| *** guoshan has joined #senlin | 00:03 | |
| *** guoshan has quit IRC | 01:05 | |
| *** Drago5 has joined #senlin | 01:22 | |
| *** guoshan has joined #senlin | 01:24 | |
| *** guoshan has quit IRC | 01:31 | |
| *** guoshan has joined #senlin | 01:32 | |
| *** yanyanhu has joined #senlin | 01:34 | |
| openstackgerrit | Allen proposed openstack/senlin-dashboard: Add angular policies table https://review.openstack.org/389076 | 01:35 |
|---|---|---|
| *** zhurong has joined #senlin | 01:35 | |
| *** elynn has joined #senlin | 01:38 | |
| *** elynn has quit IRC | 01:43 | |
| *** elynn has joined #senlin | 01:44 | |
| openstackgerrit | Allen proposed openstack/senlin-dashboard: Add angular policies table https://review.openstack.org/389076 | 01:44 |
| ruijie | hi, yanyanhu :) | 01:52 |
| yanyanhu | hi, ruijie | 01:52 |
| elynn | morning :) | 01:53 |
| ruijie | morning :). I didn't see your message yesterday | 01:53 |
| yanyanhu | morning | 01:53 |
| yanyanhu | ruijie, it's ok :) | 01:54 |
| yanyanhu | just tried to get some update about batch policy. We can make it in next meeting or you can leave some information in etherpad :) | 01:54 |
| ruijie | okay. For now, the batch policy can support cluster_delete and cluster_update | 01:55 |
| ruijie | Just have some work for functional test | 01:55 |
| yanyanhu | ruijie, great. Any further issue, please ping us in the channel | 02:04 |
| ruijie | okay, yanyanhu :) | 02:06 |
| *** Drago5 has quit IRC | 02:14 | |
| yanyanhu | hi, ruijie, around? | 02:17 |
| ruijie | yes, yanyanhu | 02:18 |
| yanyanhu | I'm checking this patch, https://review.openstack.org/#/c/390731/4/senlin/engine/actions/node_action.py | 02:19 |
| yanyanhu | I noticed the default value of check_constraint is set to False | 02:19 |
| yanyanhu | I'm thinking maybe we should set it to True by default? | 02:19 |
| ruijie | :) | 02:20 |
| yanyanhu | since that is the logic of current implementation. And actually in most cases node join/leave is invoked individually, constraint check is needed | 02:20 |
| yanyanhu | only for case of node_replace, constraint check should be ignored | 02:20 |
| ruijie | Yes, yanyanhu. That is true. Most of the actions need that check | 02:21 |
| yanyanhu | And in current patch set, you actually change the condition of constraint check. That means the constraint check will be skipped in all node leave/join | 02:22 |
| Qiming | NODE_JOIN and NODE_LEAVE were triggered only by CLUSTER_ADD_NODES and CLUSTER_DEL_NODES | 02:23 |
| ruijie | https://review.openstack.org/#/c/390752/3 | 02:23 |
| Qiming | previously we have node_join and node_leave APIs which are already removed | 02:23 |
| ruijie | I also changed the parameter in cluster action layer | 02:23 |
| Qiming | for CLUSTER_ADD_NODES and CLUSTER_DEL_NODES, size check is already done before the action was executed | 02:24 |
| yanyanhu | ah, missed this patch | 02:24 |
| yanyanhu | if so, the logic is ok | 02:25 |
| Qiming | I'd suggest we completely remove the size check logic from that two methods | 02:25 |
| ruijie | Will the cluster be locked when one action being executed | 02:26 |
| Qiming | because ... in all occasions these two methods are invoked, the size check has already been done | 02:26 |
| yanyanhu | ruijie, yes, if action targets on that cluster or a node belongs to that cluster | 02:27 |
| Qiming | http://git.openstack.org/cgit/openstack/senlin/tree/senlin/engine/actions/cluster_action.py#n912 | 02:28 |
| Qiming | the fact is that node_join and node_leave are always derived actions, there is no such a strong need to size check again | 02:29 |
| yanyanhu | yes, for those two node actions are use invisible, when they are invoked, size checking should have been done in cluster action | 02:29 |
| Qiming | exactly | 02:29 |
| yanyanhu | maybe in later patch | 02:29 |
| Qiming | this patch is about adding a logic to control wether size check should be performed | 02:30 |
| Qiming | maybe a simpler, more direct way to fix it is to remove size check directly, instead of introducing some additional conditions to be removed in a future patch | 02:31 |
| Qiming | thoughts? | 02:31 |
| ruijie | I think remove the logic is better. | 02:32 |
| yanyanhu | Qiming, yes, that is ok. Just I found size checking is not done in cluster_add_node and cluster_del_node actions? | 02:32 |
| yanyanhu | may need to add it first | 02:32 |
| Qiming | they are done | 02:32 |
| Qiming | in engine service | 02:32 |
| yanyanhu | ah, saw it | 02:33 |
| yanyanhu | check_size_params is invoked with strict as True | 02:33 |
| yanyanhu | if so, it should be safe to remove that check directly | 02:34 |
| yanyanhu | hi, ruijie, if so, I think this patch is not needed https://review.openstack.org/#/c/390752/3 | 02:36 |
| yanyanhu | only need this one https://review.openstack.org/#/c/390731/ | 02:37 |
| ruijie | yes, yanyanhu. Will fix it soon. :) | 02:45 |
| yanyanhu | thanks :) | 02:45 |
| yanyanhu | hi, Qiming, around? | 02:54 |
| Qiming | yes | 02:55 |
| yanyanhu | since brian has made 0.9.9 release of sdk, should we propose patch to requirements to bump the upper constraint of sdk to that version? | 02:55 |
| yanyanhu | or the requirement version? | 02:56 |
| Qiming | not sure | 02:56 |
| yanyanhu | or usually, sdk team will propose the patch? | 02:56 |
| Qiming | you can check the gate log for your latest patch | 02:56 |
| Qiming | and see if 0.9.9 is already used | 02:56 |
| yanyanhu | I see | 02:56 |
| Qiming | if we have to use 0.9.9 | 02:56 |
| yanyanhu | will tigger a gate job first | 02:56 |
| Qiming | that may need a version bump request | 02:57 |
| yanyanhu | got it | 02:57 |
| Qiming | you may need to convince the requirements team why 0.9.9 is a must | 02:57 |
| yanyanhu | the current upper constraint is 0.9.8, so not sure whether 0.9.9 can be applied during the test | 02:57 |
| yanyanhu | I see | 02:57 |
| yanyanhu | will make a test first | 02:57 |
| openstackgerrit | Merged openstack/senlin: functional test for batch policy https://review.openstack.org/390234 | 03:04 |
| openstackgerrit | Merged openstack/senlin: Engine support for node_update2 https://review.openstack.org/391746 | 03:05 |
| openstackgerrit | Merged openstack/senlin: API support for node_update2 https://review.openstack.org/392052 | 03:05 |
| openstackgerrit | Merged openstack/senlin: Remove dead code about node update https://review.openstack.org/392055 | 03:05 |
| openstackgerrit | Merged openstack/senlin: Add Zaqar messaging client for tempest integration test https://review.openstack.org/389024 | 03:18 |
| openstackgerrit | Carl proposed openstack/python-senlinclient: Fix syntax errors https://review.openstack.org/392409 | 03:49 |
| openstackgerrit | Merged openstack/senlin: Add a bandit environment to tox https://review.openstack.org/391630 | 03:52 |
| openstackgerrit | Merged openstack/senlin: cluster delete action handler for batch policy https://review.openstack.org/391696 | 03:52 |
| openstackgerrit | Merged openstack/senlin: Use 'openstack' command-client to replace 'senlin receiver-create' https://review.openstack.org/390001 | 03:52 |
| *** guoshan has quit IRC | 04:03 | |
| openstackgerrit | Yanyan Hu proposed openstack/senlin: Integration test for message receiver https://review.openstack.org/389098 | 04:03 |
| *** yuanying has quit IRC | 04:24 | |
| *** elynn has quit IRC | 04:30 | |
| *** yuanying has joined #senlin | 04:36 | |
| *** elynn has joined #senlin | 04:58 | |
| *** elynn has quit IRC | 05:02 | |
| *** elynn has joined #senlin | 05:03 | |
| *** guoshan has joined #senlin | 05:03 | |
| *** guoshan has quit IRC | 05:07 | |
| *** elynn_ has joined #senlin | 05:22 | |
| *** elynn has quit IRC | 05:23 | |
| *** elynn__ has joined #senlin | 05:28 | |
| *** elynn_ has quit IRC | 05:30 | |
| ruijie | hi, for the size limitation check, I still have a question. | 05:45 |
| ruijie | We do size check in engine service, then put the action to queue, and then the cluster will be locked. | 05:46 |
| ruijie | If there are 2 action, cluster with max:2 desire:0, min:0. action1 add 2 nodes, and then action2 add other 2 nodes | 05:47 |
| ruijie | hi,yanyanhu, around | 05:59 |
| yanyanhu | hi, ruijie, just came back | 05:59 |
| yanyanhu | ruijie, good question, let me see | 06:00 |
| ruijie | Will that be safe if we remove the size check login in node action layer? | 06:00 |
| ruijie | logic | 06:00 |
| yanyanhu | there could be risk I think | 06:02 |
| yanyanhu | hi, Qiming, any thought about this issue? | 06:02 |
| *** guoshan has joined #senlin | 06:04 | |
| Qiming | yes, there would be | 06:07 |
| Qiming | there seem something to do with cluster locked | 06:07 |
| yanyanhu | yes, have the same feeling | 06:08 |
| *** guoshan has quit IRC | 06:09 | |
| yanyanhu | so the point is size check done inside action is safe and accurate | 06:11 |
| yanyanhu | Maybe we let ruijie finish the current patches as a workaround. Then we can think more here for a better solution. | 06:12 |
| *** guoshan has joined #senlin | 06:21 | |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Request object for cluster policy attach operation https://review.openstack.org/392422 | 06:22 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Engine layer cluster-policy-attach2 https://review.openstack.org/392427 | 06:32 |
| *** lvdongbing has joined #senlin | 06:40 | |
| *** yuanying has quit IRC | 06:46 | |
| *** yuanying has joined #senlin | 06:49 | |
| *** elynn_ has joined #senlin | 06:50 | |
| openstackgerrit | Yanyan Hu proposed openstack/senlin: Versioned req object and engine support for node delete https://review.openstack.org/392432 | 06:52 |
| Qiming | em ... what's the point for adding some code which will be removed? | 06:52 |
| *** elynn__ has quit IRC | 06:53 | |
| openstackgerrit | Yanyan Hu proposed openstack/senlin: Integration test for message receiver https://review.openstack.org/389098 | 06:54 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: API layer support for cluster-policy-attach2 https://review.openstack.org/392433 | 06:57 |
| ruijie | So, we can check the size limitation after acquire the cluster lock? | 07:01 |
| Qiming | yes, maybe we should | 07:02 |
| Qiming | I'd suggest we reaim the current patch to "move" the check from node_join/node_leave to cluster_actions | 07:03 |
| Qiming | since we all now have a better understanding of the problem to be solved and we have got a consensus on the way to solve it, the patch won't take long to implement or review | 07:04 |
| Qiming | just a suggestion, if you prefer get the current patch in and propose another patch to add cluster leve check, then another patch to remove checks in node_join/node_leave | 07:05 |
| Qiming | that is fine also | 07:05 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Remove dead engine code wrt cluster policy attach https://review.openstack.org/392435 | 07:07 |
| ruijie | Thanks, Qiming. Will propose another patch to add the check to cluster action layer to reduce risk | 07:08 |
| Qiming | okay | 07:09 |
| Qiming | you already have my +2 | 07:09 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Request object for policy update and detach https://review.openstack.org/392441 | 07:29 |
| openstackgerrit | Kenji Ishii proposed openstack/senlin-dashboard: Add filterFacets for angular profile table https://review.openstack.org/389544 | 07:39 |
| *** openstackgerrit has quit IRC | 07:48 | |
| *** openstackgerrit has joined #senlin | 07:49 | |
| *** ChanServ sets mode: +v openstackgerrit | 07:49 | |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Engine support for policy update/detach https://review.openstack.org/392449 | 07:52 |
| xuhaiwei | hi Qiming | 07:59 |
| xuhaiwei | hi yanyanhu, elynn_, If i create a cluster with heat stack type, and in the heat template i create a new subnet, then when scaling out this cluster, many new subnet will be created, can we avoid this? | 08:01 |
| xuhaiwei | actually we only want the nodes to be in a subnet | 08:02 |
| xuhaiwei | it seems the only way is pre-creating a subnet, and then attach nodes to it, right? | 08:03 |
| Qiming | a minute | 08:04 |
| yanyanhu | xuhaiwei, um, I think currently, we can't | 08:05 |
| Qiming | it looks doesn't like a senlin problem ... | 08:06 |
| yanyanhu | if we use the current heat stack profile | 08:06 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: API layer objectified policy detach and update https://review.openstack.org/392455 | 08:13 |
| *** yanyanhu has quit IRC | 08:13 | |
| Qiming | it actually depends on whether heat has a support like: | 08:15 |
| Qiming | create a port with this name, or reuse the specified port if it exists | 08:15 |
| Qiming | xuhaiwei, what is the requirement ? | 08:15 |
| Qiming | okay, I see, it is about subnet, not port | 08:16 |
| xuhaiwei | the requirement is user wants to create a new subnet with some nodes attached to it, and the nodes are scable | 08:17 |
| Qiming | if a template is about CREATING a subnet, when you are creating stacks out of that template, it will definitely try create multiple subnet | 08:17 |
| Qiming | then you will precreate that subnet | 08:18 |
| xuhaiwei | if user doesn't want to precreate the subnet? | 08:19 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Kill dead code wrt cluster policy detach/update https://review.openstack.org/392461 | 08:19 |
| Qiming | teach heat to do that for you then | 08:19 |
| xuhaiwei | if we use senlin heat resource, this problem can be avoided, create a subnet, then pass the resource to senlin network param | 08:20 |
| Qiming | you mean heat should be smart enough to make a decision? if a subnet resource is in a template, and that named subnet already exists when creating the stack, heat should just ignore it? | 08:20 |
| Qiming | okay, if you want to do the whole thing using heat, that's fine | 08:21 |
| openstackgerrit | miaohb proposed openstack/senlin: Add filters "policy_name" and "policy_type" for policy binding list https://review.openstack.org/388505 | 08:28 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: API layer objectified policy detach and update https://review.openstack.org/392455 | 08:58 |
| *** zhurong has quit IRC | 09:06 | |
| *** elynn_ has quit IRC | 09:40 | |
| *** lvdongbing has quit IRC | 09:44 | |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Kill dead code wrt cluster policy detach/update https://review.openstack.org/392461 | 09:46 |
| *** guoshan has quit IRC | 09:48 | |
| *** guoshan has joined #senlin | 09:49 | |
| openstackgerrit | RUIJIE YUAN proposed openstack/senlin: check size constraint in cluster action layer https://review.openstack.org/392512 | 09:54 |
| openstackgerrit | RUIJIE YUAN proposed openstack/senlin: remove dead code in node action layer https://review.openstack.org/390731 | 10:05 |
| *** guoshan has quit IRC | 10:28 | |
| *** guoshan has joined #senlin | 10:46 | |
| *** test_test has joined #senlin | 11:20 | |
| *** test_test has left #senlin | 11:20 | |
| *** XueFeng has joined #senlin | 11:21 | |
| *** XueFeng has left #senlin | 11:26 | |
| openstackgerrit | Qiming Teng proposed openstack/senlin: remove dead code in node action layer https://review.openstack.org/390731 | 11:50 |
| *** guoshan has quit IRC | 11:56 | |
| *** guoshan has joined #senlin | 12:08 | |
| openstackgerrit | Merged openstack/senlin: Request object for cluster policy attach operation https://review.openstack.org/392422 | 12:17 |
| openstackgerrit | Merged openstack/senlin: Engine layer cluster-policy-attach2 https://review.openstack.org/392427 | 12:25 |
| openstackgerrit | Merged openstack/senlin: API layer support for cluster-policy-attach2 https://review.openstack.org/392433 | 12:25 |
| openstackgerrit | Merged openstack/senlin: Remove dead engine code wrt cluster policy attach https://review.openstack.org/392435 | 12:26 |
| openstackgerrit | Merged openstack/senlin: Request object for policy update and detach https://review.openstack.org/392441 | 12:26 |
| openstackgerrit | Merged openstack/senlin: Engine support for policy update/detach https://review.openstack.org/392449 | 12:26 |
| openstackgerrit | Merged openstack/senlin: API layer objectified policy detach and update https://review.openstack.org/392455 | 12:27 |
| openstackgerrit | Merged openstack/senlin: Kill dead code wrt cluster policy detach/update https://review.openstack.org/392461 | 12:27 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Versioned objects for check/recover requests https://review.openstack.org/392640 | 13:04 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Engine support to objectified cluster check/recover https://review.openstack.org/392665 | 13:39 |
| *** Drago5 has joined #senlin | 13:51 | |
| *** guoshan has quit IRC | 14:00 | |
| *** Drago5 is now known as Drago | 14:04 | |
| *** guoshan has joined #senlin | 14:09 | |
| *** guoshan has quit IRC | 14:23 | |
| *** zzxwill has joined #senlin | 14:31 | |
| *** guoshan has joined #senlin | 14:48 | |
| *** guoshan has quit IRC | 14:59 | |
| *** guoshan has joined #senlin | 15:13 | |
| openstackgerrit | Qiming Teng proposed openstack/senlin: API layer support to cluster check/recover v2 https://review.openstack.org/392712 | 15:17 |
| *** guoshan has quit IRC | 15:23 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/python-senlinclient: Updated from global requirements https://review.openstack.org/392829 | 15:40 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/senlin: Updated from global requirements https://review.openstack.org/392834 | 15:41 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/senlin-dashboard: Updated from global requirements https://review.openstack.org/392835 | 15:41 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Tune health manager to use new RPC https://review.openstack.org/392843 | 15:44 |
| *** zzxwill_ has joined #senlin | 16:01 | |
| *** zzxwill has quit IRC | 16:04 | |
| *** zzxwill_ has quit IRC | 16:24 | |
| *** Drago has quit IRC | 16:48 | |
| *** Drago5 has joined #senlin | 16:53 | |
| *** Drago5 is now known as Drago | 16:54 | |
| *** openstackgerrit has quit IRC | 18:18 | |
| *** openstackgerrit has joined #senlin | 18:19 | |
| *** ChanServ sets mode: +v openstackgerrit | 18:19 | |
| *** yuanying has quit IRC | 19:03 | |
| *** openstack has joined #senlin | 20:02 | |
| *** openstackgerrit has quit IRC | 20:18 | |
| *** openstackgerrit has joined #senlin | 20:19 | |
| *** ChanServ sets mode: +v openstackgerrit | 20:19 | |
| *** Drago has quit IRC | 21:51 | |
| *** Drago5 has joined #senlin | 21:59 | |
| *** Drago5 is now known as Drago | 22:01 | |
| *** Drago has quit IRC | 22:12 | |
| *** Drago5 has joined #senlin | 22:20 | |
| *** Drago5 is now known as Drago | 22:28 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!