Monday, 2016-09-26

*** elynn has joined #senlin00:19
*** elynn has quit IRC00:20
*** zhurong has joined #senlin01:14
*** haiwei_ has joined #senlin01:15
*** zhurong has quit IRC01:19
*** zhurong has joined #senlin01:20
haiwei_hi Qiming01:23
haiwei_Want to ask you a question about cluster-scale-in01:24
haiwei_when scale-in finished, Senlin only return a action id, right?01:24
Qimingwhen scale-in action is created, senlin engine/api returns the action id01:25
Qimingusually, it won't take long for that action to be executed01:25
haiwei_we have a use case now, that is using senlin to manage Swarm cluster, when vm nodes are deleted, want to unregister the node from swarm01:25
Qimingokay, that sounds like a deletion policy01:25
Qimingyou may want to do the deregistration in a separate policy *before* the node is actually deleted01:26
*** yanyanhu has joined #senlin01:27
Qiming * Requirements              : Tony Breeds01:28
Qiming * Sahara                    : Vitaly Gridnev01:28
Qiming * Searchlight               : Steve Mclellan01:28
Qiming * Security                  : Robert Clark[*]01:28
Qiming * Senlin                    : Yanyan Hu01:28
Qiming * Solum                     : Devdatta Kulkarni01:28
Qiming * Stable_Branch_Maintenance : Tony Breeds01:28
Qiming * Swift                     : John Dickinson01:28
QimingPTL election results01:28
Qimingcongratulations, yanyan01:28
haiwei_in the deletion policy, senlin can notice swarm which node will be deleted?01:28
haiwei_congrets, yanyan01:28
QimingI mean you can add such logics01:28
yuanyingCongrats yanyanhu01:29
Qimingor you can develop a new signoff policy or something alike01:29
Qimingdeletion policy is evaluated after scaling policy, based on their relative priority01:29
Qimingscaling policy has priority 100, deletion policy has priority 40001:30
Qimingthat means, when deletion policy is evaluated, it might have known the candidates to delete, or it is the deletion policy's resposibility to select candidates for deletion01:31
Qimingafter deletion policy is evaluated, we expect that the candidates must have been determined01:31
Qimingso, before returning from the pre_op method, you can deregister the canidate node from swarm01:32
yanyanhuthanks, guys01:32
yanyanhumore responsibility it means01:32
yanyanhuQiming, may often bother you for some guides :)01:33
haiwei_Qiming, that means I have to modify senlin's code?01:33
Qiminghaiwei, there are two ways01:33
Qimingyou can modify senlin's deletion policy code01:33
Qimingand we will help evaluate if the change is generic enough to other use cases and merge it if so01:33
haiwei_Qiming, makes sense to me, thanks01:34
Qimingif your assessment is that the use case is pretty limited, you can develop your own policy01:34
Qimingit is fine to keep it to yourself01:34
Qimingsenlin has an open architecture design that allows developers to come up with policies/profiles for some special use cases01:35
Qimingthese special policies can be kept at your hands01:35
Qimingor maybe in future we should consider adding a /contrib directory so that people can share their practices01:35
Qimingyet another note01:36
Qimingwe have some plans to workout generic notifications, i.e. whenever some important actions are taken by senlin, senlin will send out versioned notifications01:36
Qimingif swarm can be triggered by such notifications, that would be another channel01:37
yuanyingsounds interesting01:37
Qimingthanks, yuanying01:37
Qimingas an operator/user, you always want to get notified when something changed, even without looking at the log file01:38
yuanyingWe plan to use senlin for our company's booth demo01:38
Qimingoh, really? cool01:38
Qimingplease let team know if you need helps, :)01:38
*** elynn has joined #senlin01:38
haiwei_yes, you can go to NEC's booth in Barcelona summit01:39
yuanyingthanks! your suggestion is very helpful01:39
Qimingfor demo's purpose, I'd like to propose something related to senlin-dashboard01:39
Qimingyou know, I'm not a GUI developer01:39
Qimingbut I know there are quite some important features missing from senlin-dashboard01:39
Qimingfor example, the various cluster operations are not supported: scale-in, scale-out, add-nodes, del-nodes, resize ...01:40
Qimingwhen ttx asked me if we what to bump senlin-dashboard to 1.0.0, I answered no01:40
Qimingbecause without those features, the web UI is not yet fully exhibiting senlin's capability01:41
Qimingany contributions on that front would be highly appreciated, :)01:41
haiwei_in fact there is a UI profesional in our team :)01:42
Qimingshu-muto ?01:42
Qimingdon't know if he has cycles01:42
*** shu-mutou has joined #senlin01:43
haiwei_I am asking him01:43
Qimingmorning, shu-san01:43
shu-mutouHi, morning!01:44
haiwei_morning, UI man01:44
*** elynn has quit IRC01:45
Qimingshu, we have been talking about some missing features in senlin-dashboard01:45
*** elynn has joined #senlin01:45
Qiming[09:39] <Qiming> but I know there are quite some important features missing from senlin-dashboard01:46
Qiming[09:40] <Qiming> for example, the various cluster operations are not supported: scale-in, scale-out, add-nodes, del-nodes, resize ...01:46
Qiming[09:40] <Qiming> when ttx asked me if we what to bump senlin-dashboard to 1.0.0, I answered no01:46
Qiming[09:41] <Qiming> because without those features, the web UI is not yet fully exhibiting senlin's capability01:46
Qiming[09:41] <Qiming> any contributions on that front would be highly appreciated, :)01:46
shu-mutouQiming: OK, I'll contribute in next cycle!! Unfortunately, I don't have time next several weeks.01:50
Qimingthanks, shu-san01:50
yanyanhuhi, Qiming, the integration test finally worked. Just now nova server cluster test failed for some reason. Will try to fix it today. Then we can rely on it to verify some changes related to driver.01:51
Qimingis integration gate triggered everytime?01:52
yanyanhuBTW, the doc-only change won't trigger gate jobs now. I think it will be helpful to reduce the overhead of gate01:52
yanyanhuQiming, yes, it is now01:52
Qimingthat is nice01:52
yanyanhubut it doesn't vote01:52
yanyanhuQiming, yes, hope that will be helpful01:52
Qimingokay, let's see if it is helping or hurdling our progress01:52
Qimingthanks for the gate job filtering work01:53
Qimingthat one really helps us and the community01:53
yanyanhumy pleasure :)01:53
QimingI have just seen that Adrian is becoming the next Magnum PTL01:54
haiwei_yes, sorry for hongbin01:54
Qiminghere is hongbin's proposal01:55
QimingI was a little bit concerned about it01:55
Qimingthe items he proposed fell well into senlin's scope01:55
yanyanhuany chance to talk with them about coordination for these items?01:56
Qimingthere are even ideas about "node groups" ...01:56
yanyanhue.g. add a driver for senlin for cluster management?01:56
Qimingthat sound like a waste of life for Magnum team01:56
yanyanhuyes, I guess this is something tc want us to avoid in their discussion before01:57
yanyanhureally need more discussion and collaboration I feel01:58
Qimingyep, if they have a very good reason to invent a new 'clustering' thing, that is fine01:58
openstackgerritlvdongbing proposed openstack/senlin: Fix the valid keys for filtering event list in doc
Qimingthe only thing I am not sure is whether this is some team-wide understanding for Magnum01:59
yanyanhumaybe we can get a chance to make a cross-team meetup in barcelona?01:59
yanyanhufor this topic01:59
Qiminghope so02:00
Qimingfirst we need to convince them02:00
Qimingmy feeling was based on the team's previous attempt to avoid using barbican02:01
Qimingthey just try to avoid dependence on any other services as far as I can see it02:01
Qimingthe bright side is that senlin has been working on the right things, useful features for any people who want to manage resource pools02:02
Qimingand I believe we will stick to that goal02:02
yanyanhuyes, obsolutely02:02
Qimingwe have envisioned the life cycle management problem, and the need about hetero/homo-geneous clusters02:03
Qimingmaybe we still have some gaps to close, but we are already heading the right direction02:04
QimingAdrian's proposal: ... During the Ocata cycle we will work together on enabling the next generation02:04
Qimingof applications with a design for complex clusters, and better integrating02:04
Qimingour clusters with OpenStack services for networking and storage. ...02:04
Qimingany suggestions to initiate a constructive discussion with Magnum?02:05
yanyanhuI guess maybe we can ask yuan ying or hongbin to help to bridge?02:06
Qimingdon't know if it is appropriate to send emails to the core team directly02:07
*** ruijie has joined #senlin02:07
yanyanhuumm, not sure about it...02:07
Qimingintroducing senlin to them, if they haven't tried it out02:07
yanyanhuor maybe start a new thread about this topic in mailing list?02:11
Qimingnot all people checks mailinglist or express their opinions there02:12
Qimingmailinglist is an option though02:12
yanyanhuyes, that's true...02:12
Qimingor we can post to ML and cc to the core team02:13
*** R_lixh has joined #senlin02:13
yanyanhuyes, that sounds good02:13
yanyanhumore formally and can also ensure they can get the message02:13
openstackgerritYanyan Hu proposed openstack/senlin: Fix integration test
*** yuanying has quit IRC02:52
openstackgerritlvdongbing proposed openstack/python-senlinclient: Fix the issue timestamp field is none in event list
elynnyanyanhu, Qiming  I will take vacation since tomorrow, I will be on plane when the senlin meeting starts, I'm afraid I can't attend it..02:55
Qimingcannot get online when on plane?02:55
elynnI will try if there is wifi02:56
yanyanhuit's ok.03:02
yanyanhuhave a good vacation03:03
openstackgerritMerged openstack/senlin: Fix the valid keys for filtering event list in doc
Qimingelynn, was kidding03:08
Qimingenjoy your vacation, :)03:08
openstackgerritQiming Teng proposed openstack/senlin: Fix CONF.set_override for type enforcement
R_lixhWhen will you guys be Barcelona?03:29
Qiminghope so03:29
Qimingtrip request pending approval03:29
Qiminghow about you?03:29
R_lixhI submmited it to amdin just now03:30
R_lixhMaybe go to embassay to provide finger prints tomorrow03:30
R_lixhif no materisal missing03:30
R_lixhpending for approval of manager or embassay03:31
R_lixhMaybe Qiming you does not need figer prints since you traveld to Euro last year03:32
*** ruijie has quit IRC03:33
*** ruijie has joined #senlin03:38
openstackgerritMerged openstack/senlin: Fix integration test
*** yuanying has joined #senlin03:47
*** yuanying has quit IRC03:49
*** yuanying has joined #senlin03:55
*** elynn has quit IRC04:26
openstackgerritOpenStack Proposal Bot proposed openstack/python-senlinclient: Updated from global requirements
*** elynn has joined #senlin05:03
*** elynn has quit IRC05:08
*** elynn has joined #senlin05:08
openstackgerritYanyan Hu proposed openstack/senlin: Creating new keypair for integration test
*** chenfei_ has quit IRC05:20
*** chenfei has joined #senlin05:21
*** xiaozhuangqing has quit IRC06:29
openstackgerritYanyan Hu proposed openstack/senlin: Creating new keypair for integration test
*** xiaozhuangqing has joined #senlin06:43
*** xiaozhuangqing has quit IRC07:20
*** xiaozhuangqing has joined #senlin07:44
openstackgerritmiaohb proposed openstack/senlin: Revise the actions.rst
openstackgerritYanyan Hu proposed openstack/senlin: Specifying proper subscription ttl
*** shu-mutou is now known as shu-mutou-AFK08:10
openstackgerritmiaohb proposed openstack/python-senlinclient: Revise the help of "action list"
openstackgerritXueFeng Liu proposed openstack/senlin: Fix service manage cleanup
Qimingfor zaqar TTL, they even don't allow a value of -1 for example to annotate infinite expiration time?08:39
yanyanhuyes... have confirmed with them08:39
Qimingwhat are the 'from' and 'subject' fields?08:40
yanyanhutheir explanation is they need a value to check timeout although I didn't quite understand it08:40
yanyanhuQiming, thoes two fields are not positional I think08:41
Qimingnot positional?08:41
yanyanhujust for information08:41
yanyanhuyes, we can omit them when creating subscription08:41
Qimingso you mean they are optional?08:42
yanyanhulet me make a double check08:42
yanyanhuah, yes08:42
Qimingare we gonna check the 'subject' field when receiving a message?08:42
Qimingor we will never check the subscription again before we delete it?08:43
yanyanhuQiming, no, actually they are more useful for 'mailto' type of subscriber08:43
yanyanhuQiming, currently, we never check it08:44
Qimingem, okay08:44
yanyanhuI think we can remove it actually08:44
yanyanhusince we haven't find the case to use it yet08:44
QimingI'm a little bit concerned with the freeform text we assigned to 'subject'08:44
Qimingfor example08:44
Qimingif we want to use it, maybe we can just assign the receiver ID to it08:45
Qimingand get rid of 'Senlin receiver ' part08:45
yanyanhuI see08:45
yanyanhuor maybe we don't specify those two fields in current stage08:46
yanyanhuthe example given in zaqar API ref is not very clear actually...08:46
yanyanhuit is more about how those fields works for 'mailto' subscriber08:47
yanyanhulet me just remove them for we are actually using http subscriber08:47
yanyanhuoh, BTW, Qiming, I have created an etherpad to track the idea for our presentation in summit.
yanyanhuhave sent the link to eldon to ask his idea08:52
yanyanhubut he said he may not be able to join the summit08:52
yanyanhubut junwei will be there08:52
yanyanhuhe said he will read the etherpad tomorrow and give his feedback08:53
openstackgerritmiaohb proposed openstack/senlin: Revise the bindings.rst
openstackgerritYanyan Hu proposed openstack/senlin: Specifying proper subscription ttl
openstackgerritxiaozhuangqing proposed openstack/senlin: Use LOG.warning replace LOG.warn
openstackgerritmiaohb proposed openstack/senlin: Revise the clusters.rst
*** yanyanhu has quit IRC09:51
openstackgerritQiming Teng proposed openstack/senlin: Add to_json_schema support to Object fields
ruijieHi, Qiming09:55
ruijieI'll propose a patch about the batch policy09:55
ruijieIt's a basic version about my understanding of this policy09:56
ruijieCan you help me review it09:56
*** elynn has quit IRC09:58
*** zhurong has quit IRC10:01
*** xiaozhuangqing_ has joined #senlin10:03
*** xiaozhuangqing has quit IRC10:03
*** xiaozhuangqing_ is now known as xiaozhuangqing10:04
openstackgerritMerged openstack/senlin-dashboard: Use isotime format in senlin-dashboard
openstackgerritRUIJIE YUAN proposed openstack/senlin: Support batch policy
*** ruijie has quit IRC10:25
*** zhurong has joined #senlin11:59
openstackgerritXueFeng Liu proposed openstack/senlin: change cb to cm
*** zhurong has quit IRC12:09
*** zhurong has joined #senlin12:11
*** elynn has joined #senlin13:51
*** zhurong has quit IRC14:01
*** qwebirc85264 has joined #senlin14:17
*** qwebirc85264 has quit IRC14:18
*** Drago has joined #senlin14:46
*** openstackstatus has joined #senlin15:28
*** ChanServ sets mode: +v openstackstatus15:28
-openstackstatus- NOTICE: Earlier job failures for "zuul-cloner: error: too few arguments" should now be solved, and can safely be rechecked15:43
-openstackstatus- NOTICE: Launchpad SSO is offline, preventing login to, and many other sites; no ETA has been provided by the LP admin team15:49
*** ChanServ changes topic to "Launchpad SSO is offline, preventing login to, and many other sites; no ETA has been provided by the LP admin team"15:49
*** ChanServ changes topic to "IRCLog: | Bugs: | Review:,n,z"16:12
-openstackstatus- NOTICE: Launchpad SSO logins are confirmed working correctly again16:12
*** elynn has quit IRC16:40
*** Drago has quit IRC22:52
*** zhurong has joined #senlin23:48

Generated by 2.14.0 by Marius Gedminas - find it at!