| *** jdandrea has quit IRC | 00:46 | |
| *** lawrancejing has joined #senlin | 01:34 | |
| *** Qiming has joined #senlin | 01:41 | |
| openstackgerrit | xu-haiwei proposed stackforge/senlin: Make deletion policy work for cluster resize action https://review.openstack.org/231885 | 02:05 |
|---|---|---|
| openstackgerrit | xu-haiwei proposed stackforge/senlin: Make deletion policy work for cluster resize action https://review.openstack.org/231885 | 02:14 |
| openstackgerrit | Merged stackforge/senlin-dashboard: Hide metadata while metadata dict is empty https://review.openstack.org/232596 | 02:31 |
| openstackgerrit | Yanyan Hu proposed stackforge/senlin: Add a TODO item about document https://review.openstack.org/231490 | 03:09 |
| openstackgerrit | Merged stackforge/senlin: Make deletion policy work for cluster resize action https://review.openstack.org/231885 | 03:12 |
| openstackgerrit | Merged stackforge/senlin: Fix alarm rule parsing behavior https://review.openstack.org/228802 | 03:16 |
| *** Qiming_ has joined #senlin | 03:27 | |
| *** Qiming has quit IRC | 03:27 | |
| *** Qiming_ is now known as Qiming | 03:39 | |
| openstackgerrit | lawrancejing proposed stackforge/senlin-dashboard: Add cluster create action https://review.openstack.org/232575 | 03:54 |
| openstackgerrit | Merged stackforge/senlin: Add a TODO item about document https://review.openstack.org/231490 | 04:07 |
| *** mathspanda has joined #senlin | 04:23 | |
| openstackgerrit | Merged stackforge/senlin-dashboard: Add cluster create api https://review.openstack.org/232487 | 04:28 |
| openstackgerrit | Merged stackforge/senlin-dashboard: Add cluster create action https://review.openstack.org/232575 | 04:29 |
| *** Zhenqi has joined #senlin | 04:37 | |
| openstackgerrit | Merged stackforge/senlin: Revise node index assignment https://review.openstack.org/231317 | 04:39 |
| *** mathspanda has quit IRC | 05:17 | |
| *** Zhenqi has quit IRC | 05:48 | |
| openstackgerrit | Merged stackforge/senlin: Updated from global requirements https://review.openstack.org/232908 | 06:15 |
| *** gongysh has joined #senlin | 06:54 | |
| *** xuhaiwei has quit IRC | 07:09 | |
| *** xuhaiwei has joined #senlin | 07:10 | |
| openstackgerrit | lawrancejing proposed stackforge/senlin-dashboard: Add node create api https://review.openstack.org/232939 | 07:48 |
| *** Shijia has quit IRC | 07:49 | |
| *** Shijia has joined #senlin | 07:53 | |
| xuhaiwei | Qiming | 08:03 |
| Qiming | hi | 08:04 |
| xuhaiwei | still about deletion policy | 08:04 |
| xuhaiwei | I found cluster_del_nodes action just remove specified node out of cluster, so deletion policy should not take effect? | 08:05 |
| Qiming | it doesn't hurt if the policy is implemented correctly | 08:05 |
| Qiming | in previous comments I mentioned that some of the code in the deletion policy can be optimized | 08:07 |
| Qiming | I was referring to the case when CLUSTER_DEL_NODES is the action, we don't need to compute candidates | 08:08 |
| xuhaiwei | I think the deletion policy's TARGET should not have CLUSTER_DEL_NODES | 08:08 |
| Qiming | fine | 08:08 |
| Qiming | that makes sense too | 08:09 |
| xuhaiwei | we just dont need to check the pre_op at all | 08:09 |
| Qiming | it the TARGET doesn't include CLUSTER_DEL_NODES, the policy won't be effected | 08:10 |
| xuhaiwei | yes | 08:10 |
| xuhaiwei | should it be? | 08:10 |
| Qiming | I'm okay with removing CLUSTER_DEL_NODES from target | 08:10 |
| xuhaiwei | ok | 08:11 |
| Qiming | if you are looking into the deletion policy (at least it seems so) | 08:11 |
| Qiming | you may want to think of other possibilities | 08:11 |
| xuhaiwei | yes? | 08:12 |
| Qiming | when doing SCALE_IN or RESIZE, we may want to remove nodes that are not ACTIVE first | 08:12 |
| xuhaiwei | currently deleting ERROR nodes first | 08:13 |
| Qiming | right | 08:13 |
| Qiming | that logic is not incorrect | 08:13 |
| Qiming | we may want to remove nodes that are still being created | 08:14 |
| xuhaiwei | you mean the deletion policy should be improved? | 08:14 |
| Qiming | yes | 08:15 |
| Qiming | if you are looking into that policy, there are something not yet done | 08:15 |
| xuhaiwei | ok, I will make a change | 08:15 |
| Qiming | another aspect that can be improved is about grace period | 08:15 |
| Qiming | and reduce_desired_capacity | 08:16 |
| Qiming | those two parameters are not handled | 08:16 |
| Qiming | grace period is an interval during which you will not delete the node | 08:17 |
| Qiming | you want to grant the app/service running in those nodes some extra time to complete their on-going services | 08:17 |
| xuhaiwei | yes | 08:18 |
| xuhaiwei | I will think about this too | 08:19 |
| Qiming | cool | 08:19 |
| xuhaiwei | by the way, we will have a three days vocation at this weekend | 08:20 |
| Qiming | that means only one day off, ;) | 08:20 |
| xuhaiwei | yes, one day more | 08:22 |
| Qiming | good for you | 08:22 |
| Qiming | we have 10+ days off | 08:22 |
| Qiming | still working from at home now | 08:22 |
| xuhaiwei | that is IBM's special vocation? | 08:23 |
| Qiming | no, it's China | 08:24 |
| Qiming | public holiday | 08:24 |
| Qiming | plus IBM's forced annual leave | 08:24 |
| xuhaiwei | IBM is good | 08:24 |
| Qiming | still, not everyone is happy about being forced to spend their annual leaves, ;) | 08:25 |
| xuhaiwei | why not take it freely? | 08:26 |
| Qiming | it's company's policy | 08:26 |
| Qiming | they "help" you to balance work and life | 08:26 |
| xuhaiwei | IBM is not so good | 08:26 |
| Qiming | lol | 08:26 |
| xuhaiwei | anyway, you have vocation | 08:27 |
| Qiming | yep, it's fun, watching how people get jammed on road, pressed by crowds, cheated/threatened by restaurants | 08:29 |
| xuhaiwei | it really needs time to improve this | 08:30 |
| Qiming | it's safer to stay at home | 08:31 |
| Qiming | though boring | 08:31 |
| xuhaiwei | yap | 08:31 |
| openstackgerrit | xu-haiwei proposed stackforge/senlin: Remove CLUSTER_DEL_NODES from deletion policy's TARGET https://review.openstack.org/232953 | 08:40 |
| Qiming | xuhaiwei, just saw your patch | 08:48 |
| Qiming | I'm wondering if we are doing the right thing | 08:49 |
| Qiming | maybe we can give it a second thought | 08:50 |
| Qiming | still there? | 08:50 |
| xuhaiwei | yes | 08:51 |
| Qiming | probably okay, probably not | 08:52 |
| xuhaiwei | I just dont understand why you think cluster_del_nodes need deletion policy at the design state | 08:52 |
| Qiming | we are not only choosing candidates for deletion in the policy | 08:52 |
| xuhaiwei | yes? | 08:52 |
| Qiming | the policy also outputs or even enforce 'grace_period' | 08:53 |
| Qiming | so I'm thinking whether these values are relevant to CLUSTER_DEL_NODES use case | 08:54 |
| xuhaiwei | it makes a little sense to me now | 08:54 |
| xuhaiwei | for outputs, should we do it in the action? | 08:54 |
| Qiming | yes, a "tiny" sense to me | 08:54 |
| lawrancejing | Qiming: where could get the node/cluster/profile status list (active, creating....) | 08:56 |
| Qiming | node-list, cluster-list will show their statuses? | 08:57 |
| xuhaiwei | for grace_period, because CLUSTER_DEL_NODES just remove nodes, but not delete them, there should be ok I think | 08:57 |
| Qiming | lawrancejing, also node-show, cluster-show | 08:57 |
| Qiming | xuhaiwei, think from a user's perspective | 08:57 |
| Qiming | when you are typing (for example): senlin cluster-del-nodes -n n1 -n n2 cluster1 | 08:58 |
| Qiming | what would you expect? | 08:58 |
| xuhaiwei | remove these two nodes from cluster1, maybe add them to other clusters | 08:59 |
| xuhaiwei | or just delete them | 08:59 |
| Qiming | it is kind of a semantic dilemma, right? | 08:59 |
| Qiming | but if we provide options like this: senlin cluster-del-nodes --destroy-after-delete -n n1 -n n2 cluster1 | 09:00 |
| Qiming | there won't be any ambiguity | 09:01 |
| xuhaiwei | yes | 09:01 |
| Qiming | sounds a better option to me | 09:02 |
| xuhaiwei | what opinion? | 09:03 |
| Qiming | provide a more flexible command interface (also API interface) for deleting nodes from cluster | 09:03 |
| xuhaiwei | add a destroy_after_deletion option? | 09:05 |
| Qiming | yes | 09:05 |
| lawrancejing | node-show could show only one status of node..i want to get the all status and distinguish loading and end-state status so that the node table row could be loading status(yellow) while the node is loading status(creating,deleting) | 09:05 |
| Qiming | lawrancejing, node-list is not giving you the status? | 09:06 |
| lawrancejing | Qiming, it does, but i wanna to get all status choices of node. | 09:08 |
| Qiming | oh I see | 09:08 |
| xuhaiwei | just notice 'destroy_after_deletion' is also in the deletion policy, this is especially for cluster_del_nodes action I think | 09:08 |
| Qiming | http://git.openstack.org/cgit/stackforge/senlin/tree/senlin/engine/node.py#n37 | 09:08 |
| Qiming | okay, they are orthogonal | 09:09 |
| lawrancejing | OK, got that | 09:09 |
| lawrancejing | thanks | 09:10 |
| Qiming | lawrancejing, you are terrific! | 09:10 |
| xuhaiwei | yes, maybe remove it from deletion policy? | 09:10 |
| Qiming | xuhaiwei, on one hand, we can enhance the api/cli to add this option | 09:10 |
| Qiming | on the other hand, we cannot force users to specify everything from api calls or commands | 09:11 |
| lawrancejing | ::::) | 09:11 |
| xuhaiwei | you mean remain both? | 09:11 |
| Qiming | so ... the option 'destroy_after_deletion' and 'grace_period' still useful | 09:11 |
| Qiming | yep | 09:11 |
| Qiming | maybe we should keep CLUSTER_DEL_NODES one of the TARGET in deletion policy? | 09:14 |
| Qiming | it can be treated as a special case of CLUSTER_SCALE_IN | 09:14 |
| Qiming | it is special because it know very well which nodes are to be removed from the cluster | 09:15 |
| xuhaiwei | yes | 09:15 |
| xuhaiwei | that means only some policies of deletion policy will take effect | 09:16 |
| Qiming | we really need more discussions of this kind | 09:16 |
| Qiming | yep | 09:16 |
| xuhaiwei | for most users, say deletion policy, only the deletion order comes to their minds | 09:16 |
| xuhaiwei | like me | 09:17 |
| Qiming | yes | 09:17 |
| Qiming | we added two more options in the policy, ;) | 09:17 |
| Qiming | and so far we believe they are useful, if implemented correctly | 09:18 |
| xuhaiwei | just leave the patch there, I will try to fix grace_period and reduce_desired_capacity | 09:18 |
| Qiming | ok | 09:19 |
| xuhaiwei | I will go home now, see u | 09:19 |
| Qiming | when those features are implemented | 09:19 |
| Qiming | we can gain a better understanding how things inter-related | 09:19 |
| Qiming | see you man! | 09:19 |
| openstackgerrit | lawrancejing proposed stackforge/senlin-dashboard: Add node creation action https://review.openstack.org/232979 | 09:38 |
| openstackgerrit | lawrancejing proposed stackforge/senlin-dashboard: Add cluster delete action https://review.openstack.org/232980 | 09:39 |
| *** lawrancejing has quit IRC | 09:43 | |
| *** lawrancejing has joined #senlin | 10:02 | |
| *** zhenguo has quit IRC | 10:03 | |
| openstackgerrit | lawrancejing proposed stackforge/senlin-dashboard: Add some help text for profle create form https://review.openstack.org/232985 | 10:04 |
| openstackgerrit | Merged stackforge/senlin-dashboard: Add cluster delete action https://review.openstack.org/232980 | 10:11 |
| *** mathspanda has joined #senlin | 10:12 | |
| *** mathspan_ has joined #senlin | 10:13 | |
| *** mathspan_ has quit IRC | 10:13 | |
| *** mathspan_ has joined #senlin | 10:13 | |
| *** mathspan_ has quit IRC | 10:13 | |
| *** mathspan_ has joined #senlin | 10:13 | |
| *** mathspanda has quit IRC | 10:17 | |
| openstackgerrit | lawrancejing proposed stackforge/senlin-dashboard: Add node creation action https://review.openstack.org/232979 | 10:21 |
| *** lawrancejing has quit IRC | 10:23 | |
| *** mathspan_ has quit IRC | 10:30 | |
| *** mathspanda has joined #senlin | 10:31 | |
| *** mathspan_ has joined #senlin | 10:37 | |
| *** mathspanda has quit IRC | 10:37 | |
| *** mathspan_ has quit IRC | 10:46 | |
| *** mathspanda has joined #senlin | 10:47 | |
| *** mathspan_ has joined #senlin | 11:40 | |
| *** mathspanda has quit IRC | 11:44 | |
| *** mathspan_ has quit IRC | 12:31 | |
| *** mathspanda has joined #senlin | 12:31 | |
| *** mathspan_ has joined #senlin | 12:43 | |
| *** mathspanda has quit IRC | 12:46 | |
| *** lawrancejing has joined #senlin | 13:02 | |
| *** mathspan_ has quit IRC | 13:05 | |
| *** mathspanda has joined #senlin | 13:05 | |
| openstackgerrit | lawrancejing proposed stackforge/senlin-dashboard: Add some help text for profle create form https://review.openstack.org/232985 | 13:07 |
| *** mathspan_ has joined #senlin | 13:54 | |
| *** mathspanda has quit IRC | 13:57 | |
| *** Shijia has quit IRC | 14:23 | |
| *** Shijia has joined #senlin | 14:23 | |
| *** Qiming has quit IRC | 14:26 | |
| *** Shijia has quit IRC | 14:32 | |
| *** mathspan_ has quit IRC | 15:10 | |
| openstackgerrit | lawrancejing proposed stackforge/senlin-dashboard: Detemine the cluster row status base on cluster status via ajax https://review.openstack.org/233113 | 15:11 |
| *** Shijia has joined #senlin | 15:31 | |
| *** Shijia has quit IRC | 15:37 | |
| openstackgerrit | Merged stackforge/senlin-dashboard: Add node create api https://review.openstack.org/232939 | 15:49 |
| *** lawrancejing has quit IRC | 16:11 | |
| *** gongysh has quit IRC | 17:09 | |
| *** gongysh has joined #senlin | 17:09 | |
| *** Shijia has joined #senlin | 17:42 | |
| *** Shijia has quit IRC | 17:47 | |
| *** Shijia has joined #senlin | 18:46 | |
| *** Shijia has quit IRC | 18:51 | |
| *** Shijia has joined #senlin | 19:52 | |
| *** Shijia has quit IRC | 19:56 | |
| *** Shijia has joined #senlin | 20:56 | |
| *** Shijia has quit IRC | 21:02 | |
| *** Shijia has joined #senlin | 22:01 | |
| *** Shijia has quit IRC | 22:06 | |
| *** Shijia has joined #senlin | 23:06 | |
| *** Shijia has quit IRC | 23:12 | |
| *** Qiming has joined #senlin | 23:42 | |
| openstackgerrit | Merged stackforge/python-senlinclient: Updated from global requirements https://review.openstack.org/232907 | 23:51 |
| *** Shijia has joined #senlin | 23:58 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!