Tuesday, 2016-12-20

openstackgerritGhanshyam Mann proposed openstack/senlin: Fix Senlin tempest plugin  https://review.openstack.org/41265800:17
*** guoshan has joined #senlin00:27
*** catintheroof has quit IRC00:29
openstackgerritGhanshyam Mann proposed openstack/senlin: Fix Senlin tempest plugin  https://review.openstack.org/41265800:30
*** XueFeng has quit IRC00:42
*** XueFeng has joined #senlin00:43
*** xiangxinyong is now known as edisonxiang01:02
*** XueFeng has quit IRC01:03
*** guoshan has quit IRC01:05
*** XueFeng has joined #senlin01:06
*** guoshan has joined #senlin01:26
*** yanyanhu has joined #senlin01:34
openstackgerritHa Van Tu proposed openstack/senlin: Fix typo in examples  https://review.openstack.org/41269902:24
*** dinghh1 has joined #senlin02:55
*** dinghh has quit IRC02:56
*** chohoor has joined #senlin03:09
openstackgerritGhanshyam Mann proposed openstack/senlin: DNM: just testing  https://review.openstack.org/41272203:13
openstackgerritGhanshyam Mann proposed openstack/senlin: DNM: just testing  https://review.openstack.org/41272203:14
openstackgerritRUIJIE YUAN proposed openstack/senlin: rework profile-get2/list2  https://review.openstack.org/41272503:35
openstackgerritShan Guo proposed openstack/senlin: Add parameter in CLUSTER_DEL_NODES action  https://review.openstack.org/41272903:42
*** guoshan has quit IRC04:46
*** gmann has joined #senlin05:24
gmannQiming: ping, anything changes for APi permission as it failing  - https://review.openstack.org/#/c/412658/205:24
Qiminggmann, nothing I'm aware of05:36
Qimingyanyanhu, can you help look into this as well?05:37
Qimingrecent changes at tempest side is breaking our api/functional/integration tests ...05:37
gmannQiming: yea, but with fix it gives 403 error even on tempest revet patch also - https://review.openstack.org/#/c/412722/05:40
yanyanhuQiming, sure, will check it05:40
gmanneven it works on my local devstack setup05:40
gmannyanyanhu:  this also have same 403 issue. this is depends-on tempest revert one -  https://review.openstack.org/#/c/412722/05:41
yanyanhugmann, ok, let me have a look05:42
*** guoshan has joined #senlin05:46
*** guoshan has quit IRC05:51
yanyanhugmann, Qiming, that is weird, senlin-api service can not bind to port 8778  when it is started05:51
yanyanhuit guess this error is not related to change on tempest05:51
yanyanhulet me check the test log of other failed patches05:52
gmannthis too same issue - http://logs.openstack.org/58/412658/2/check/gate-senlin-dsvm-tempest-api/03d369c/logs/screen-sl-api.txt.gz05:54
yanyanhulooks like 8778 port is occupied by another service. that why 403 exception was raised05:57
yanyanhuand when this patch was proposed and verified in gate, this error haven't happened yet05:58
gmannyea placement one - https://github.com/openstack-dev/devstack/blob/51a225c5d7f7d8655be62cab0e4d677b6839f35a/lib/placement#L5005:59
yanyanhuI guess we need a better way to manage those ports consumed by different services :)06:01
gmannand this one was enabled for multinode here - https://review.openstack.org/#/c/411510/06:01
gmannyanyanhu: i think we have but not for all services06:01
gmannlet me find wiki i looked last week06:01
yanyanhugmann, so any way to handle this conflict?06:01
yanyanhugmann, thanks :)06:01
Qimingit is at gate side ...06:02
gmannbut not upto date seems :)06:03
gmannit should be via this wiki06:03
gmannlet me talk to sean on this06:03
yanyanhugmann, yes... it not uptodate06:03
yanyanhugmann, great, thank you so much, please just leave us message if there is any action we need to take06:04
gmannmay be you can add for senlin and then palcement API we can change as it was later one06:04
yanyanhugmann, yes, will try update the wiki06:04
gmannlet me change placement API port on devstack06:05
*** guoshan has joined #senlin06:15
Qimingalright, gmann is right, it looks like placement service is 'robbing' port 8778 from us, :)06:30
gmannQiming: yea, changing here hope 8780 not being used - https://review.openstack.org/#/c/412770/06:31
QimingI'm proposing a patch to the openstack-manuals project so that people will know that 8778 is used06:37
yanyanhuQiming, great06:38
yanyanhuI'm trying download the openstack-manual repo, it has costed 30mins and it still in progress...06:38
Qimingit is BIG06:38
yanyanhuyes... need another 30mins I guess06:39
*** dinghh has joined #senlin07:11
*** dinghh1 has quit IRC07:13
openstackgerritmiaohb proposed openstack/senlin: Trivial: Arrange the imported objects in alphabetical order  https://review.openstack.org/41284307:43
openstackgerritShan Guo proposed openstack/senlin: Add parameter in CLUSTER_DEL_NODES action  https://review.openstack.org/41272908:12
openstackgerritShan Guo proposed openstack/senlin: Add parameter in CLUSTER_DEL_NODES action  https://review.openstack.org/41272908:13
openstackgerritShan Guo proposed openstack/senlin: Add parameter in CLUSTER_DEL_NODES action  https://review.openstack.org/41272908:15
*** shu-mutou has joined #senlin08:34
yanyanhuhi, Qiming, around?08:37
yanyanhudo you know what is placement service for? Didn't find this project in project list or git.openstack.org. Is it a sub proj of Nova?08:39
openstackgerritmiaohb proposed openstack/senlin: Trivial:  Arrange the imported objects in alphabetical order  https://review.openstack.org/41284308:39
Qimingnot quite sure08:40
Qimingcould be an extension of nova08:40
shu-mutouHi! Does API for updating resources register timestamp into updated_at?08:43
yanyanhuI saw -1 Sylvain Bauza08:44
yanyanhuhi, shu-mutou, I think so08:44
Qimingsee this: https://github.com/openstack-dev/devstack/blob/51a225c5d7f7d8655be62cab0e4d677b6839f35a/lib/placement#L5008:44
yanyanhuthe timestamp property of updated resource should be refreshed08:44
Qimingin the first few lines08:44
yanyanhuQiming, yes, but still not quite sure what it is for08:45
yanyanhuI saw Sylvain Bauza's comment  in that patch. Anyway, we can consider to disable placement service if we don't need it in our test. Although I feel unhappy that a service is enabled in devstack core lib and breaks our job without any hint.08:45
shu-mutouyanyanhu: thanks!!08:46
yanyanhushu-mutou, you're welcome :)08:46
Qimingsomething related to this: http://specs.openstack.org/openstack/nova-specs/specs/newton/implemented/generic-resource-pools.html08:47
yanyanhuwill wait for more feedback on that patch08:47
gmannQiming: yanyanhu can you join openstack-dev and ping bauzas for port discussion08:56
yanyanhugmann, sure08:56
openstackgerritSylvain Bauza proposed openstack/senlin: DO NOT MERGE: Testing the gate  https://review.openstack.org/41293009:03
*** xuhaiwei has joined #senlin09:04
openstackgerrityunfeng zhou proposed openstack/senlin-dashboard: add CONTRIBUTING.rst  https://review.openstack.org/41296609:22
openstackgerritShu Muto proposed openstack/senlin-dashboard: Add node create/update actions  https://review.openstack.org/39697509:36
*** shu-mutou is now known as shu-mutou-AWAY09:44
*** yanyanhu has quit IRC09:54
*** guoshan has quit IRC10:24
*** chohoor has quit IRC11:00
*** guoshan has joined #senlin11:25
*** guoshan has quit IRC11:29
openstackgerritmiaohb proposed openstack/senlin: [WIP] Versioned request and engine support for credential_update  https://review.openstack.org/41305012:14
*** guoshan has joined #senlin12:26
*** guoshan has quit IRC12:30
*** catintheroof has joined #senlin12:37
*** yanyanhu has joined #senlin12:57
yanyanhuhi, guys, meeting will start in openstack-meeting channel in minutes12:58
yanyanhuhi, meeting has started13:00
*** elynn has joined #senlin13:00
*** lxinhui has joined #senlin13:13
*** yanyan has joined #senlin13:23
yanyanhello, am I still online?13:24
*** yanyanhu has quit IRC13:26
*** guoshan has joined #senlin13:26
*** guoshan has quit IRC13:31
*** Ruijie_ has joined #senlin14:00
yanyanhi, just a quick update, our gate is broken....14:01
yanyanfrom this morning I think14:01
Qiminglast night14:01
Qiming9pm or so14:01
yanyanyea, last night14:01
Qimingbefore some patches and after some patches14:01
yanyanthat is because a new added service "placement" is enabled by default in devstack lib14:02
Qimingsuddenly ... tempest jobs all fail14:02
elynnplacement is what...14:02
yanyanand it uses 8778 port as senlin-api service14:02
yanyanelynn, it part of nova currently14:02
Qiminghopefully, it will be all okay again tomorrow14:03
Qiming[21:39] <sdague> bauzas: ok, cool, just let me know when you have stuff up for review, will ensure we get that in today14:03
yanyanand will be splitted as an invidual API in future based on my understanding14:03
lxinhuiQiming, yanyan14:03
yanyanQiming, yes, hope it will receover tomorrow14:03
Qiming[21:40] <bauzas> I need to cut https://github.com/openstack-dev/devstack/blob/master/files/apache-placement-api.template#L1-L16 and put some directives in the location directive itself14:03
Qiming[21:40] <jroll> ah, I assumed if you only had the location directive it wouldn't be needed14:03
Qiming[21:40] <bauzas> yeah, we're honestly blocking senlin14:03
lxinhuiplease help to explain nowdays14:03
lxinhuisenlin capcaity management fucntions14:03
elynnokay too many projects flying in openstack world now14:03
Qimingyes, they are working hard on that14:03
lxinhuiI remember we need to define the numbers14:04
Qimingplacement is a boy from nova14:04
lxinhuiabout the14:04
yanyanhere are the efforts to address this issue14:04
lxinhuicluster size14:04
lxinhuior something14:04
lxinhuimax, min14:04
Qimingyes? desired_capacity?14:04
lxinhui desired_capacity?14:05
lxinhuiplease share more about the design of desired_capacity14:05
*** zzxwill has joined #senlin14:05
yanyanlxinhui, we do make a little change on it actually14:05
lxinhuithat means the first size14:05
yanyanabout the conception of "desired" capacity14:05
lxinhuiwhen create14:05
yanyanmaybe Qiming can give you a more clear and accurate definition on it14:06
*** catinthe_ has joined #senlin14:06
lxinhuican I use it for reserve purpose?14:06
yanyanwe discussed on it a lot before reaching consensus14:06
Qimingyes, later, it will be increased or decreased when you scale a cluster14:06
lxinhuiCan I use senlin to reserve some reasurces for some tenant?14:07
Qimingno, xinhui, its current meaning is: user desired capacity of a cluster, which may and may not be the actual capacity14:07
Qimingyes, you can create a separate cluster14:07
Qimingsenlin gives you a tool, we don't mandate on the way you use it14:08
Qimingwe considered that use case of reservation14:08
lxinhuibut little bit upgly14:08
Qimingyou can create a standby cluster14:08
lxinhuiyes, I see14:08
Qimingshared by many active clusters14:08
yanyanyes, as Qiming said, that all depends on how you use it14:08
*** catintheroof has quit IRC14:08
lxinhuibut you know for the resource pool management purpose14:08
yanyanhow you use senlin to manage resource14:08
Qimingwhen you want to scale out one of them, you can migrate one out of the standby cluster14:09
lxinhuithere are several numbers user need to set14:09
lxinhuione the max14:09
lxinhuithe other is the mini14:09
yanyanwe provides abundant features for cluster membership management14:09
lxinhuithe third one is the resesered14:09
Qimingyes, you can omit them14:09
Qimingthe 3rd is desired_capacity14:09
lxinhuithe reserved should be less than the max14:09
Qimingreserved will become the 4th14:10
lxinhuiand larger than mini14:10
lxinhuibut have keep the resource can not be used by others14:10
Qimingwhy should I reserve some nodes in a cluster if I'm not using them14:10
lxinhuiI say standy solution is ugly is becuase14:10
Qimingfor reservation, you can create a standby one14:10
yanyanlxinhui, I think we need to define "reserve" first  :)14:11
lxinhuiit waste much before someone really wanna it14:11
Qimingif it is shared, it may be not a waste14:11
Qimingon the contrary, if everyone "reserves" some capacity by themselves, that is a huge waste14:12
yanyanso I feel we just need to ensure users can get the resource promptly when they need it14:13
lxinhuithat is what I mean14:13
yanyanbut that is more about policy I feel14:13
lxinhuithat is not to find a solution14:14
lxinhuibut wanna senlin to fill in some gap14:14
yanyanlxinhui, yes, agree14:14
lxinhuiinstead of just some mitigate14:14
yanyanwe need to provide enough flexibility for users to support their resource management policy14:14
yanyanbut we won't make decision for user14:15
yanyanabout how to make that reservation14:15
lxinhuiwe do not14:15
lxinhuiactaully we can not14:15
yanyanwe let them to decide the strategy14:15
lxinhuiprovide support for waht I am discussed14:15
lxinhuinot strategy14:15
Qimingtell me, if we have min=1, max=10, desired=5, reserve=7, current=614:16
lxinhuibut resoruce calculation instead of real provisioing them14:16
Qimingif I'm scaling out the cluster by 2 nodes14:16
yanyanlxinhui, resource calculation means?14:16
lxinhuikeep the source14:16
Qiminghow should I calculate the new number?14:16
lxinhuifor the user14:16
lxinhuiinstead of really create the vm14:16
Qimingthe way to keep resource is to raise the 'min' value14:16
lxinhuiQiming, easy anwser, 7 can be ensured14:17
lxinhuibut 8 can fail14:17
Qimingthen you mean, max is useless?14:17
lxinhuithat is max14:17
lxinhuijust allow or not14:18
lxinhuibut not ensure or not14:18
Qimingwe cannot reserve without create ...14:18
lxinhuisince virtual resource allow to over claim14:18
lxinhuiQiming, I see14:18
yanyanimo, occupying resource is the only way to really reserve resource :)14:18
yanyanreserve the word itself means "it is only for you"14:18
Qimingthere have been some service proposal on that a long time ago14:19
lxinhuiYanyan, please refer to above mention about poor points of real creation14:19
yanyanotherwise, you can only say, we have higher priority to use it :)14:19
Qimingit is named climate or something14:19
Qimingdoing resource reservation14:19
lxinhuiwhy it failed14:19
Qimingand their have been proposals across community about quota14:20
*** Ruijie__ has joined #senlin14:20
lxinhuifor the limits of resource?14:20
Qimingno one works on it14:20
lxinhuior else14:20
yanyanpoor points, you mean?14:20
lxinhuieveryone still wanna vsphere14:20
Qimingeveryone brings in new idea and say: this is important, this is better14:20
Qimingthen resources distracted14:20
lxinhuican not use senlin14:20
Qimingyou can see projects blooming14:21
lxinhuibut use vsphere directly14:21
yanyanresource redundancy, is one important prerequisite of HA :)14:21
lxinhuiconsider this in another way14:21
Qimingvsphere is good14:21
lxinhuiyes, yanyan14:21
Qimingyou got control of everything14:21
lxinhuimany user rely on this for similar reasons14:21
Qimingopenstack is not14:21
lxinhuiI am asking for everything :)14:22
lxinhuiI am not14:22
lxinhuiI mean14:22
lxinhuijust reserve14:22
lxinhuiAnyway, Thanks, Qiming and Yanyan14:22
yanyanlxinhui, you're welcome14:22
QimingI mean vSphere can control everything14:22
*** Ruijie_ has quit IRC14:22
lxinhuiI will read qiming's code and add more in the next week14:23
yanyanalway pleased to discuss :)14:23
Qimingbut in OpenStack, no service can do that14:23
lxinhuiQiming, actaully not14:23
Qimingit is run by a single company14:23
lxinhuiyou need think it in a bigger stack14:23
yanyanhi, I have to leave. see u guys tomorrow. have a good night14:23
lxinhuiso what I am asking actally is a kind of service14:23
Qimingyou won't come up with two monitoring solutions: ceilometer and monasca14:23
lxinhuieven only API14:23
*** yanyan has quit IRC14:23
Qimingyou won't waste time on wsme, flask, pecan, wsgi all these variants14:24
lxinhuiwhat you are talking just naughty argue14:24
lxinhuiWe can find other time14:24
Qimingit IS a BIG problem of the community14:24
lxinhuidiscuss about API things14:24
lxinhuisee you, have to run for another meeting14:25
Qimingceilometer, when returns version info, does tthis:  {'versions': {'values': []}}14:25
lxinhuiit almost ending14:25
lxinhuibye see you14:25
*** lxinhui has quit IRC14:25
*** guoshan has joined #senlin14:27
*** Ruijie__ has quit IRC14:28
*** Ruijie__ has joined #senlin14:28
*** guoshan has quit IRC14:31
*** zzxwill has quit IRC14:32
*** Ruijie_ has joined #senlin14:50
*** Ruijie__ has quit IRC14:50
*** Ruijie_ has quit IRC15:00
*** zhouzhengxi has joined #senlin15:23
*** guoshan has joined #senlin15:28
*** guoshan has quit IRC15:32
*** zhouzhengxi has quit IRC15:58
*** Drago1 has joined #senlin16:22
*** guoshan has joined #senlin16:29
*** guoshan has quit IRC16:33
*** catintheroof has joined #senlin16:39
*** catinthe_ has quit IRC16:42
*** Drago1 has quit IRC17:01
*** guoshan has joined #senlin17:29
*** guoshan has quit IRC17:34
*** catinthe_ has joined #senlin18:03
*** catintheroof has quit IRC18:06
*** guoshan has joined #senlin18:30
*** guoshan has quit IRC18:35
*** elynn has quit IRC19:01
*** elynn has joined #senlin19:02
*** guoshan has joined #senlin19:31
*** guoshan has quit IRC19:35
*** catintheroof has joined #senlin19:37
*** catinthe_ has quit IRC19:40
*** guoshan has joined #senlin20:32
*** guoshan has quit IRC20:37
*** catintheroof has quit IRC20:47
*** Drago1 has joined #senlin21:32
*** guoshan has joined #senlin21:32
*** guoshan has quit IRC21:37
*** guoshan has joined #senlin22:33
*** guoshan has quit IRC22:38
*** guoshan has joined #senlin23:34
*** guoshan has quit IRC23:39

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