| *** limao has joined #openstack-kuryr | 00:03 | |
| *** tonanhngo has joined #openstack-kuryr | 00:12 | |
| *** tonanhngo has quit IRC | 00:34 | |
| *** tonanhngo has joined #openstack-kuryr | 00:35 | |
| *** tonanhngo has quit IRC | 00:40 | |
| *** tonanhngo has joined #openstack-kuryr | 00:43 | |
| *** tonanhngo has quit IRC | 00:43 | |
| *** yedongcan has joined #openstack-kuryr | 01:09 | |
| *** hongbin has joined #openstack-kuryr | 01:38 | |
| *** caowei has joined #openstack-kuryr | 01:47 | |
| *** edisonxiang_ has joined #openstack-kuryr | 02:10 | |
| *** hongbin_ has joined #openstack-kuryr | 02:11 | |
| *** dougbtv_ has joined #openstack-kuryr | 02:15 | |
| *** hongbin has quit IRC | 02:18 | |
| *** dougbtv has quit IRC | 02:18 | |
| *** edisonxiang has quit IRC | 02:18 | |
| *** kzaitsev1pi has quit IRC | 02:18 | |
| *** kzaitsev_pi has joined #openstack-kuryr | 02:19 | |
| *** huats has joined #openstack-kuryr | 02:19 | |
| *** kzaitsev1pi has joined #openstack-kuryr | 02:25 | |
| *** edisonxiang_ is now known as edisonxiang | 02:27 | |
| *** huats_ has quit IRC | 02:28 | |
| *** kzaitsev_pi has quit IRC | 02:28 | |
| *** hongbin_ has quit IRC | 02:43 | |
| *** yamamoto has quit IRC | 02:52 | |
| *** yamamoto has joined #openstack-kuryr | 02:53 | |
| janonymous | zengchen: sry could not reply earlier, i saw your discussion in sig-machinery though | 02:54 |
|---|---|---|
| *** hongbin has joined #openstack-kuryr | 02:55 | |
| *** limao has quit IRC | 03:05 | |
| *** kiennt has joined #openstack-kuryr | 03:11 | |
| *** limao has joined #openstack-kuryr | 03:36 | |
| *** aojea has joined #openstack-kuryr | 03:41 | |
| *** limao has quit IRC | 03:42 | |
| *** aojea has quit IRC | 03:46 | |
| *** hongbin has quit IRC | 03:56 | |
| *** tonanhngo has joined #openstack-kuryr | 04:02 | |
| *** openstackgerrit has joined #openstack-kuryr | 04:44 | |
| openstackgerrit | Jaivish Kothari(janonymous) proposed openstack/kuryr-kubernetes master: CNI Daemon template https://review.openstack.org/480028 | 04:44 |
| *** caowei has quit IRC | 04:56 | |
| *** yamamoto has quit IRC | 05:03 | |
| *** yamamoto has joined #openstack-kuryr | 05:04 | |
| *** yamamoto has quit IRC | 05:11 | |
| *** limao has joined #openstack-kuryr | 05:24 | |
| *** caowei has joined #openstack-kuryr | 05:33 | |
| *** kiennt has quit IRC | 05:50 | |
| *** kiennt has joined #openstack-kuryr | 06:08 | |
| *** janki has joined #openstack-kuryr | 06:16 | |
| *** yamamoto has joined #openstack-kuryr | 06:19 | |
| *** kiennt has quit IRC | 06:22 | |
| *** kiennt has joined #openstack-kuryr | 06:23 | |
| *** yamamoto has quit IRC | 06:24 | |
| *** kzaitsev_ws has joined #openstack-kuryr | 06:39 | |
| *** ashishb has joined #openstack-kuryr | 06:42 | |
| *** ajo has joined #openstack-kuryr | 06:44 | |
| *** yamamoto has joined #openstack-kuryr | 06:52 | |
| *** yamamoto has quit IRC | 07:07 | |
| zengchen | janonymous:yes, After discussing with them, I think the design of watch framework is almost right. Why need 'list + watch' pattern? Because 'watch' may failed with the exception of 'resource version too old'. That means | 07:37 |
| zengchen | the resource version sending to K8S by 'watch' is smaller than the one k8s keep in its cache. Why it will happen? The capacity of cache of K8S is limited. The 'watch' will fail when the speed of fetching the events(by watch) is less than the speed of generating events. At this scenario, it can not get events by 'watch' anymore and should 'list' all the resources and watch again. | 07:37 |
| apuimedo | zengchen: resourceversion=0 can never get 'version too old' | 07:48 |
| zengchen | apuimedo: it will get a new resource version when calling 'watch' one time. Then call 'watch' with the new resource version which will be too old. | 07:50 |
| apuimedo | zengchen: IIRC, if you do watch with resourceversion=0 it will get the current state and continue watching | 07:55 |
| apuimedo | you don't need to send two watches | 07:55 |
| zengchen | apuimedo:how do you get the results when the watch does not return? there is a parameter of timeout for watch interface. IF you don't set, k8s may give it a default value. the request will return after that time. | 08:00 |
| *** garyloug has joined #openstack-kuryr | 08:01 | |
| apuimedo | zengchen: I'm not sure I got the question | 08:01 |
| janonymous | apuimedo: http://paste.openstack.org/show/618133/ | 08:01 |
| apuimedo | what does it matter if they are in order | 08:02 |
| apuimedo | ? | 08:02 |
| *** openstackgerrit has quit IRC | 08:02 | |
| apuimedo | what does it matter to know when you've caught up? | 08:03 |
| apuimedo | that is to say | 08:03 |
| apuimedo | what do you do with the information that you've caught up? | 08:03 |
| apuimedo | line 6 makes a point though | 08:04 |
| janonymous | apuimedo: i pasted reply for zenchen's query by k8s guy, zenchen may explain now :) | 08:04 |
| *** yamamoto has joined #openstack-kuryr | 08:04 | |
| zengchen | janonymous:ho, you have copyed the explanation liggitt said. | 08:05 |
| janonymous | zengchen: ya , just for apuimedo to get context :) | 08:05 |
| zengchen | janonymous:do you understand liggitt's comments? | 08:06 |
| janonymous | zengchen: yea, but i have similar doubts like apuimedo's though.. | 08:07 |
| zengchen | janonymous:I think liggitt's meaning is that the watch interface is not very reliable. it should use list + watch together. | 08:07 |
| apuimedo | zengchen: that was my intention with the PoC | 08:07 |
| apuimedo | but I worried that between the list and the watch we may have enough events to go out of the cache | 08:07 |
| apuimedo | (resourceversion wise) | 08:08 |
| janonymous | apuimedo: As per ivc comment on 480028 , should i upload a dependent patch as many changes were needed for that and would take more time to test.. | 08:09 |
| apuimedo | https://github.com/kubernetes/kubernetes/pull/852 | 08:09 |
| apuimedo | janonymous: link? | 08:10 |
| janonymous | https://review.openstack.org/#/c/480028/12 | 08:10 |
| zengchen | apuimedo: it can not use only 'watch' at the current disign of k8s. | 08:11 |
| apuimedo | zengchen: I would really like more information as to why | 08:11 |
| zengchen | apuimedo:yes, I will try. | 08:12 |
| *** yamamoto has quit IRC | 08:17 | |
| apuimedo | zengchen: regarding "If you're watching a rarely changed resource, the most recent resourceVersion on a particular object in a stream of events you get by watching from 0 may still be old enough to be outside the watch window or have already been compacted." it seems to me that if we keep a resourceversion per resource type, this does not affect us | 08:20 |
| apuimedo | zengchen: reading https://github.com/kubernetes/kubernetes/pull/13910 | 08:21 |
| apuimedo | they are discussing the safety of reasourceversion=0 | 08:21 |
| apuimedo | http://paste.openstack.org/show/618136/ | 08:24 |
| zengchen | yes, it seems a little complex. I need some time to understand the design of watch. | 08:27 |
| apuimedo | zengchen: you know what | 08:30 |
| apuimedo | ? | 08:30 |
| apuimedo | since there is https://github.com/kubernetes/kubernetes/issues/13969 anyway. which means they want to kill the watch with resourceversion=0 | 08:31 |
| apuimedo | do it with list + watch | 08:31 |
| zengchen | apuimedo:I think we should reach an agreement first before doing next work. | 08:33 |
| *** yamamoto has joined #openstack-kuryr | 08:44 | |
| apuimedo | zengchen: I thought you wanted to do list+watch, so my previous message was me agreeing to it | 08:45 |
| apuimedo | :P | 08:46 |
| *** yamamoto has quit IRC | 08:56 | |
| janonymous | apuimedo: suggestions on how to proceed with patch | 09:00 |
| apuimedo | janonymous: I'll look | 09:02 |
| janonymous | apuimedo: thanks | 09:03 |
| apuimedo | is this about ivc's comment to https://review.openstack.org/#/c/480028/10/kuryr_kubernetes/cni/api.py ? | 09:03 |
| janonymous | apuimedo: yeah the last comment: CNIDaemonizedRunner | 09:05 |
| janonymous | apuimedo: and CNIStandaloneRunner | 09:06 |
| janonymous | apuimedo: Last comment in https://review.openstack.org/#/c/480028/ | 09:06 |
| janonymous | so, should i upload a dependent patch or in same patch | 09:10 |
| apuimedo | janonymous: should be in the same patch | 09:12 |
| janonymous | apuimedo: ohkay! | 09:13 |
| apuimedo | thanks janonymous | 09:13 |
| *** openstackgerrit has joined #openstack-kuryr | 09:13 | |
| openstackgerrit | Jaivish Kothari(janonymous) proposed openstack/kuryr-kubernetes master: CNI Daemon template https://review.openstack.org/480028 | 09:13 |
| janonymous | apuimedo: but tests are still required :( , and release date is short.. | 09:14 |
| apuimedo | janonymous: don't worry about that | 09:16 |
| janonymous | apuimedo: :) | 09:16 |
| apuimedo | we'll make Pike a 0.2.0 and then early queens we'll release 1.0.0 | 09:16 |
| janonymous | apuimedo: oh ohkay, thanks! | 09:17 |
| *** yamamoto has joined #openstack-kuryr | 09:24 | |
| apuimedo | thanks to you ;-) | 09:25 |
| *** limao has quit IRC | 09:26 | |
| *** yamamoto has quit IRC | 09:35 | |
| *** kiennt has quit IRC | 10:01 | |
| *** egonzalez has joined #openstack-kuryr | 10:03 | |
| zengchen | apuimedo:sorry, i left for a conference. I will update the patch and keep on studying the pattern of list + watch. I hope I will solve your questions. | 10:12 |
| *** zengchen has quit IRC | 10:18 | |
| *** yamamoto has joined #openstack-kuryr | 10:32 | |
| *** limao has joined #openstack-kuryr | 10:40 | |
| *** yamamoto has quit IRC | 10:48 | |
| *** limao_ has joined #openstack-kuryr | 11:03 | |
| *** limao has quit IRC | 11:03 | |
| *** limao_ has quit IRC | 11:09 | |
| *** limao has joined #openstack-kuryr | 11:09 | |
| *** garyloug has quit IRC | 11:12 | |
| *** garyloug has joined #openstack-kuryr | 11:12 | |
| *** caowei has quit IRC | 11:15 | |
| *** garyloug_ has joined #openstack-kuryr | 11:20 | |
| *** garyloug has quit IRC | 11:20 | |
| *** ajo has quit IRC | 11:22 | |
| *** limao has quit IRC | 11:24 | |
| *** egonzalez has quit IRC | 11:24 | |
| *** garyloug_ has quit IRC | 11:25 | |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: documentation: Document how to install services https://review.openstack.org/492959 | 11:27 |
| apuimedo | janonymous: kzaitsev_ws: please, take a look at this patch ^^ | 11:28 |
| apuimedo | it was painful to do | 11:28 |
| apuimedo | (I found two bugs in Octavia while writing it that thankfully got fixed in the last two days) | 11:28 |
| janonymous | apuimedo : Great! i will review shortly :D | 11:29 |
| *** limao has joined #openstack-kuryr | 11:30 | |
| *** garyloug has joined #openstack-kuryr | 11:30 | |
| apuimedo | thanks | 11:32 |
| *** yamamoto has joined #openstack-kuryr | 11:35 | |
| *** dougbtv_ is now known as dougbtv | 11:39 | |
| *** yamamoto has quit IRC | 11:48 | |
| *** yamamoto has joined #openstack-kuryr | 11:48 | |
| *** garyloug has quit IRC | 11:54 | |
| *** yamamoto has quit IRC | 11:55 | |
| *** yamamoto has joined #openstack-kuryr | 11:57 | |
| *** egonzalez has joined #openstack-kuryr | 12:05 | |
| *** janonymous has quit IRC | 12:19 | |
| *** janonymous has joined #openstack-kuryr | 12:19 | |
| *** gouthamr has joined #openstack-kuryr | 12:19 | |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: documentation: Document how to install services https://review.openstack.org/492959 | 12:30 |
| *** janki has quit IRC | 12:58 | |
| *** egonzalez has quit IRC | 13:13 | |
| *** egonzalez has joined #openstack-kuryr | 13:16 | |
| *** irenab has quit IRC | 13:18 | |
| *** lihi has quit IRC | 13:18 | |
| *** ajo has joined #openstack-kuryr | 13:18 | |
| *** oanson has quit IRC | 13:19 | |
| *** irenab has joined #openstack-kuryr | 13:19 | |
| *** oanson has joined #openstack-kuryr | 13:19 | |
| *** lihi has joined #openstack-kuryr | 13:20 | |
| *** hongbin has joined #openstack-kuryr | 13:44 | |
| hongbin | limao: hi liping | 13:44 |
| limao | hongbin: Hi | 13:45 |
| hongbin | limao: need your help for one thing, i need this patch to cut the stable/pike branch for fuxi: https://review.openstack.org/#/c/489844/ | 13:45 |
| limao | sure, let me see | 13:45 |
| hongbin | limao: :) | 13:45 |
| limao | hongbin: this is OK for me , at that time one CI is failed, it is ok now, merging | 13:46 |
| hongbin | limao: thx | 13:46 |
| hongbin | limao: have a good weekend | 13:46 |
| limao | hongbin: wlc, you too ;-) | 13:47 |
| *** ajo has quit IRC | 14:30 | |
| *** egonzalez has quit IRC | 16:12 | |
| *** kzaitsev_ws has quit IRC | 16:13 | |
| *** ajo has joined #openstack-kuryr | 16:40 | |
| *** egonzalez has joined #openstack-kuryr | 16:46 | |
| *** ajo has quit IRC | 17:04 | |
| *** ajo has joined #openstack-kuryr | 17:09 | |
| *** ajo has quit IRC | 17:11 | |
| *** yamamoto has quit IRC | 17:43 | |
| *** ajo has joined #openstack-kuryr | 17:46 | |
| *** egonzalez has quit IRC | 17:56 | |
| *** egonzalez has joined #openstack-kuryr | 18:04 | |
| *** egonzalez has quit IRC | 18:08 | |
| *** yamamoto has joined #openstack-kuryr | 18:44 | |
| *** yamamoto has quit IRC | 18:49 | |
| *** limao has quit IRC | 19:32 | |
| *** limao has joined #openstack-kuryr | 19:32 | |
| *** limao has quit IRC | 19:37 | |
| *** yamamoto has joined #openstack-kuryr | 19:46 | |
| *** egonzalez has joined #openstack-kuryr | 19:51 | |
| *** yamamoto has quit IRC | 19:55 | |
| *** dougbtv_ has joined #openstack-kuryr | 19:56 | |
| *** dougbtv has quit IRC | 19:59 | |
| *** pc_m has quit IRC | 20:04 | |
| *** pc_m has joined #openstack-kuryr | 20:04 | |
| *** gouthamr has quit IRC | 20:14 | |
| *** vikasc has quit IRC | 20:27 | |
| *** vikasc has joined #openstack-kuryr | 20:28 | |
| *** gouthamr has joined #openstack-kuryr | 20:32 | |
| *** dougbtv__ has joined #openstack-kuryr | 21:02 | |
| *** dougbtv_ has quit IRC | 21:06 | |
| openstackgerrit | Merged openstack/fuxi master: Deploy fuxi-server via uwsgi https://review.openstack.org/489844 | 21:09 |
| *** egonzalez has quit IRC | 21:11 | |
| *** atoth has quit IRC | 21:25 | |
| *** hongbin has quit IRC | 23:12 | |
| *** limao has joined #openstack-kuryr | 23:34 | |
| *** limao has quit IRC | 23:40 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!