| *** dixiaoli has joined #senlin | 00:54 | |
| *** dixiaoli has quit IRC | 00:55 | |
| *** dixiaoli has joined #senlin | 00:55 | |
| *** elynn has joined #senlin | 02:06 | |
| *** yuanying has quit IRC | 02:51 | |
| openstackgerrit | Merged openstack/senlin-dashboard master: Updated from global requirements https://review.openstack.org/464490 | 03:13 |
|---|---|---|
| *** shu-mutou has joined #senlin | 03:24 | |
| *** yuanying has joined #senlin | 03:58 | |
| *** yuanying has quit IRC | 04:23 | |
| *** yuanying has joined #senlin | 04:23 | |
| *** yuanying has quit IRC | 05:18 | |
| *** yuanying has joined #senlin | 05:19 | |
| *** yuanying has quit IRC | 05:19 | |
| *** yuanying has joined #senlin | 05:20 | |
| *** yuanying has quit IRC | 06:06 | |
| openstackgerrit | Qiming Teng proposed openstack/senlin master: Change requirement assertion to 'must' https://review.openstack.org/464943 | 06:38 |
| *** yuanying has joined #senlin | 06:59 | |
| *** dixiaoli has quit IRC | 08:04 | |
| openstackgerrit | Qiming Teng proposed openstack/senlin master: Change requirement assertion to 'must' https://review.openstack.org/464943 | 08:26 |
| *** dixiaoli has joined #senlin | 08:28 | |
| *** chenyb has joined #senlin | 08:31 | |
| openstackgerrit | Qiming Teng proposed openstack/senlin master: Trivial: Use correct words in inequality assertions https://review.openstack.org/464976 | 08:36 |
| *** shu-mutou is now known as shu-mutou-AWAY | 08:44 | |
| *** dixiaoli has quit IRC | 09:15 | |
| *** dixiaoli has joined #senlin | 09:16 | |
| *** Guest67875 has joined #senlin | 09:33 | |
| openstackgerrit | Qiming Teng proposed openstack/senlin master: Add NodeAdoptPreviewRequest https://review.openstack.org/465000 | 09:33 |
| *** Guest67875 is now known as XueFeng | 09:36 | |
| *** elynn has quit IRC | 09:46 | |
| *** elynn has joined #senlin | 09:47 | |
| *** elynn has quit IRC | 09:47 | |
| *** elynn has joined #senlin | 09:48 | |
| *** XueFeng has quit IRC | 09:51 | |
| *** XueFeng has joined #senlin | 09:53 | |
| *** elynn has quit IRC | 10:05 | |
| *** dixiaoli has quit IRC | 10:14 | |
| *** chenyb has quit IRC | 10:16 | |
| *** chenyb has joined #senlin | 10:33 | |
| *** chenyb has quit IRC | 11:27 | |
| *** chenyb has joined #senlin | 12:09 | |
| *** chenyb has quit IRC | 12:54 | |
| *** elynn has joined #senlin | 12:56 | |
| Qiming | hi | 12:59 |
| Qiming | not sure we will have a weekly meeting today | 12:59 |
| Qiming | since most of us have just returned from a long week of summit activities | 13:00 |
| Qiming | elynn, haiwei, XueFeng ? | 13:00 |
| *** chenyb has joined #senlin | 13:01 | |
| elynn | I have nothing to update. | 13:01 |
| Qiming | okay | 13:02 |
| Qiming | two things from my side | 13:02 |
| *** Ruijie_ has joined #senlin | 13:02 | |
| Qiming | improving node adoption, just proposed a patch, separating the preview request from adopt request | 13:02 |
| elynn | Saw your email that this meeting was cancelled, if you have sth we could attend this meeting. | 13:03 |
| Qiming | another thing is trying to fix py35 gate jobs: https://review.openstack.org/464568 | 13:04 |
| Qiming | other than those two items, I don't have a lot to update | 13:04 |
| Ruijie_ | https://review.openstack.org/#/c/463276/ please help review if you have time :) | 13:05 |
| Ruijie_ | Didn't find a better way other than this one to do health check when processing scaling actions. | 13:06 |
| Qiming | ruijie, I checked that patch a few days ago | 13:06 |
| Qiming | it looks ... ugly | 13:07 |
| Qiming | I was thinking about taking a step back | 13:07 |
| Qiming | what is the usage scenario? | 13:07 |
| Ruijie_ | yes Qiming, a little bit .. for the current model .. | 13:08 |
| Qiming | right | 13:08 |
| *** XueFengLiu has joined #senlin | 13:08 | |
| XueFengLiu | hi | 13:08 |
| Qiming | so .. the requirement is this, (trying to sort it out), when you are scaling in a cluster | 13:09 |
| Ruijie_ | the decision should be made excatly after scaling policy and before deletion and other policies | 13:09 |
| Qiming | you want to make sure the nodes' health status updated before removing nodes | 13:09 |
| Qiming | no matter how many nodes you want to remove eventually, the need is about making sure senlin knows the exact state of each node, right? | 13:10 |
| Ruijie_ | yes Qiming | 13:10 |
| Qiming | then how about this ... | 13:10 |
| Qiming | add an optional parameter to the scale in api | 13:11 |
| Qiming | do health check if 'health_check' is specified, before triggering the cluster scale in action | 13:11 |
| Ruijie_ | Qiming, you mean in engine.cluster_action.execute()/_execute() or even more ahead ? | 13:13 |
| Qiming | before triggering that action | 13:13 |
| Qiming | I cannot envision a case where health check operation itself needs a policy, so it can be thrown out of the action execution logic | 13:14 |
| Ruijie_ | https://review.openstack.org/gitweb?p=openstack/senlin.git;a=blob;f=senlin/engine/scheduler.py;h=6f2d0e1dea042fda077b8726b915864490107420;hb=356269980b5b1659a54a3c5b6035f4906c29deb4#l87 | 13:15 |
| Qiming | how about here: http://git.openstack.org/cgit/openstack/senlin/tree/senlin/engine/service.py#n1344 | 13:17 |
| Qiming | you know what | 13:18 |
| Qiming | that brings me to another thought | 13:18 |
| Qiming | maybe we should leave this to users | 13:18 |
| Qiming | if they want to do a health check before scaling in a cluster | 13:18 |
| Qiming | they should do cluster_check call before cluster_scale_in call | 13:18 |
| Qiming | two calls | 13:19 |
| Qiming | I have seen another patch that adds health check to the cluster_recover api, which is ... not consistent with our current design | 13:20 |
| Ruijie_ | actually, that makes sense ... | 13:20 |
| Qiming | merging two calls into one will always create a lot headaches | 13:20 |
| openstackgerrit | Qiming Teng proposed openstack/senlin master: Add NodeAdoptPreviewRequest https://review.openstack.org/465000 | 13:23 |
| Qiming | we cannot, by any means, ensure that nodes that were active remain active between the time it was checked and the time cluster is scaled in | 13:23 |
| Qiming | after health check, a node can still fail | 13:24 |
| Ruijie_ | after all, we want to provide primitive functions but not the services | 13:24 |
| Qiming | right, that is another way to put it | 13:24 |
| Ruijie_ | I mean to be atomic | 13:25 |
| Qiming | if we can find a way to solve a user problem easily, we can do it | 13:25 |
| Qiming | for the scaling operation in an unstable environment, we can promise nothing | 13:25 |
| Ruijie_ | okay Qiming, that makes sense | 13:26 |
| Qiming | I have seen you trying this in many ways, none of them makes us feel comfortable, but still, many thanks | 13:27 |
| Ruijie_ | so the work will be: 1. improve cluster_check (lb faiure need to be handled), 2. remove cluster_check login in cluster_recover, 3. improve cluster_recover to check desired_capacity | 13:27 |
| Ruijie_ | np Qiming, that's my job :) | 13:28 |
| Qiming | item 2 came from a patch | 13:28 |
| Ruijie_ | yes Qiming, I sae that "check" parameter | 13:29 |
| Ruijie_ | saw | 13:29 |
| Qiming | this one: https://review.openstack.org/#/c/462802/ | 13:29 |
| Ruijie_ | that may change the right value to wrong, like the NS_WARNING which was set by lb_policy | 13:30 |
| Qiming | I didn't think about it for long time | 13:31 |
| Qiming | it is acceptable, but ... I'm a little bit concerned about the direction we are going | 13:31 |
| *** elynn has quit IRC | 13:36 | |
| *** XueFengLiu has quit IRC | 13:38 | |
| *** XueFengLiu has joined #senlin | 13:38 | |
| chenyb | https://review.openstack.org/#/c/464027/ please help review if you have time :) | 13:59 |
| *** chenyb has quit IRC | 14:23 | |
| *** chenyb has joined #senlin | 14:23 | |
| *** chenyb has quit IRC | 14:27 | |
| *** dixiaoli has joined #senlin | 14:28 | |
| *** dixiaoli has quit IRC | 14:28 | |
| *** dixiaoli has joined #senlin | 14:29 | |
| *** XueFengLiu has quit IRC | 14:30 | |
| *** XueFengLiu has joined #senlin | 14:30 | |
| *** dixiaoli has quit IRC | 14:36 | |
| *** Ruijie_ has quit IRC | 15:17 | |
| *** XueFengLiu has quit IRC | 16:15 | |
| *** openstackgerrit has quit IRC | 18:17 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!