| *** yamamoto has joined #openstack-kuryr | 00:08 | |
| *** kiennt has joined #openstack-kuryr | 00:35 | |
| *** limao has joined #openstack-kuryr | 00:54 | |
| *** gouthamr has joined #openstack-kuryr | 01:01 | |
| *** aojea has joined #openstack-kuryr | 01:08 | |
| *** aojea has quit IRC | 01:13 | |
| *** edisonxiang has quit IRC | 01:18 | |
| *** limao has quit IRC | 01:31 | |
| *** limao has joined #openstack-kuryr | 01:33 | |
| *** limao has quit IRC | 01:33 | |
| *** limao has joined #openstack-kuryr | 01:33 | |
| *** edisonxiang has joined #openstack-kuryr | 01:53 | |
| *** yedongcan has joined #openstack-kuryr | 01:54 | |
| *** gouthamr has quit IRC | 02:49 | |
| *** aojea has joined #openstack-kuryr | 03:09 | |
| *** aojea has quit IRC | 03:14 | |
| *** yedongcan has quit IRC | 03:50 | |
| *** yedongcan has joined #openstack-kuryr | 03:52 | |
| *** yboaron has joined #openstack-kuryr | 04:08 | |
| janonymous | apuimedo: hi, tried privsep but got another error ..trying to work on it : http://paste.openstack.org/show/616262/ | 04:16 |
|---|---|---|
| *** limao has quit IRC | 04:17 | |
| *** limao has joined #openstack-kuryr | 04:17 | |
| *** limao has quit IRC | 04:22 | |
| *** limao has joined #openstack-kuryr | 05:03 | |
| *** tonanhngo has joined #openstack-kuryr | 05:10 | |
| *** aojea has joined #openstack-kuryr | 05:10 | |
| *** aojea has quit IRC | 05:15 | |
| *** janki has joined #openstack-kuryr | 05:25 | |
| *** kiennt has quit IRC | 05:32 | |
| *** aojea has joined #openstack-kuryr | 05:44 | |
| *** kiennt has joined #openstack-kuryr | 06:04 | |
| *** aojea has quit IRC | 06:10 | |
| *** longfei_zhang has joined #openstack-kuryr | 06:16 | |
| *** kiennt has quit IRC | 06:18 | |
| *** kiennt has joined #openstack-kuryr | 06:18 | |
| longfei_zhang | Hi, do we support the fixed ip address for clusterip service type? i mean i need to use a fixed ip when k8s creating the kube-dns service. | 06:28 |
| *** yedongcan has quit IRC | 06:29 | |
| *** yedongcan has joined #openstack-kuryr | 06:30 | |
| *** yamamoto has quit IRC | 06:34 | |
| *** kzaitsev_ws has joined #openstack-kuryr | 06:34 | |
| longfei_zhang | kzaitsev_ws, Hi, do we support the fixed ip address for clusterip service type? i mean i need to use a fixed ip when k8s creating the kube-dns service. | 06:35 |
| janonymous | apuimedo: making this change:paste.openstack.org/show/616267/ and moving it out of eventlet dir it was able to work atleast..... | 06:36 |
| longfei_zhang | janonymous, Hi, do we support the fixed ip address for clusterip service type? i mean i need to use a fixed ip when k8s creating the kube-dns service. | 06:37 |
| kzaitsev_ws | longfei_zhang: I believe kuryr-k8s does not support clusterip | 06:41 |
| longfei_zhang | so we just support ClusterIP type, but can not specify the fix ip for ClusterIP service , right? | 06:44 |
| kzaitsev_ws | longfei_zhang: https://github.com/openstack/kuryr-kubernetes/blob/master/kuryr_kubernetes/controller/handlers/lbaas.py#L56 | 06:44 |
| longfei_zhang | kzaitsev_ws, I know we only support ClusterIP service type, i'm not sure if we support use a fixed ip for this service? | 06:45 |
| longfei_zhang | i mean usually we have a fixed ip for dns service | 06:46 |
| longfei_zhang | kube-dns as a service in k8s cluster, so we fixed this ip in the yaml file | 06:47 |
| longfei_zhang | i'm not sure if kuryr will take this fixed ip when creating lb for it ? | 06:48 |
| irenab | longfei_zhang, please submit the bug if it does not work | 06:50 |
| janonymous | longfei_zhang: ^^ | 06:50 |
| longfei_zhang | irenab, ok | 06:55 |
| *** tonanhngo has quit IRC | 07:13 | |
| openstackgerrit | Hangdong Zhang proposed openstack/fuxi master: Update the documentation link for doc migration https://review.openstack.org/486510 | 07:25 |
| *** aojea has joined #openstack-kuryr | 07:28 | |
| *** janki is now known as janki|lunch | 07:37 | |
| *** apuimedo has quit IRC | 07:43 | |
| *** yamamoto has joined #openstack-kuryr | 07:45 | |
| *** apuimedo has joined #openstack-kuryr | 07:47 | |
| *** egonzalez has joined #openstack-kuryr | 07:47 | |
| *** tonanhngo has joined #openstack-kuryr | 08:04 | |
| *** tonanhngo has quit IRC | 08:08 | |
| *** kural_ has joined #openstack-kuryr | 08:09 | |
| *** kural has joined #openstack-kuryr | 08:09 | |
| openstackgerrit | Hangdong Zhang proposed openstack/kuryr master: Update the documentation link for doc migration https://review.openstack.org/486523 | 08:24 |
| *** tonanhngo has joined #openstack-kuryr | 08:32 | |
| *** tonanhngo has quit IRC | 08:37 | |
| *** janki|lunch is now known as janki | 08:40 | |
| *** c00281451 has joined #openstack-kuryr | 08:41 | |
| *** c00281451 is now known as zengchen | 08:41 | |
| *** garyloug has joined #openstack-kuryr | 08:47 | |
| *** yamamoto has quit IRC | 09:02 | |
| *** yamamoto has joined #openstack-kuryr | 09:15 | |
| longfei_zhang | garyloug, ping | 09:23 |
| *** limao has quit IRC | 09:23 | |
| *** tonanhngo has joined #openstack-kuryr | 09:24 | |
| longfei_zhang | i have a question about MACVLAN mode, there are some pods that have defined livenessprobe, which means the kubelet need to communicate with the pods through the pods ip, kubelet process and pods are on the same vm, so for macvlan mode , the kubelet can not communicate with the pods through the pod ip | 09:28 |
| *** tonanhngo has quit IRC | 09:28 | |
| garyloug | Hi longfei_zhang, this is a limitation of the Linux MACVLAN driver. The slave devices (pods) cannot reach the master (VM) and vice versa. | 09:41 |
| apuimedo | garyloug: longfei_zhang note that you can always add an extra port for the VM for that purpose | 09:44 |
| apuimedo | so that kubernetes probes work | 09:44 |
| longfei_zhang | garyloug, slave pods can not reach the master vm? | 09:45 |
| longfei_zhang | garyloug, do you mean the pod (on the slave vm )can not reach the slave vm or master vm ? | 09:47 |
| *** yamamoto has quit IRC | 09:47 | |
| garyloug | longfei_zhang, when I say master and slave I am referring to MACVLAN only. The MACVLAN master device is the VM interface. On top of this master interface we create slave devices, these are inserted into the pod namespace. | 09:49 |
| garyloug | The limitation with the driver is the slaves cannot contact their own master and master cannot contact it's own slaves. All other communication is possible. | 09:50 |
| *** longfei_zhang has quit IRC | 09:54 | |
| *** longfei_zhang has joined #openstack-kuryr | 09:59 | |
| longfei_zhang | garyloug, oh i got | 10:00 |
| *** yamamoto has joined #openstack-kuryr | 10:02 | |
| *** yamamoto has quit IRC | 10:02 | |
| *** tonanhngo has joined #openstack-kuryr | 10:18 | |
| *** tonanhngo_ has joined #openstack-kuryr | 10:20 | |
| *** tonanhngo has quit IRC | 10:23 | |
| *** tonanhngo_ has quit IRC | 10:25 | |
| longfei_zhang | apuimedo, thank you for the information, and i found one solution in this article http://backreference.org/2014/03/20/some-notes-on-macvlanmacvtap/ | 10:28 |
| longfei_zhang | " just create another macvlan on the host on the same lower device, and use this to communicate with the guest" | 10:28 |
| apuimedo | right | 10:29 |
| apuimedo | that's what I do | 10:29 |
| *** yamamoto has joined #openstack-kuryr | 10:33 | |
| *** kiennt has quit IRC | 10:33 | |
| *** tonanhngo has joined #openstack-kuryr | 11:02 | |
| *** tonanhngo has quit IRC | 11:07 | |
| apuimedo | janonymous: so the problem was on the unix server you were using to listen to the requests? | 11:27 |
| apuimedo | janonymous: yes. from the explanation of the unixstreamserver you were using before "These four classes process requests synchronously; each request must be completed before the next request can be started. This isn’t suitable if each request takes a long time to complete, because it requires a lot of computation, or because it returns a lot of data which the client is slow to process. The solution | 11:31 |
| apuimedo | is to create a separate process or thread to handle each request; the ForkingMixIn and ThreadingMixIn mix-in classes can be used to support asynchronous behaviour." | 11:31 |
| apuimedo | it was a not a good fit for cni -> cni_daemon comm | 11:32 |
| apuimedo | where a lot of cni execs can talk to the cni_daemon at once | 11:32 |
| *** tonanhngo has joined #openstack-kuryr | 11:34 | |
| *** tonanhngo has quit IRC | 11:39 | |
| irenab | janonymous, ping | 11:53 |
| irenab | about https://review.openstack.org/#/c/484754/, the gate is failing while waiting for the k8s-api to respond | 11:54 |
| irenab | see in http://logs.openstack.org/54/484754/1/check/gate-install-dsvm-default-kuryr-kubernetes/18df2b1/logs/devstacklog.txt.gz | 11:55 |
| irenab | "Waiting for Kubernetes API Server to respond++ /opt/stack/new/kuryr-kubernetes/devstack/plugin.sh:wait_for:363 : curl -o /dev/null -sIf http://158.69.84.20:8080" | 11:55 |
| *** yamamoto has quit IRC | 12:19 | |
| openstackgerrit | Merged openstack/kuryr-kubernetes master: Updated from global requirements https://review.openstack.org/481186 | 12:19 |
| openstackgerrit | Kirill Zaitsev proposed openstack/kuryr-kubernetes master: Allow passing multiple VIFs to CNI https://review.openstack.org/471012 | 12:21 |
| irenab | kzaitsev_ws, hi | 12:24 |
| irenab | I have some time slot to go through the list of awaiting revie patches. I wonder if you plan to follow-up on https://review.openstack.org/#/c/478458/ | 12:25 |
| *** yamamoto has joined #openstack-kuryr | 12:36 | |
| *** yamamoto has quit IRC | 12:36 | |
| kzaitsev_ws | irenab: I am | 12:36 |
| kzaitsev_ws | but | 12:36 |
| kzaitsev_ws | it's a patch that sits on top of 3 other | 12:36 |
| kzaitsev_ws | I thought I'll update it after we sort out the implementation | 12:37 |
| kzaitsev_ws | there's not much sense in updating the docs all the time. | 12:37 |
| kzaitsev_ws | I can -WF it if you like | 12:37 |
| * kzaitsev_ws fighting ironic at the moment %) | 12:38 | |
| openstackgerrit | Berezovsky Irena proposed openstack/kuryr-kubernetes master: devstack: Move to devstack-plugin-container https://review.openstack.org/474238 | 12:41 |
| *** yedongcan has left #openstack-kuryr | 12:42 | |
| irenab | kzaitsev_ws, no problem. Will try to progress with the former 3 patches | 12:44 |
| kzaitsev_ws | irenab: while we're at it we can try and discuss the plugin mechanism for the vifs | 12:45 |
| irenab | sure | 12:47 |
| kzaitsev_ws | also: the additional-subnets https://review.openstack.org/#/c/482544 is there mainly for demo reasons (and in case we decide to go with phased approach before TPR/CRD) | 12:47 |
| kzaitsev_ws | like phase-1 simple annotations, phase-2 TPR/CRD | 12:47 |
| irenab | Let me try to sum-on what I have in mind | 12:47 |
| irenab | maybe worth to mark as WIP or something, so we will not merge by accident | 12:48 |
| irenab | Back to the pluggable VIF | 12:48 |
| irenab | I do not like that now we need to modify VIF to add handling for any new annoation we add | 12:49 |
| irenab | I think somethign similar to ML2 mech/extensions drivers will be a good way to provide extandability | 12:49 |
| irenab | VIF will call some sort of 'processorsManager' that will propogate Pod to all existing 'processors' | 12:50 |
| kzaitsev_ws | smth like 'enabled_drivers' in the config, that is a list of drivers, that get a pod object and do whatever they want with it? | 12:50 |
| irenab | yes | 12:51 |
| kzaitsev_ws | I'm not sure we need a separate entity for that. it'll be a singleton anyway | 12:51 |
| irenab | so VIF stays without the change, while any exotic annotation is added for some new use case | 12:51 |
| irenab | we already have SR_IOV/Requested-port/subnets | 12:52 |
| kzaitsev_ws | yeah, VIFHandler's on_present would become unmanagable pretty soon. | 12:53 |
| kzaitsev_ws | agree on that | 12:53 |
| irenab | we need to clarify the API on this callback and then as you said, list of drivers to invoke | 12:56 |
| kzaitsev_ws | I can draft my idea with 'enabled_drivers' or 'vif_drivers' or 'vif_pipeline' (thinking about paste) and we can check if we can go somewhere from there | 12:57 |
| irenab | kzaitsev_ws, this will be great. | 12:59 |
| irenab | If you won't get to it during coming days, I may have some time later this week or next to draft something | 12:59 |
| *** tonanhngo has joined #openstack-kuryr | 13:00 | |
| *** rwallner has joined #openstack-kuryr | 13:00 | |
| *** rwallner has quit IRC | 13:01 | |
| *** rwallner has joined #openstack-kuryr | 13:01 | |
| kzaitsev_ws | yeah. I guess it would need to know about already allocate vif-names and some way to request new names, but apart from that a pod should suffice | 13:05 |
| kzaitsev_ws | the only problem is that it would be a rather invasive refactor | 13:06 |
| kzaitsev_ws | probably ) | 13:06 |
| kzaitsev_ws | irenab: I'll try to get to it tomorrow (gotta have a short day today and a long one tomorrow) and I'll ping you as soon as I upload the thing | 13:07 |
| irenab | kzaitsev_ws, thanks! | 13:11 |
| *** longfei_zhang has quit IRC | 13:14 | |
| *** yboaron_ has joined #openstack-kuryr | 13:15 | |
| kzaitsev_ws | irenab: ah, I see where the problem is. The drivers (vif/subnets/secgroups) are currently loosely coupled | 13:17 |
| *** rwallner has quit IRC | 13:17 | |
| kzaitsev_ws | we gotta think of a way for them to work together interchangeably | 13:17 |
| *** rwallner has joined #openstack-kuryr | 13:17 | |
| *** yboaron has quit IRC | 13:17 | |
| kzaitsev_ws | maybe also through config | 13:18 |
| *** yamamoto has joined #openstack-kuryr | 13:36 | |
| *** gouthamr has joined #openstack-kuryr | 13:40 | |
| *** yamamoto has quit IRC | 13:44 | |
| *** limao has joined #openstack-kuryr | 13:47 | |
| *** yboaron_ has quit IRC | 13:49 | |
| *** yboaron_ has joined #openstack-kuryr | 13:49 | |
| apuimedo | I'll have to start the minutes 5 minutes later | 13:58 |
| apuimedo | dmellado: can you start it and I'll join when I finish the current one? | 13:59 |
| dmellado | apuimedo: sure | 13:59 |
| apuimedo | thanks | 13:59 |
| *** kzaitsev_ws has quit IRC | 14:03 | |
| *** yboaron_ has quit IRC | 14:08 | |
| *** yboaron has joined #openstack-kuryr | 14:12 | |
| *** zengchen1 has joined #openstack-kuryr | 14:23 | |
| janonymous | irenab: aah, i will try to check what is wrong..last time gate was failing due to bug in gate.. | 14:24 |
| irenab | dmellado: apuimedo : in zun it is what apuimedo did: https://github.com/openstack-infra/project-config/blob/master/jenkins/jobs/zun.yaml#L31 | 14:40 |
| dmellado | irenab: ack, I'll need to check what does this do and how it would affect the gates | 14:40 |
| *** yboaron has quit IRC | 14:52 | |
| *** reedip has quit IRC | 14:56 | |
| *** zengchen1 has quit IRC | 14:57 | |
| *** apuimedo has quit IRC | 15:00 | |
| *** apuimedo has joined #openstack-kuryr | 15:01 | |
| *** reedip has joined #openstack-kuryr | 15:03 | |
| *** janki has quit IRC | 15:35 | |
| *** lihi has quit IRC | 15:37 | |
| *** lihi has joined #openstack-kuryr | 15:38 | |
| *** apuimedo has quit IRC | 15:41 | |
| *** apuimedo has joined #openstack-kuryr | 15:42 | |
| *** apuimedo has quit IRC | 15:43 | |
| *** apuimedo has joined #openstack-kuryr | 15:43 | |
| garyloug | Hi all, we have updated TPR/CDR section in this doc with our findings. https://docs.google.com/document/d/1GNZbmVxE2n5j07YRlkx4zP-XbFSPI9UEcVifX1MRXyY/edit#heading=h.12z7zo4b90h7 | 15:49 |
| apuimedo | thansk! | 15:51 |
| *** atoth has joined #openstack-kuryr | 15:52 | |
| *** aojea has quit IRC | 16:03 | |
| *** egonzalez has quit IRC | 16:08 | |
| *** aojea has joined #openstack-kuryr | 16:12 | |
| *** aojea has quit IRC | 16:17 | |
| *** vikasc has joined #openstack-kuryr | 16:18 | |
| *** aojea has joined #openstack-kuryr | 16:21 | |
| *** aojea has quit IRC | 16:26 | |
| *** aojea has joined #openstack-kuryr | 16:31 | |
| *** aojea has quit IRC | 16:36 | |
| *** aojea has joined #openstack-kuryr | 16:40 | |
| *** aojea has quit IRC | 16:44 | |
| *** garyloug has quit IRC | 17:06 | |
| *** alraddarla_ has joined #openstack-kuryr | 17:14 | |
| *** alraddarla has quit IRC | 17:17 | |
| *** alraddarla_ has quit IRC | 17:17 | |
| *** alraddarla has joined #openstack-kuryr | 17:18 | |
| *** atoth has quit IRC | 17:53 | |
| *** kural_ has quit IRC | 18:10 | |
| *** kural has quit IRC | 18:10 | |
| *** vikasc has quit IRC | 18:24 | |
| *** atoth has joined #openstack-kuryr | 19:21 | |
| *** alraddarla has quit IRC | 20:02 | |
| *** alraddarla has joined #openstack-kuryr | 20:06 | |
| *** aojea has joined #openstack-kuryr | 20:09 | |
| *** aojea has quit IRC | 20:13 | |
| *** aojea has joined #openstack-kuryr | 20:18 | |
| *** aojea has quit IRC | 20:22 | |
| *** aojea has joined #openstack-kuryr | 20:45 | |
| *** aojea has quit IRC | 20:50 | |
| *** alraddarla has quit IRC | 20:58 | |
| *** aojea has joined #openstack-kuryr | 21:03 | |
| *** aojea has quit IRC | 21:09 | |
| *** yamamoto_ has joined #openstack-kuryr | 21:09 | |
| *** pc_m has quit IRC | 21:11 | |
| *** yamamoto_ has quit IRC | 21:20 | |
| *** aojea has joined #openstack-kuryr | 21:21 | |
| *** aojea has quit IRC | 21:26 | |
| *** aojea has joined #openstack-kuryr | 21:40 | |
| *** pc_m has joined #openstack-kuryr | 21:40 | |
| *** aojea has quit IRC | 21:44 | |
| *** rwallner has quit IRC | 22:01 | |
| *** yamamoto has joined #openstack-kuryr | 22:01 | |
| *** aojea has joined #openstack-kuryr | 22:07 | |
| *** jgriffith has quit IRC | 22:10 | |
| *** jgriffith has joined #openstack-kuryr | 22:10 | |
| *** aojea has quit IRC | 22:12 | |
| *** jgriffith has quit IRC | 22:12 | |
| *** jgriffith has joined #openstack-kuryr | 22:15 | |
| *** jgriffith has quit IRC | 22:16 | |
| *** jgriffith has joined #openstack-kuryr | 22:19 | |
| *** jgriffith has quit IRC | 22:19 | |
| *** jgriffith has joined #openstack-kuryr | 22:23 | |
| *** gouthamr has quit IRC | 22:27 | |
| *** gouthamr has joined #openstack-kuryr | 22:41 | |
| *** aojea has joined #openstack-kuryr | 23:20 | |
| *** aojea has quit IRC | 23:24 | |
| *** aojea has joined #openstack-kuryr | 23:29 | |
| *** aojea has quit IRC | 23:33 | |
| *** aojea has joined #openstack-kuryr | 23:38 | |
| *** aojea has quit IRC | 23:42 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!