Tuesday, 2016-09-20

« Monday, 2016-09-19 Index Wednesday, 2016-09-21 »
*** zhurong has quit IRC00:04
*** guoshan has joined #senlin00:11
*** shu-mutou-AFK is now known as shu-mutou00:18
openstackgerritMerged openstack/senlin: Enable senlin policy for dashboard via devstack installation  https://review.openstack.org/36741700:31
*** guoshan has quit IRC00:52
openstackgerritXueFeng Liu proposed openstack/senlin: Fix service manage cleanup  https://review.openstack.org/37192201:02
*** zhurong has joined #senlin01:12
*** guoshan has joined #senlin01:22
*** elynn has joined #senlin01:38
*** elynn has quit IRC01:43
*** elynn has joined #senlin01:43
*** yanyanhu has joined #senlin01:47
openstackgerritmiaohb proposed openstack/python-senlinclient: Fix error in action list  https://review.openstack.org/37282901:56
openstackgerritmiaohb proposed openstack/senlin: Fix error in actions.inc  https://review.openstack.org/37283202:03
openstackgerritXueFeng Liu proposed openstack/senlin: Fix FEATURES.rst error  https://review.openstack.org/37284402:34
openstackgerritQiming Teng proposed openstack/senlin: Rework NODE_JOIN logic  https://review.openstack.org/37247302:38
openstackgerritMerged openstack/senlin: Support cluster replace action  https://review.openstack.org/36853402:41
openstackgerritMerged openstack/senlin: Fix error in senlin-manage.rst  https://review.openstack.org/37222502:41
openstackgerritMerged openstack/senlin: Fix some typos in doc  https://review.openstack.org/37221002:42
openstackgerritMerged openstack/senlin: Fix error in policies.rst  https://review.openstack.org/37220502:42
openstackgerritMerged openstack/senlin: Fix error in receivers.rst  https://review.openstack.org/37219702:42
openstackgerritMerged openstack/senlin: Select roles for trust building  https://review.openstack.org/36708802:42
openstackgerritMerged openstack/senlin: API and engine support for receiver notifying  https://review.openstack.org/36720902:42
openstackgerritmiaohb proposed openstack/python-senlinclient: Fix error in action list  https://review.openstack.org/37284802:42
openstackgerritQiming Teng proposed openstack/senlin: Rework NODE_LEAVE action logic  https://review.openstack.org/37248202:46
openstackgerritXueFeng Liu proposed openstack/senlin: Fix error in conf.py  https://review.openstack.org/37285503:02
openstackgerritQiming Teng proposed openstack/senlin: Rework NODE_JOIN logic  https://review.openstack.org/37247303:34
openstackgerritmiaohb proposed openstack/senlin: Fix error in actions.inc  https://review.openstack.org/37286503:39
openstackgerritmiaohb proposed openstack/senlin: Fix misspelling in parameters.yaml  https://review.openstack.org/37286803:45
openstackgerritMerged openstack/senlin: Rework NODE_CREATE action logic  https://review.openstack.org/37230103:51
openstackgerritMerged openstack/senlin: Rework NODE_DELETE action logic  https://review.openstack.org/37235103:54
*** guoshan has quit IRC03:55
openstackgerritlvdongbing proposed openstack/senlin: Delete engine from db when stop engine  https://review.openstack.org/37287503:59
*** elynn has quit IRC04:22
openstackgerritMerged openstack/senlin: Fix error in conf.py  https://review.openstack.org/37285504:24
openstackgerritmiaohb proposed openstack/senlin: Fix error in status.yaml  https://review.openstack.org/37288104:40
*** zhurong has quit IRC04:43
*** elynn has joined #senlin04:52
openstackgerritShu Muto proposed openstack/senlin-dashboard: Remove unnecessary spaces  https://review.openstack.org/37288504:54
*** guoshan has joined #senlin04:56
*** elynn has quit IRC04:56
*** elynn has joined #senlin04:57
openstackgerritXueFeng Liu proposed openstack/senlin: Fix error in registry.py  https://review.openstack.org/37288605:00
openstackgerritmiaohb proposed openstack/senlin: Fix error in FEATURES.rst  https://review.openstack.org/37288905:13
openstackgerritXueFeng Liu proposed openstack/senlin: Modify unit test function name  https://review.openstack.org/37289205:23
openstackgerritXueFeng Liu proposed openstack/senlin: Modify test_service_manage_report_cleanup  https://review.openstack.org/37289205:25
openstackgerritQiming Teng proposed openstack/senlin: Rework NODE_LEAVE action logic  https://review.openstack.org/37248205:34
*** zhurong has joined #senlin05:35
*** zhurong has quit IRC05:47
*** zhurong has joined #senlin05:48
*** elynn_ has joined #senlin05:50
*** elynn has quit IRC05:52
*** elynn_ has quit IRC05:54
*** elynn_ has joined #senlin05:55
openstackgerritzhufl proposed openstack/senlin: Remove unnecessary setUp  https://review.openstack.org/37290706:07
openstackgerritmiaohb proposed openstack/senlin: Fix error in container-cluster  https://review.openstack.org/37291806:18
openstackgerritOpenStack Proposal Bot proposed openstack/senlin-dashboard: Imported Translations from Zanata  https://review.openstack.org/37292006:19
*** elynn_ has quit IRC06:33
*** elynn has joined #senlin06:36
openstackgerritmelissaml proposed openstack/senlin: Fix error in actions.rst  https://review.openstack.org/37293806:39
openstackgerritmiaohb proposed openstack/senlin: Fix error in template.rst  https://review.openstack.org/37293906:39
*** elynn has quit IRC06:41
*** elynn has joined #senlin06:41
openstackgerritmelissaml proposed openstack/senlin: Fix error in actions.rst  https://review.openstack.org/37293806:47
openstackgerritRUIJIE YUAN proposed openstack/senlin: Service support node replace  https://review.openstack.org/36925907:18
openstackgerritMerged openstack/senlin-dashboard: Remove unnecessary spaces  https://review.openstack.org/37288507:24
openstackgerritYanyan Hu proposed openstack/senlin: Fix nova_v2 test driver  https://review.openstack.org/37298007:29
openstackgerritYanyan Hu proposed openstack/senlin: Fix nova_v2 test driver  https://review.openstack.org/37298007:39
openstackgerritzhangyanxian proposed openstack/senlin: Fix a typo in utils.py  https://review.openstack.org/37299607:41
openstackgerritzhangyanxian proposed openstack/senlin: Fix a typo in utils.py  https://review.openstack.org/37299607:42
openstackgerritYanyan Hu proposed openstack/senlin: [WIP]Implement receiver notification handling  https://review.openstack.org/37300407:46
*** guoshan has quit IRC07:54
*** guoshan has joined #senlin07:54
*** zhurong has quit IRC07:55
openstackgerritQiming Teng proposed openstack/senlin: Use new desired when eval cluster status  https://review.openstack.org/37301508:05
openstackgerritQiming Teng proposed openstack/senlin: Tweak CLUSTER_ADD_NODE action for capacity update  https://review.openstack.org/37301608:05
*** zhurong has joined #senlin08:13
*** shu-mutou is now known as shu-mutou-AFK08:36
openstackgerritYanyan Hu proposed openstack/senlin: Remove a finished TODO in comment  https://review.openstack.org/37303908:40
openstackgerritYanyan Hu proposed openstack/senlin: [WIP]Implement receiver notification handling  https://review.openstack.org/37300408:50
openstackgerritRUIJIE YUAN proposed openstack/senlin: Service support node replace  https://review.openstack.org/36925909:01
*** rvasilets__ has left #senlin09:15
openstackgerritShan Guo proposed openstack/senlin: fix typos in doc  https://review.openstack.org/37308409:19
openstackgerritShan Guo proposed openstack/senlin: fix typos in doc  https://review.openstack.org/37308409:20
openstackgerritXueFeng Liu proposed openstack/senlin: Fix cluster_check from health manage  https://review.openstack.org/36993709:27
openstackgerritXueFeng Liu proposed openstack/senlin: Fix cluster_check from health manage  https://review.openstack.org/36993709:29
openstackgerritQiming Teng proposed openstack/senlin: Tweak CLUSTER_ADD_NODE action for capacity update  https://review.openstack.org/37301609:29
openstackgerritRUIJIE YUAN proposed openstack/senlin: Service support node replace  https://review.openstack.org/36925909:36
openstackgerritMerged openstack/senlin-dashboard: Imported Translations from Zanata  https://review.openstack.org/37292009:37
openstackgerritlvdongbing proposed openstack/senlin: Delete engine from db when stop engine  https://review.openstack.org/37287509:41
openstackgerritRUIJIE YUAN proposed openstack/senlin: Service support node replace  https://review.openstack.org/36925909:45
openstackgerritMerged openstack/senlin: Remove a finished TODO in comment  https://review.openstack.org/37303909:47
*** yanyanhu has quit IRC09:53
openstackgerritMerged openstack/senlin: Fix error in container-cluster  https://review.openstack.org/37291809:59
*** zhurong has quit IRC10:03
openstackgerritRUIJIE YUAN proposed openstack/senlin: Service support node replace  https://review.openstack.org/36925910:03
*** guoshan has quit IRC10:05
openstackgerritRUIJIE YUAN proposed openstack/senlin: Service support node replace  https://review.openstack.org/36925910:16
openstackgerritRUIJIE YUAN proposed openstack/senlin: API support node replace  https://review.openstack.org/36929110:16
openstackgerritmelissaml proposed openstack/senlin: Fix error in deletion_v1.rst  https://review.openstack.org/37313910:25
*** elynn has quit IRC10:37
openstackgerritXueFeng Liu proposed openstack/senlin: Fix error in tutorial/receivers.rst  https://review.openstack.org/37316510:57
*** guoshan has joined #senlin11:06
openstackgerritRUIJIE YUAN proposed openstack/senlin: API support node replace  https://review.openstack.org/36929111:12
*** ruijie has quit IRC11:18
openstackgerritmiaohb proposed openstack/python-senlinclient: Fix error in build-info  https://review.openstack.org/37319311:30
*** guoshan has quit IRC11:44
*** guoshan has joined #senlin11:45
openstackgerritmiaohb proposed openstack/senlin: Fix error in authorization.rst  https://review.openstack.org/37321111:50
*** zhurong has joined #senlin11:54
*** guoshan has quit IRC11:54
openstackgerritMerged openstack/python-senlinclient: Fix error in action list  https://review.openstack.org/37284811:54
*** guoshan has joined #senlin11:55
openstackgerritXueFeng Liu proposed openstack/senlin: modify profile_update  https://review.openstack.org/37324112:25
*** guoshan has quit IRC12:34
*** yanyanhu has joined #senlin12:40
*** guoshan has joined #senlin12:49
*** elynn has joined #senlin12:59
Qimingyanyan, can you help start the meeting?13:00
yanyanhuQiming, sure13:00
QimingI don't have any peculiar agenda today13:00
Qimingwill join you in a few minutes13:00
yanyanhubut I can't edit the agenda, always can't see the captcha...13:00
yanyanhusure, will start it13:00
Qimingbabysitting ....13:00
yanyanhuI see13:00
yanyanhuhi, guys, meeting has been started13:01
*** ruijie has joined #senlin13:02
yanyanhuhi, ruijie, please join #openstack-meeting channel if convenient :)13:03
ruijieok13:03
openstackgerritMerged openstack/senlin: Service support node replace  https://review.openstack.org/36925913:14
openstackgerritMerged openstack/senlin: Delete engine from db when stop engine  https://review.openstack.org/37287513:15
openstackgerritXueFeng Liu proposed openstack/senlin: modify profile_update  https://review.openstack.org/37324113:32
*** openstackstatus has joined #senlin13:37
*** ChanServ sets mode: +v openstackstatus13:37
*** guoshan has quit IRC13:41
*** guoshan has joined #senlin13:43
-openstackstatus- NOTICE: OpenStack Infra now has a Twitter bot, follow it at https://twitter.com/openstackinfra13:45
openstackgerritXueFeng Liu proposed openstack/senlin: modify profile_update  https://review.openstack.org/37324113:48
openstackgerritMerged openstack/senlin: Fix nova_v2 test driver  https://review.openstack.org/37298013:50
*** guoshan has quit IRC13:53
*** guoshan has joined #senlin13:57
elynnI'm thinking treating user and ceilometer differently...14:01
yanyanhuelynn, umm, imho, once user let ceilometer make decision on behalf of him/she, we should treat ceilometer as a real user...14:02
yanyanhuthe only difference is the user is now a robot14:02
elynnWe can use metadata to tell scaling action based on what, desired or current?14:03
yanyanhuhi, Qiming, will go through the table and think it more thoroughtly14:03
elynnI will go through it too...14:03
*** guoshan has quit IRC14:03
yanyanhuelynn, it could be difficult to keep consistency if we apply that way...14:04
elynnkeep consistency on what?14:05
yanyanhuI mean in that case, user need to keep in mind that there are two possible ways to recalculate the desired14:06
Qiminglet's take two steps back14:06
Qimingimagine this14:06
Qimingif we now remove desired capacity property from a cluster14:07
yanyanhuyes14:07
Qimingor rename it to initial_capacity14:07
yanyanhuok14:07
Qimingthen we never show it when do cluster-show14:07
yanyanhuQiming, yes, that makes sense14:08
Qimingall logics becomes a lot more clearer right?14:08
elynnyes, sounds great.14:08
yanyanhuright14:08
Qimingusers (or robots) always do scaling decisions based on the real (current) capacity14:08
yanyanhusince in that case, users themselves will take the responsibility to keep the value of "desired"14:08
Qimingthat is something we can agree upon14:08
yanyanhuyep14:09
Qimingokay, let's move one step forward14:09
yanyanhuuser need to remind what they desired, e.g. in a note book :)14:09
Qimingdoes it matter at all?14:09
Qimingdesired is only used to derive the 'health' status of a cluster14:10
Qimingit is never the fact14:10
Qimingwhen desired=current, you consider yourself a luck dog14:10
yanyanhuJust user need to tell whether a cluster is healthy by themselves in that case?14:10
yanyanhuyes14:10
Qimingso desired_capacity can be kept14:10
Qimingit is user's expectation, previous expectation14:10
yanyanhuuser need to check the cluster real size and compare with what they expected to decide it14:10
Qimingsenlin is calculating health status based on it14:11
Qimingyes14:11
Qimingthen back to14:11
Qiming[22:09] <Qiming> okay, let's move one step forward14:11
Qimingwe now add a 'current_capacity' property when users do cluster-show14:11
Qimingin addition to the 'nodes' property14:12
yanyanhuQiming, I see what you mean. But that means there is an implication that "user will change their desired once they found cluster creating result is different from they expected"14:12
Qimingnext time, when he/she makes decisions to scale a cluster, they should use the 'current_capacity' as the basis14:12
Qimingeverytime a resize operation is issued, users ARE chaning their desired capacity, chaning their expetation14:13
Qiminga new bar for senlin to meet14:13
Qimings/chaning/changing/14:13
yanyanhuexactlly, that's why I said there is an "implication" here14:13
Qimingexplain this to users14:13
Qimingthen we save our lives14:14
yanyanhuwe need to assume that user will "change/adjust" their desired once they found the result is different from expected14:14
Qimingdo NOT chase desired capacity14:14
Qimingyou will find the code getting more and more complicated14:14
elynnThat's good, use 3 capacities instead of 1.14:14
Qimingusers will always be told the 'actual' and 'desired'14:14
elynninitial_capacity, health_capacity, current_capacity.14:15
Qimingexactly, just in current implementation, the newton release, we have initial=health=desired14:15
yanyanhuumm, that means user should always accept "real" as their "desired"14:15
Qimingreal is not their desired ...14:16
Qimingwhy are you saying that?14:16
Qimingwhen you have a cluster-show command showing you this:14:16
Qiming|  current_capacity   | 3                 |14:16
yanyanhusince after failure happened, user will ajust their expection to the real one14:16
Qiming| desired_capacity |  5      |14:16
Qimingnot 'to the real one', it is 'based on the real one'14:17
yanyanhuyes, then user will know, oh, the actually size is 3, then the following operation will be done based on this14:17
Qimingwe are letting users take all responsibilities, instead of pretending that senlin is very smart to handle all those differences automatically14:18
yanyanhuQiming, exactly14:18
elynnQiming: you are right.14:18
Qiminggetting things right, in my opinion, is top priority14:19
yanyanhuthat's why I said we assume user will understand it and change their expection14:19
Qimingthen when the whole implementation is solid and consistent14:19
yanyanhuonce they saw failure happen14:19
Qimingwe can learn how to become a little bit smarter, step by step14:19
Qimingfor example, maybe a new cluster_reset operation?14:19
Qimingmaybe health policy v2, which can create/delete nodes and even do some convergence14:20
yanyanhuso maybe we should completely drop "desired_capacity"?14:20
elynnsounds good, so the only problem is how do we get rid of desired_capacity...14:20
Qimingwith this mindset shift, we are actually making 'desired capacity' just a reference number14:20
elynnHow to keep backward compatible...14:21
Qiminga hint telling users why you have your cluster in a WARNING status14:21
yanyanhuanother possible issue is if a cluster has 5 nodes, but one of them is in error status, the real size should be 4 or 5?14:21
Qimingyanyanhu, finally, a great question14:21
Qimingthe current implementation is still not perfect14:21
elynnyanyanhu: for current implementation it's 5...14:21
Qimingyes14:21
elynnthen cluster_recover will correct it..14:22
Qiminggive me a second14:22
yanyanhuok14:22
Qiminghttps://review.openstack.org/#/c/372016/14:23
Qimingfound it14:23
QimingI have noticed this issue14:23
Qimingbut cannot get it done by newton I think14:23
Qimingin the above patch, the node_count_by_cluster was augmented with keyword arguments14:24
Qimingthe plan was to do node_count_by_cluster(cluster_id, status='ACTIVE')14:24
yanyanhuyes, that is very useful14:24
Qimingonce I get the CLUSTER_DEL_NODES thing fixed14:25
QimingI can do a global change14:25
yanyanhuso IMHO, maybe we should completely drop "desired_capacity". And let user to define a init_size when cluster is first created. Then we only maintain min/max later.14:25
Qimingadd that 'status='ACTIVE'' parameters where appropriate14:25
Qimingyanyanhu, that is a mindset14:25
yanyanhuand maybe a current_size which could be unnecessary since user can get it by count nodes lenght14:25
yanyanhulength14:26
Qimingwe can keep it as a reference14:26
Qimingas for current_capacity14:26
Qimingit can be a simple client side fix14:26
yanyanhusince the word "desired" could be confusing :)14:26
yanyanhudifferent people could have different explanation on it14:26
Qiminghowever, if we define current_capacity as number of 'ACTIVE' nodes, it has to be fixed at server side14:26
yanyanhuQiming, right14:26
Qimingwe have the final say on this14:27
yanyanhuQiming, that is about health I think.14:27
yanyanhuso may not very urgent14:27
Qimingit is 'desired', never think of it as 'guaranteed'14:27
Qimingit is a something derived from your previous operation14:27
yanyanhuQiming, yes. since nobody can guarantee anything in cloud :)14:28
Qimingand a number used to derive cluster status14:28
Qimingperiod. that is all14:28
elynnat least we reach some consensus: to get rid of desired_capacity.14:28
Qimingafter we add 'current_capacity', or 'active_nodes'14:28
Qimingit will be much easier to explain this14:28
yanyanhuso after this fix. the desired will actually the inital. after cluster creation is done, we don't consider it for any further scaling/resizing/adding node/deleting node14:28
yanyanhuwe just keep checking min/max14:28
elynnor maybe current_capacity is actually node_count ...14:29
Qimingno resize operation is based on desired capacity14:29
elynnto be less confusing14:29
Qimingdesired capacity is updated though, as a result of any resize related operation14:29
Qiming'or maybe current_capacity is actually node_count', YES14:29
yanyanhuthen user themselves will be responsible to decide what size is they expecting14:29
Qimingyes14:29
yanyanhusenlin will just give them information about the "current" cluster size14:30
yanyanhuI see14:30
yanyanhuclear for me now14:30
Qimingin my opinion, it will be a lot easier to attempt basing scaling operation on a previous 'desired' capacity14:30
Qimingyes14:30
yanyanhuyes, that will be much easier14:30
yanyanhufor senlin14:30
elynnso we should be care about the word 'current' here...14:31
Qimingem, good point14:31
yanyanhuelynn, yes, need a statement about it14:31
elynnto make our lives easier in the future...14:31
yanyanhuand it depends on how we handle health and recovery14:31
Qimingyep, a long way to go regarding health and recovery14:32
yanyanhubut without HA, the "size" related logic is clear14:32
Qimingem, seems we can reallocate one Barcelona session for something else14:32
yanyanhusenlin just tells enduser: look, your cluster is like this, it has 4 nodes and size limit is 1 to 514:33
yanyanhudone14:33
yanyanhuQiming, yes, definitely need session for it14:33
Qimingright, that is the world users live14:33
QimingI mean we have gone thru the size thing today, we can use that timeslot for other topics14:34
Qimingmaybe more discussion on health14:34
yanyanhuthen we put all health logic to HA framework, e.g. making senlin smarter to be aware user's "desired"14:34
yanyanhuQiming, oh I see14:34
Qimingbut you are the PTL now, yanyan14:34
yanyanhuQiming, ah? not after summit14:35
QimingI don't see anyone else submitting a nomination in db14:35
elynnhaha14:35
QimingI'll help you plan for the summit14:35
yanyanhuQiming, thanks a lot :)14:35
yanyanhuneed to learn some new skills...14:35
Qimingand I'll (hopefully) play a good assistant in the coming cycle14:35
yanyanhunot just assistant maybe :)14:36
yanyanhuneed some decisions14:36
elynnand a good father too ;)14:36
QimingI was planning to step down even at the beginning of newton14:36
Qimingno single person is supposed to hold a role forever, not good for project, not good for people14:37
yanyanhuok. maybe need your help to familiar with something like release progress, etc.14:37
Qiminghopefully I can be a good father, :D14:37
yanyanhuyou will :)14:37
yanyanhuhaha14:37
yanyanhuwill leave, my wife will yell at me...14:38
Qimingand ... during the summit season, I'd propose we consider some new core members14:38
yanyanhumaybe we can have some further discussion on summit plan later14:38
yanyanhuQiming, yes14:38
Qimingbased on their exhibited skills and dedication14:38
Qimingmore recognition <=> more responsibility14:39
yanyanhuyep14:39
Qimingwill let you go hug your wife, :)14:39
yanyanhu:) actually I want to hug my baby14:39
QimingLOL14:39
yanyanhuthis msg will be invisible for my wife14:39
Qimingalright14:39
Qiminggood night then14:39
yanyanhuhave a good night14:40
yanyanhutalk to you guys later14:40
elynngood night~14:40
*** yanyanhu has quit IRC14:40
openstackgerritXueFeng Liu proposed openstack/senlin: 8/tgEtz93v3dKigrfRvd4UdUs8jRqG+OxcSYpz/hNQmodify profile_update  https://review.openstack.org/37324114:49
*** ruijie has quit IRC14:55
openstackgerritXueFeng Liu proposed openstack/senlin: modify profile_update  https://review.openstack.org/37324114:57
*** Drago has joined #senlin14:58
*** elynn has quit IRC15:02
*** zhurong has quit IRC15:11
openstackgerritMerged openstack/senlin: Fix a typo in utils.py  https://review.openstack.org/37299616:22
openstackgerritMerged openstack/senlin: Fix error in actions.inc  https://review.openstack.org/37286516:25
openstackgerritMerged openstack/senlin: Fix error in authorization.rst  https://review.openstack.org/37321116:25
openstackgerritMerged openstack/senlin: Modify test_service_manage_report_cleanup  https://review.openstack.org/37289216:25
*** Drago has quit IRC22:44
*** Alexey_Abashkin_ has joined #senlin23:45
*** Alexey_Abashkin has quit IRC23:46
« Monday, 2016-09-19 Index Wednesday, 2016-09-21 »

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!