| *** Qiming has quit IRC | 00:16 | |
| *** bdrich has quit IRC | 00:44 | |
| *** Qiming has joined #senlin | 01:21 | |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Implement do_check method for nova profile https://review.openstack.org/261990 | 01:50 |
|---|---|---|
| *** elynn has joined #senlin | 01:58 | |
| *** Qiming_ has joined #senlin | 02:00 | |
| *** Qiming has quit IRC | 02:00 | |
| *** pratikmallya has quit IRC | 02:04 | |
| elynn | Hi Qiming_ | 02:09 |
| Qiming_ | yes | 02:09 |
| elynn | Saw your comment at https://review.openstack.org/#/c/261971/ | 02:09 |
| Qiming_ | right | 02:10 |
| Qiming_ | location is not a cluster property | 02:10 |
| Qiming_ | the usage of location should be transparent to client | 02:10 |
| elynn | If don't use location, what do you suggest? | 02:11 |
| elynn | I notice that resize will return action | 02:12 |
| elynn | and client can receive that. | 02:12 |
| elynn | https://github.com/openstack/python-openstacksdk/blob/master/openstack/cluster/v1/cluster.py#L79 | 02:13 |
| Qiming_ | receiving an action is not a good practice either | 02:13 |
| elynn | e... | 02:13 |
| Qiming_ | say we have cluster_resize api, exposed from openstack/cluster/v1/_proxy.py | 02:14 |
| Qiming_ | we may want to return action from that api | 02:14 |
| Qiming_ | so client has no idea of 'location', where it comes from | 02:15 |
| Qiming_ | returning action to client at this layer doesn't break any assumption about REST APIs | 02:15 |
| Qiming_ | the key point, from client's perspective, is about sync/async calls | 02:16 |
| Qiming_ | that is the problem we are solving | 02:16 |
| Qiming_ | I'm thinking if we should do this: cluster_resize(async=True, **params) | 02:16 |
| Qiming_ | if async is False, we do action status polling inside that function | 02:17 |
| Qiming_ | still thinking | 02:17 |
| elynn | I'm not sure... | 02:18 |
| elynn | We put it in senlinclient or openstacksdk? | 02:20 |
| Qiming_ | for better control, put it in senlinclient | 02:37 |
| Qiming_ | for better generality, put it in openstacksdk | 02:37 |
| Qiming_ | it's a tradeoff | 02:37 |
| openstackgerrit | Ethan Lynn proposed openstack/senlin: Add new cluster status RESIZING https://review.openstack.org/261934 | 02:46 |
| *** gongysh has quit IRC | 03:18 | |
| *** gongysh has joined #senlin | 03:35 | |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Implement do_check method for nova profile https://review.openstack.org/261990 | 04:34 |
| *** Qiming_ has quit IRC | 04:38 | |
| *** pratikmallya has joined #senlin | 06:00 | |
| *** Qiming has joined #senlin | 06:18 | |
| openstackgerrit | Qiming Teng proposed openstack/python-senlinclient: Fix formatters used for listing actions https://review.openstack.org/262138 | 06:49 |
| *** Qiming_ has joined #senlin | 07:03 | |
| *** Qiming has quit IRC | 07:06 | |
| *** pratikmallya has quit IRC | 07:20 | |
| openstackgerrit | Ethan Lynn proposed openstack/senlin: Re-enable lock breaker https://review.openstack.org/262151 | 08:13 |
| *** gongysh has quit IRC | 08:23 | |
| openstackgerrit | Qiming Teng proposed openstack/python-senlinclient: Fix formatters used for listing actions https://review.openstack.org/262138 | 08:27 |
| *** gongysh has joined #senlin | 08:36 | |
| *** Yanyanhu has joined #senlin | 09:00 | |
| *** branw has joined #senlin | 09:16 | |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Allow action list to be filtered by 'status'. https://review.openstack.org/262169 | 09:40 |
| *** Qiming_ has quit IRC | 09:55 | |
| *** Qiming has joined #senlin | 09:55 | |
| openstackgerrit | Ethan Lynn proposed openstack/senlin: Catch ProfileNotFound exception when load_runtime_data in Node https://review.openstack.org/262178 | 10:05 |
| Qiming | git review is failing? | 10:07 |
| Qiming | I'm getting authentication failure now when doing 'git review' | 10:07 |
| elynn | Seems so | 10:08 |
| Yanyanhu | let me try | 10:08 |
| elynn | Very slow | 10:08 |
| Yanyanhu | I got 502 error | 10:08 |
| Qiming | sigh | 10:08 |
| Qiming | going home | 10:08 |
| elynn | Maybe it will recover later, I met once today and later it will come back. | 10:08 |
| Qiming | not a time for working | 10:08 |
| Qiming | it's from Great Wall I think | 10:09 |
| Qiming | sigh | 10:09 |
| Yanyanhu | guess so... | 10:09 |
| elynn | maybe... | 10:09 |
| elynn | But time to go home :) | 10:10 |
| *** elynn has quit IRC | 10:10 | |
| *** Qiming has quit IRC | 10:13 | |
| *** Yanyanhu has quit IRC | 10:34 | |
| *** Qiming has joined #senlin | 11:04 | |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Add user doc for actions https://review.openstack.org/262189 | 11:08 |
| *** Liuqing has joined #senlin | 11:27 | |
| *** Liuqing has quit IRC | 12:01 | |
| *** Liuqing has joined #senlin | 12:49 | |
| *** elynn has joined #senlin | 12:50 | |
| *** yanyanhu has joined #senlin | 12:53 | |
| Qiming | meeting time | 13:01 |
| Qiming | #openstack-meeting channel | 13:01 |
| *** pratikmallya has joined #senlin | 13:38 | |
| *** lixinhui_ has joined #senlin | 13:52 | |
| *** haiwei has joined #senlin | 13:59 | |
| Qiming | thanks for joining | 14:00 |
| yanyanhu | no problem | 14:00 |
| elynn | Anyway, add a response_headers to resource property is a good idea. | 14:00 |
| Qiming | I'd still prefer doing wait inside sdk | 14:01 |
| elynn | If I can get action from response_headers, then I won't need a wait function I guess. | 14:01 |
| Qiming | if that is becoming a blocker for heat | 14:01 |
| elynn | We can do it, but I need to figure out how to use it in heat. | 14:01 |
| Qiming | yep | 14:01 |
| Qiming | you will have the Cluster object directly | 14:02 |
| Qiming | that object has a response_headers for you to check | 14:02 |
| Qiming | do your own sleep in heat, possibly with a yield | 14:02 |
| *** yanyanhu has quit IRC | 14:02 | |
| elynn | Yes | 14:03 |
| elynn | You still need to discuss it with sdk guys? | 14:03 |
| Qiming | let's talk via code reviews | 14:03 |
| elynn | ok, I will prepare codes tomorrow. | 14:04 |
| Qiming | seldom can we show up on IRC at the same time | 14:04 |
| Qiming | cool | 14:04 |
| elynn | And I might still need this patch https://review.openstack.org/261970 | 14:05 |
| elynn | to expose headers. | 14:05 |
| Liuqing | Qiming: while the `health management` implemented and we use the nova_server.yaml, senlin will have the recreate-after-delete logic? | 14:07 |
| elynn | Will take a shower, ttyl. | 14:07 |
| Qiming | Liuqing, yes | 14:07 |
| Liuqing | ok | 14:07 |
| Liuqing | automate or we should call some api? | 14:08 |
| *** lixinhui_ has quit IRC | 14:13 | |
| Qiming | Liuqing, via config file | 14:15 |
| Liuqing | Qiming: not very clear, which config option will handle it ? | 14:16 |
| Qiming | an option to be added | 14:17 |
| Liuqing | i just thought the option should in our profile.. | 14:17 |
| Liuqing | in config is fine | 14:17 |
| Qiming | health check will be don automatically | 14:17 |
| Qiming | but recovery has to be configured | 14:18 |
| Qiming | we are thinking of a global config that do_recover() will be invoked automatically when node failures are detected | 14:18 |
| Qiming | a more finer granularity of control should be done via a health policy attached to your cluster | 14:19 |
| Liuqing | yes, granularity is needed. | 14:20 |
| Qiming | ya | 14:21 |
| Qiming | not all clusters are born equal | 14:21 |
| Qiming | some are paid, :) | 14:21 |
| Liuqing | yes, so the granularity will in profile rather than config? | 14:22 |
| Qiming | not in profile | 14:22 |
| Qiming | it will be about a cluster | 14:23 |
| Qiming | still building the foundations at this stage | 14:24 |
| Qiming | will make tradeoffs about configuration options and policy specs | 14:24 |
| Liuqing | ok | 14:24 |
| Liuqing | Qiming: so we will have some periodic task for do_check() in somewhere? | 14:29 |
| Qiming | yes, that is the health_manager | 14:29 |
| *** elynn has quit IRC | 14:33 | |
| haiwei | Qiming, in Tokyo summit we are agreed to not to do health check automatically, it is changed? | 14:35 |
| Liuqing | hi haiwei, if we have time, we want to recycle rocket hahaha.. | 14:37 |
| Qiming | haiwei, we don't do monitoring | 14:38 |
| haiwei | what do you mean by 'recycle rocket' | 14:38 |
| Qiming | we have removed all trigger related code | 14:39 |
| Liuqing | haiwei, a joke. | 14:40 |
| Qiming | we have do something to make sure we know the real status of the objects we created, especially nodes | 14:40 |
| haiwei | so the node's state will be updated periodically ? | 14:41 |
| Qiming | or else, we cannot do a good job regarding auto-scaling | 14:41 |
| haiwei | ok | 14:42 |
| *** Liuqing has quit IRC | 14:43 | |
| *** Liuqing has joined #senlin | 15:02 | |
| *** haiwei has quit IRC | 15:22 | |
| *** Qiming has quit IRC | 15:33 | |
| *** Liuqing has quit IRC | 16:48 | |
| *** openstackgerrit has quit IRC | 21:17 | |
| *** openstackgerrit has joined #senlin | 21:18 | |
| *** pratikmallya has quit IRC | 21:55 | |
| *** pratikmallya has joined #senlin | 22:56 | |
| *** pratikmallya has quit IRC | 23:00 | |
| *** pratikmallya has joined #senlin | 23:34 | |
| *** Qiming has joined #senlin | 23:35 | |
| *** pratikmallya has quit IRC | 23:43 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!