| *** Qiming has quit IRC | 00:17 | |
| *** pratikmallya has joined #senlin | 00:35 | |
| *** Qiming has joined #senlin | 00:51 | |
| *** Liuqing has joined #senlin | 00:53 | |
| *** pratikmallya has quit IRC | 01:03 | |
| *** junxu has joined #senlin | 01:06 | |
| *** jun_ has quit IRC | 01:09 | |
| *** junxu has quit IRC | 01:11 | |
| *** junxu has joined #senlin | 01:11 | |
| Qiming | morning | 01:17 |
|---|---|---|
| xuhaiwei | morning | 01:33 |
| openstackgerrit | Merged openstack/python-senlinclient: The param endpoint in doc string is deprecated https://review.openstack.org/268903 | 01:35 |
| junxu | morning | 01:47 |
| *** Liuqing has quit IRC | 01:47 | |
| *** elynn__ has joined #senlin | 01:52 | |
| *** elynn__ has quit IRC | 01:57 | |
| *** elynn__ has joined #senlin | 01:58 | |
| *** elynn__ has quit IRC | 02:00 | |
| *** elynn has joined #senlin | 02:00 | |
| elynn | Morning | 02:01 |
| *** Yanyanhu has joined #senlin | 02:04 | |
| *** Liuqing has joined #senlin | 02:11 | |
| openstackgerrit | junxu proposed openstack/senlin: Warning when delete cluster with reciver https://review.openstack.org/269392 | 02:13 |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Add Check and Recover into Node Action https://review.openstack.org/264568 | 02:30 |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Add Recover as a Cluster Action https://review.openstack.org/267419 | 02:36 |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Add Recover into RPC API https://review.openstack.org/267922 | 02:43 |
| *** elynn has quit IRC | 03:11 | |
| *** yuanying has quit IRC | 03:15 | |
| *** elynn has joined #senlin | 03:16 | |
| Qiming | lixinhui, there? | 03:23 |
| lixinhui | yes | 03:23 |
| Qiming | the above patches you submitted | 03:23 |
| Qiming | are there dependencies among them? | 03:23 |
| lixinhui | Rpc depends on cluster depends on node | 03:24 |
| Qiming | did you introduce patch dependencies explicitly? | 03:25 |
| lixinhui | yes | 03:25 |
| lixinhui | I add the deepndency into commit message | 03:25 |
| Qiming | gerrit new version sucks, cannot see dependency easily | 03:25 |
| lixinhui | 267922 depends on 267419 depends on 264568 | 03:26 |
| Qiming | okay | 03:26 |
| Qiming | I'm on 264568 at the moment | 03:26 |
| lixinhui | thanks | 03:27 |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Add Check and Recover as a Cluster Action https://review.openstack.org/267419 | 03:49 |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Add Check and Recover as a Cluster Action https://review.openstack.org/267419 | 03:52 |
| *** xuhaiwei has quit IRC | 04:01 | |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Add Check and Recover as a Cluster Action https://review.openstack.org/267419 | 04:02 |
| *** yuanying has joined #senlin | 04:07 | |
| *** xuhaiwei has joined #senlin | 04:07 | |
| *** elynn has quit IRC | 04:21 | |
| *** elynn has joined #senlin | 05:04 | |
| *** gongysh has joined #senlin | 05:23 | |
| xuhaiwei | saw this mail? http://lists.openstack.org/pipermail/openstack-dev/2016-January/084133.html | 05:40 |
| Qiming | xuhaiwei, yep | 05:42 |
| elynn | Finally someone stand out and stop this behavior :) | 05:50 |
| Qiming | hi, just post some comments to lixinhui's patch | 05:52 |
| Qiming | https://review.openstack.org/#/c/267922/3/senlin/tests/unit/engine/service/test_clusters.py | 05:52 |
| Qiming | there are some problems not specific to this patch in particular | 05:53 |
| Qiming | we have been doing something wrong in unit tests | 05:53 |
| Qiming | for each unit test, we should mimic all possible return values from the functions called, and test if the test target behaves as expected | 05:54 |
| Qiming | it NEVER means we have to create something for real | 05:54 |
| Qiming | it is too heavy weight, it is (sometimes) masking the fact a function hasn't been thoroughly tested | 05:54 |
| Qiming | please everyone help pay attention to this when reviewing patches | 05:55 |
| Qiming | it would be even better if someone can help improve tests | 05:56 |
| Qiming | again, it is not a problem of this patch in particular, it was a mistake we made in the early days, so ... we simply need to stop doing that | 05:57 |
| xuhaiwei | yes, I have done the similar thing that creating a real cluster, profile and node for test | 05:59 |
| Qiming | me too | 05:59 |
| elynn | I get it, we should avoid storing *real* record in unittests, but for some unittests like create/update/delete, do we still avoid to create real object? | 06:01 |
| *** elynn_ has joined #senlin | 06:02 | |
| Qiming | if we don't stop this, we will see more and more test cases bumping in with real objects created for no good | 06:02 |
| Qiming | https://review.openstack.org/#/c/269392/1/senlin/tests/unit/engine/service/test_clusters.py | 06:02 |
| Qiming | elynn, good question, we should focus on the logic inside the target function only, testing it with all possible return values (even exceptions) from other functions | 06:04 |
| Qiming | each layer should do its own job, instead of involving too much entities into that test | 06:04 |
| Qiming | we have seen too much of that in Heat I guess | 06:05 |
| *** elynn__ has joined #senlin | 06:06 | |
| elynn | Qiming: Yes, agree, focus on the logic inside function is enough. | 06:06 |
| Qiming | please be picky at this when review | 06:06 |
| *** elynn_ has quit IRC | 06:07 | |
| elynn | so besides db api tests, we should be more careful about creating real objects in unittests, right? | 06:07 |
| Qiming | yes | 06:12 |
| Qiming | for example, creating a cluster is an expensive operation | 06:13 |
| Qiming | it creates an action and queues that action then schedule that action for execution ... | 06:13 |
| *** elynn__ has quit IRC | 06:13 | |
| Qiming | we almost never need to do that in an unit test | 06:13 |
| openstackgerrit | Merged openstack/python-senlinclient: Add back show_details for node-show https://review.openstack.org/268938 | 06:17 |
| openstackgerrit | Merged openstack/senlin: Revise version json format https://review.openstack.org/268862 | 06:20 |
| lixinhui | got it, Qiming | 06:25 |
| Qiming | Yanyanhu, I'm working on region placement policy to enable it to handle CLUSTER_SCALE_IN and CLUSTER_RESIZE | 06:26 |
| Yanyanhu | ok | 06:27 |
| Yanyanhu | I'm now working on adding priority back and fix priority value for builtin policies | 06:27 |
| Yanyanhu | I first tried to work on action data issue but found related code in placement policy and deletion policy is not clean | 06:28 |
| xuhaiwei | Qiming, about the deletion policy for cross az/region, can we pass data from different actions? | 06:28 |
| Qiming | Yanyanhu, yes, I'm fixing that | 06:28 |
| Yanyanhu | cool! | 06:28 |
| Yanyanhu | so I will first address priority issue | 06:28 |
| Qiming | Yanyanhu, the current data transfer for placement policy introduced a new key in action.data -- 'placement' | 06:29 |
| Qiming | I'm moving the policy decisions to 'creation' or 'deletion' accordingly, so that the downstream checking would be easier | 06:29 |
| Yanyanhu | yes, I feel maybe put placements into 'creation'/'deletion' field is better? | 06:29 |
| Qiming | hoho, we are saying the same language, :) | 06:29 |
| Yanyanhu | exact what I'm thinking | 06:29 |
| Yanyanhu | right :) | 06:30 |
| Yanyanhu | then the action data struct will be clearer | 06:30 |
| Qiming | xuhaiwei, yes, we are passing policy decisions using the action.data field | 06:30 |
| Qiming | xuhaiwei, all policies should read the action.data field and see if there are inputs they can understand and consume | 06:30 |
| xuhaiwei | how can different actions share the same action.data? since when one action is doing the other action is already over | 06:31 |
| Qiming | those policies are also expected to yield their decicions back to the action.data field so that downstream checkings can be performed | 06:31 |
| Qiming | no two actions will share policy decisions | 06:32 |
| xuhaiwei | for example, placement policy is triggered by CLUSTER_SCALE_OUT, but deletion policy is triggered by CLUSTER_SCALE_IN, how can they share data by action.data? | 06:33 |
| xuhaiwei | the two policies are triggered by different actions | 06:33 |
| Qiming | xuhaiwei, no, they won | 06:34 |
| Qiming | they won't share data | 06:34 |
| Qiming | they even won't execute in parallel | 06:34 |
| xuhaiwei | yes | 06:34 |
| Qiming | one action already locked the cluster | 06:34 |
| xuhaiwei | so as we discussed at the meetup, placement policy will tell the placement info to deletion policy, right? | 06:35 |
| Qiming | yes | 06:35 |
| xuhaiwei | that can't be done by sharing action.data, so I put the placement info into cluster.data | 06:35 |
| Qiming | that can be done | 06:36 |
| xuhaiwei | can you review this patch again? https://review.openstack.org/#/c/267969/ | 06:36 |
| Qiming | xuhaiwei, we really don't need to store this kind of data to cluster.data | 06:39 |
| Qiming | currently, we are reserving cluster.data for other purposes | 06:39 |
| xuhaiwei | then how to share it ? | 06:39 |
| openstackgerrit | Liuqing Jing proposed openstack/senlin-dashboard: Fix the broken senlin dashboard https://review.openstack.org/269451 | 06:39 |
| Qiming | we save it into action.data | 06:39 |
| Qiming | the node distribution data is only meaningful for policy decisions | 06:40 |
| Qiming | policies are checked when actions are performed | 06:40 |
| Qiming | after an action is executed, we don't have to save the node distribution data | 06:40 |
| Qiming | as for cluster.data | 06:41 |
| Qiming | we are retaining them for node operations | 06:41 |
| xuhaiwei | but when deletion policy is triggered, it doesn't know the distribution info, where to get it? | 06:42 |
| Qiming | refresh it | 06:42 |
| xuhaiwei | collecting them again from the nodes? | 06:42 |
| Qiming | yes | 06:42 |
| Qiming | now I see where you are from | 06:43 |
| Qiming | you want to optimize this | 06:44 |
| xuhaiwei | it will make many duplicated source from placement policy | 06:44 |
| Qiming | that is a good point | 06:44 |
| xuhaiwei | yes, I want to save it to the cluster.data, and fresh it when placement policy is attached | 06:45 |
| Qiming | that should work | 06:45 |
| Qiming | however, my current focus is to enable the basics | 06:46 |
| xuhaiwei | and when handling deletion, in deletion policy we only need fetch the data from cluster.data | 06:46 |
| Qiming | I mean enabling region_placement policy to handle CLUSTER_SCALE_IN and CLUSTER_RESZIE | 06:46 |
| xuhaiwei | I know that | 06:46 |
| Qiming | once this workflow is there, we can optimize it and see if the data can be shared to save an additional cluster-wide refresh | 06:47 |
| Qiming | xuhaiwei and Yanyanhu, please check if this is okay? http://paste.openstack.org/show/484236/ | 06:48 |
| Qiming | it is the revised docstring for the region_placement policy I'm working on | 06:49 |
| xuhaiwei | Qiming, 'count' is not in the sample josn | 06:51 |
| Qiming | a simplified version: http://paste.openstack.org/show/484237/ | 06:52 |
| xuhaiwei | I like the second one | 06:52 |
| Qiming | me too | 06:53 |
| xuhaiwei | since you are thinking about region_placement, I will think about zone_placement for cluster_resize action | 06:53 |
| Qiming | em, the same logic can be copied there later on | 06:54 |
| xuhaiwei | ok | 06:54 |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Add Check and Recover as a Cluster Action https://review.openstack.org/267419 | 07:13 |
| openstackgerrit | Yanyan Hu proposed openstack/senlin: Fix getting node addrress logic in os lbaas driver https://review.openstack.org/269477 | 07:52 |
| openstackgerrit | Yanyan Hu proposed openstack/senlin: Add fixed priority for built-in policies https://review.openstack.org/269478 | 07:52 |
| Yanyanhu | faint, accidentally added dependency between these two patches | 07:53 |
| openstackgerrit | Ethan Lynn proposed openstack/senlin: Make senlin-api a console script entry point https://review.openstack.org/267375 | 07:59 |
| openstackgerrit | Ethan Lynn proposed openstack/senlin: Make senlin-manage a console_script entry point https://review.openstack.org/267374 | 07:59 |
| openstackgerrit | Ethan Lynn proposed openstack/senlin: Make senlin-engine a console script entry point https://review.openstack.org/267376 | 07:59 |
| openstackgerrit | Ethan Lynn proposed openstack/senlin: Fix devstack script to use console scripts https://review.openstack.org/269484 | 07:59 |
| openstackgerrit | Merged openstack/senlin-dashboard: Fix the broken senlin dashboard https://review.openstack.org/269451 | 07:59 |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Add Check and Recover into Node Action https://review.openstack.org/264568 | 08:00 |
| openstackgerrit | Merged openstack/senlin: Add 'cooldown' for scaling policy https://review.openstack.org/268918 | 08:05 |
| openstackgerrit | Yanyan Hu proposed openstack/senlin: Rework lb policy https://review.openstack.org/264670 | 08:24 |
| Yanyanhu | hi, Qiming, have proposed the patch for reenabling priority attr of cluster policy bonding. Will wait for your patch on placement policy before making further change on action data transimission logic. | 08:42 |
| Yanyanhu | about the output format, how about this one? http://paste.openstack.org/show/484243/ | 08:45 |
| Yanyanhu | sorry, just noticed the msg you pasted before | 08:45 |
| Yanyanhu | or this one: http://paste.openstack.org/show/484245/ | 08:48 |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Add check and Recover into RPC API https://review.openstack.org/267922 | 08:49 |
| Yanyanhu | just feel maybe puting the placement results into 'placements' field is clearer? | 08:49 |
| openstackgerrit | junxu proposed openstack/senlin: Warning when delete cluster with receivers https://review.openstack.org/269392 | 08:59 |
| *** elynn has quit IRC | 09:04 | |
| *** elynn has joined #senlin | 09:04 | |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Add Check and Recover as a Cluster Action https://review.openstack.org/267419 | 09:06 |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Add Check and Recover into Node Action https://review.openstack.org/264568 | 09:07 |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Add Check and Recover as a Cluster Action https://review.openstack.org/267419 | 09:11 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Add region validation to keystone driver https://review.openstack.org/269504 | 09:16 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Enable cluster to do node distribution refresh https://review.openstack.org/269505 | 09:17 |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Add Check and Recover into Node Action https://review.openstack.org/264568 | 09:35 |
| Yanyanhu | hi, Qiming, just saw your patch 269505. I found the dist result contains the node amount in each zones/regions. Do we need to fill the nodes list into the dist result as well? | 09:37 |
| Yanyanhu | https://review.openstack.org/#/c/269505/1/senlin/engine/cluster.py | 09:38 |
| Qiming | why? | 09:38 |
| openstackgerrit | Ayush Garg proposed openstack/python-senlinclient: Update the word 'Senin' to 'Senlin' https://review.openstack.org/269514 | 09:39 |
| Yanyanhu | since the nodes list is some policies expect I think, like lb policy | 09:39 |
| Qiming | don't get it | 09:40 |
| Qiming | lb policy care about node distribution across regions? across azs? | 09:41 |
| Yanyanhu | I mean lb policy need deleted nodes list as input, so if there is no deletion policy attached, it may need to invoke get_region_distribution/get_zone_distribution to get it | 09:42 |
| Yanyanhu | no | 09:42 |
| Yanyanhu | it just needs the deleted nodes list, no matter where these nodes are | 09:42 |
| Qiming | these two methods are utility functions for clusters | 09:42 |
| Qiming | it tells you how current nodes are distributed | 09:42 |
| Yanyanhu | ok, I see | 09:42 |
| Qiming | that is all | 09:42 |
| Yanyanhu | it's not for this purpose | 09:43 |
| Qiming | right | 09:43 |
| Qiming | even if there is a function that converts {'deletion': {'azs': ['az1', 'az1', 'az2']}} to {'deletion': {'candidates': ['node12', 'node13', 'node25']}} | 09:44 |
| Qiming | that function should not belong to cluster module, right? | 09:45 |
| Yanyanhu | yes | 09:45 |
| Yanyanhu | I think cluster_action is better place for it | 09:45 |
| Qiming | yep | 09:45 |
| Qiming | maybe someone can start working on such a function | 09:46 |
| Yanyanhu | ok, I can take over it | 09:46 |
| Qiming | it can be shared between deletion policy and lb policy | 09:46 |
| Qiming | great | 09:46 |
| Qiming | I'm really not good at writing algorithms ... sigh | 09:46 |
| Yanyanhu | yes, that will be a kind of util function | 09:46 |
| Yanyanhu | me neither... | 09:47 |
| Yanyanhu | you should be much better than me on algorithms | 09:47 |
| openstackgerrit | venkatamahesh proposed openstack/python-senlinclient: Update the word 'Senin' to 'Senlin' https://review.openstack.org/269514 | 09:47 |
| Qiming | ... | 09:47 |
| Yanyanhu | at least, both your bachelor and master degree are computer science :) | 09:47 |
| Yanyanhu | I'm a layman for this, haha | 09:48 |
| Qiming | ... | 09:48 |
| openstackgerrit | Ethan Lynn proposed openstack/senlin: Fix devstack script to use console scripts https://review.openstack.org/269484 | 09:51 |
| openstackgerrit | Merged openstack/senlin: Warning when delete cluster with receivers https://review.openstack.org/269392 | 09:58 |
| openstackgerrit | Liuqing Jing proposed openstack/senlin-dashboard: Add receivers panel https://review.openstack.org/269528 | 10:02 |
| Qiming | hi, Yanyanhu | 10:06 |
| Yanyanhu | hi | 10:07 |
| Qiming | I don't think the revision to priority work should be blocked by the placement policies | 10:07 |
| Yanyanhu | sure | 10:07 |
| Qiming | just go ahead assuming that we won't get placement policies work before release | 10:07 |
| Yanyanhu | these two works have no dependency on each other | 10:07 |
| Qiming | I'll do my best to catch up | 10:07 |
| Yanyanhu | many thanks :) | 10:08 |
| *** Qiming has quit IRC | 10:12 | |
| openstackgerrit | venkatamahesh proposed openstack/python-senlinclient: Update the word 'Senin' to 'Senlin' https://review.openstack.org/269514 | 10:12 |
| openstackgerrit | Ayush Garg proposed openstack/python-senlinclient: Update the word 'Senin' to 'Senlin' https://review.openstack.org/269514 | 10:12 |
| *** elynn has quit IRC | 10:13 | |
| *** Ayush has joined #senlin | 10:19 | |
| Ayush | Hello All, I have freshly installed senlin using devstack and try to execute the command $senlin profile-type-list but it exits with an error "The plugin None could not be found". Can anyone help me to resolve this? | 10:22 |
| *** Yanyanhu has quit IRC | 10:24 | |
| openstackgerrit | Merged openstack/python-senlinclient: Update the word 'Senin' to 'Senlin' https://review.openstack.org/269514 | 10:35 |
| *** Qiming has joined #senlin | 11:05 | |
| Ayush | Hello All, I have freshly installed senlin using devstack and try to execute the command $senlin profile-type-list but it exit with an error "The plugin None could not be found". Can anyone help me to resolve this? I am a newbie in this project. | 11:09 |
| Qiming | Ayush, can you help paste the stack trace? | 11:11 |
| Qiming | you can do a $senlin -d profile-type-list | 11:12 |
| Qiming | and try figure out which plugin is missing | 11:13 |
| xuhaiwei | I am afraid he is not using the latest source | 11:15 |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Add check and Recover into RPC API https://review.openstack.org/267922 | 11:53 |
| *** Liuqing has quit IRC | 12:06 | |
| *** gongysh has quit IRC | 12:07 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/python-senlinclient: Updated from global requirements https://review.openstack.org/269350 | 12:08 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/senlin: Updated from global requirements https://review.openstack.org/269353 | 12:09 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/senlin-dashboard: Updated from global requirements https://review.openstack.org/268539 | 12:09 |
| *** elynn has joined #senlin | 12:15 | |
| *** elynn_ has joined #senlin | 12:28 | |
| *** elynn has quit IRC | 12:30 | |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Add check and Recover into RPC API https://review.openstack.org/267922 | 12:31 |
| *** lixinhui_ has joined #senlin | 12:52 | |
| *** yanyanhu has joined #senlin | 12:53 | |
| elynn_ | meeting? | 13:01 |
| yanyanhu | yes | 13:01 |
| openstackgerrit | Merged openstack/senlin: Updated from global requirements https://review.openstack.org/269353 | 13:41 |
| openstackgerrit | Merged openstack/senlin: Make senlin-manage a console_script entry point https://review.openstack.org/267374 | 13:50 |
| *** yanyanhu has quit IRC | 13:53 | |
| *** lixinhui_ has quit IRC | 14:11 | |
| *** elynn_ has quit IRC | 14:12 | |
| openstackgerrit | Merged openstack/python-senlinclient: Updated from global requirements https://review.openstack.org/269350 | 14:21 |
| openstackgerrit | Merged openstack/senlin: Add region validation to keystone driver https://review.openstack.org/269504 | 14:23 |
| *** pratikmallya has joined #senlin | 14:27 | |
| *** pratikma_ has joined #senlin | 14:29 | |
| *** pratikmallya has quit IRC | 14:32 | |
| openstackgerrit | Merged openstack/senlin: Enable cluster to do node distribution refresh https://review.openstack.org/269505 | 14:58 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Rework region placement policy https://review.openstack.org/269715 | 15:41 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Rework region placement policy https://review.openstack.org/269715 | 15:57 |
| openstackgerrit | Merged openstack/senlin: Fix getting node addrress logic in os lbaas driver https://review.openstack.org/269477 | 16:02 |
| *** Qiming has quit IRC | 16:09 | |
| *** Liuqing has joined #senlin | 16:16 | |
| *** bdrich has joined #senlin | 16:28 | |
| *** Liuqing has quit IRC | 16:46 | |
| *** Liuqing has joined #senlin | 16:50 | |
| *** Liuqing has quit IRC | 16:51 | |
| *** Liuqing has joined #senlin | 16:53 | |
| *** Liuqing has quit IRC | 16:53 | |
| *** pratikma_ has quit IRC | 17:42 | |
| *** pratikmallya has joined #senlin | 18:10 | |
| *** bdrich has quit IRC | 18:19 | |
| *** bdrich has joined #senlin | 19:14 | |
| *** pratikmallya has quit IRC | 22:01 | |
| *** pratikmallya has joined #senlin | 22:19 | |
| *** pratikmallya has quit IRC | 22:40 | |
| *** bdrich has quit IRC | 22:43 | |
| *** pratikmallya has joined #senlin | 23:48 | |
| *** Qiming has joined #senlin | 23:51 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!