*** salv-orlando has quit IRC18:32
*** salv-orl_ has quit IRC19:27
*** oanson has joined #openstack-kuryr20:01
*** oanson has quit IRC20:09
*** yamamoto has joined #openstack-kuryr20:39
*** yamamoto has quit IRC20:43
*** diogogmt has quit IRC21:58
*** salv-orlando has joined #openstack-kuryr22:39
*** salv-orlando has quit IRC23:13

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