Thursday, 2017-03-16

*** yuanying has quit IRC00:10
*** yuanying has joined #openstack-watcher00:15
openstackgerritYumengBao proposed openstack/watcher master: Use https to access git
*** sanfern has quit IRC01:04
*** sanfern has joined #openstack-watcher01:05
openstackgerritYumengBao proposed openstack/watcher master: Use https to access git
*** sanfern has quit IRC01:39
*** pshedimb has quit IRC01:56
*** pshedimb has joined #openstack-watcher02:02
*** karthikpr has joined #openstack-watcher02:15
*** karthikpr has quit IRC02:20
*** pshedimb has quit IRC02:45
*** yuanying has quit IRC03:06
*** zhurong has joined #openstack-watcher03:44
*** sanfern has joined #openstack-watcher04:03
*** adisky_ has joined #openstack-watcher04:17
*** yuanying has joined #openstack-watcher04:46
*** pshedimb_ has joined #openstack-watcher05:14
*** zhurong has quit IRC05:25
*** adisky_ has quit IRC06:18
*** adisky_ has joined #openstack-watcher06:18
*** hidekazu has quit IRC06:21
*** adisky_ has quit IRC06:30
openstackgerritBéla Vancsics proposed openstack/watcher master: Reduced the code complexity
*** alexchadin has joined #openstack-watcher07:13
*** hidekazu has joined #openstack-watcher07:26
*** pshedimb_ has quit IRC07:33
*** adisky_ has joined #openstack-watcher08:05
*** vincentfrancoise has joined #openstack-watcher08:14
*** dtardivel has joined #openstack-watcher08:15
dtardivelYumeng: hi08:15
*** yuanying has quit IRC08:19
adisky_alexchadin: hi08:22
alexchadinhi adisky_08:22
adisky_its about installation guide ine openstack manuals.08:23
adisky_do we need make linux distro specific packages for watcher??08:24
alexchadinadisky_: good question08:29
adisky_alexchadin: :|08:30
alexchadinadisky_: I think yes, since there are different package management software: yum and apt-get08:32
adisky_alexchadin: yes, i can make packages but dont know how to upload them08:32
alexchadinadisky_: being on your side, I would ask openstack-manuals team about it08:33
alexchadinadisky_: I think, they will give you good starting point here08:33
adisky_alexchadin: thnx, i will try to search if there is a guide available.08:34
alexchadinadisky_: take a look
adisky_alexchadin: ok08:36
*** yuanying has joined #openstack-watcher08:55
YumengI was just having a meeting09:27
dtardivelYumeng: yesterday, on IRC meeting, we wrote about splitting your spec in 3 different specs. Are you agree with this proposition ?09:29
YumengI am OK with that.09:30
dtardivelOk, great :)09:30
dtardivelYumeng: when you will write the spec about how to build a 'baremetal' data model, could you explain the prerequisite that, in Ironic, node should be created with the extra metadata hypervisor ID.09:33
Yumengdtardivel: do we use ironic? and following this  From a Watcher point of view, It comes down to this: . Implements a new Action to power off/on a physical server by using Ironic API . Build a new baremetal (ironic) data model . Implement a new strategy based on 'baremetal' and 'compute' data models, which could trigger 'compute' and 'baremetal' actions.09:33
vincentfrancoiseYumeng: we could also extend the current data model to add a baremetal dimension09:34
dtardivelvincentfrancoise: I not fan of your proposal.09:35
vincentfrancoiseYumeng: as suggested during the PTG ( line 253)09:35
vincentfrancoisedtardivel: ^09:35
dtardivelI prefer to build different data models, and use some of them in strategy level09:36
vincentfrancoiseI personally don't care but I was just referring to what seemed to have been decided during the PTG09:36
Yumengdtardivel: ok. I will propose new specs.09:37
vincentfrancoisealexchadin: hi09:38
alexchadinhere is brainstorming-etherpad:
alexchadinI've also added it to Forum wiki:
alexchadinfeel free to share your thoughts and questions09:39
dtardivelYumeng: vincentfrancoise: is the new unified data model will be a new plugin ?09:44
vincentfrancoisedtardivel: I don't really know what's been said on the matter at the PTG09:45
vincentfrancoisedtardivel: but the way Intel designed it  is actually within the current CDM graph model09:46
vincentfrancoisedtardivel: I guess you would have some optional branches of the graph that would be built if the service is up09:47
vincentfrancoisedtardivel: but this is still unclear to me09:47
*** alexchadin has quit IRC09:49
openstackgerritPradeep Kumar Singh proposed openstack/watcher master: Prevent the migration of VM with 'optimize' False in VM metadata
Yumengvincentfrancoise:dtardivel: I just quickly go over   we agreed that we should have a separate CDM for storage(line 239), and for other topics concerning CDM we just adding new data model into current CDM (line 243,253)09:59
Yumeng vincentfrancoise:dtardivel: and I got a question: when we add more and more data models into current CDM, it might get very large. Will this affect efficiency of watcher function? or any other impact?10:05
vincentfrancoiseYumeng: The model is maintain in the background, so this background processing will grow slower10:06
dtardivelYumeng: yes it was one of my remarks during the PTG10:07
vincentfrancoiseYumeng: but the execution of a strategy only consist in cloning the current representation at execution time10:07
vincentfrancoiseYumeng: so the execution time of an audit should grow fairly linearly as compared to the size of the infra10:08
vincentfrancoiseYumeng: although that's a guess that is difficult to accurately assess10:09
Yumengvincentfrancoise:dtardivel: I am interested in the unified model of the CDM, my only concern is the potential impact. dtardivel, you just mentioned "is the new unified data model will be a new plugin ?" do you have any ideas? Is there any way we might have to reduce this impact as much as possible?10:15
vincentfrancoiseYumeng: having multiple models just means splitting the problem in many graph which will all run in the background together10:16
Yumengvincentfrancoise: so finally these two ways are expected to have same impact if there is any10:18
vincentfrancoiseYumeng: the same I don't know, but the same order of magnitude is probable10:18
Yumengvincentfrancoise: Make sense to me.10:19
vincentfrancoiseYumeng: by making separate CDMs, you can play around with their refresh rate to slightly lower their maintenance cost but that would be at the expense of its correctness10:21
vincentfrancoiseYumeng: but I would guess that doing such a thing will only provide a slight gain10:21
vincentfrancoiseYumeng: codewise both approaches have advantages and drawbacks10:22
vincentfrancoiseYumeng: 1 CDM means that you can do CDM filtering more effectively because 1 graph = 1 filtering logic10:23
vincentfrancoiseYumeng: but splitting into many CDMs reduces the complexity of each model which eases the maintenance / the number of possible bugs10:24
vincentfrancoisethat's a discussion to have with alexchadin too I guess10:25
Yumengvincentfrancoise:dtardivel: Fair enough. Thx vincent. So currently, I will follow what has been agreed in 253).10:27
vincentfrancoiseYumeng: what's important with supporting Ironic is to also find the notifications they emit, and listen to them to update the CDM10:29
vincentfrancoiseYumeng: because I personally will downvote any information that doesn't get updated via a notification10:30
vincentfrancoiseYumeng: because the risk of staleness would be too big10:30
Yumengvincentfrancoise: I agree. Will you have more discussion with Alex about CDMs? next time when you discuss, please let me know. and BTW shall we put this into open discussion in our next irc meeting  ?10:35
vincentfrancoiseYumeng: maybe the IRC open discussion is a good opportunity?10:37
vincentfrancoiseYumeng: if I see him and also have time, I willm but if you disuss with him too, that'd be good too :)10:38
Yumengvincentfrancoise: I think so. but we usually don't have much time left for open discussion  haha10:38
Yumengvincentfrancoise: No problem, I will.10:40
Yumengvincentfrancoise:see you! time to go home for dinner :)10:40
vincentfrancoiseYumeng: enjoy your meal10:41
Yumengvincentfrancoise: :)10:41
*** Yumeng has quit IRC10:41
*** sanfern has quit IRC10:48
*** sanfern has joined #openstack-watcher11:38
*** pkoniszewski has left #openstack-watcher13:54
*** vincentfrancois1 has joined #openstack-watcher13:55
*** vincentfrancois1 has quit IRC14:35
*** logan- has quit IRC14:46
*** pshedimb_ has joined #openstack-watcher14:46
*** logan- has joined #openstack-watcher14:46
*** danpawlik has quit IRC14:55
*** pshedimb_ has quit IRC14:57
*** karthikpr has joined #openstack-watcher15:04
*** vincentfrancois1 has joined #openstack-watcher15:53
*** pshedimb_ has joined #openstack-watcher16:00
*** karthikpr has quit IRC16:23
*** karthikpr has joined #openstack-watcher16:27
*** karthikpr has quit IRC16:38
*** vincentfrancois1 has quit IRC16:40
*** karthikpr has joined #openstack-watcher17:01
*** karthikpr has quit IRC17:12
openstackgerritMerged openstack/watcher master: Updated from global requirements
*** sanfern has quit IRC17:45
*** karthikpr has joined #openstack-watcher17:57
*** aspiers has quit IRC18:23
*** pshedimb_ has quit IRC18:26
*** karthikpr has quit IRC18:31
*** aspiers has joined #openstack-watcher18:35
*** pshedimb_ has joined #openstack-watcher18:51
*** karthikpr has joined #openstack-watcher19:00
*** karthikpr has quit IRC19:06
*** dtardivel has quit IRC19:10
*** karthikpr has joined #openstack-watcher19:11
*** pshedimb_ has quit IRC19:40
*** pshedimb_ has joined #openstack-watcher19:50
*** karthikpr has quit IRC19:50
*** hidekazu has quit IRC20:25
*** karthikpr has joined #openstack-watcher20:26
*** karthikpr has quit IRC20:27
*** karthikpr has joined #openstack-watcher20:27
*** karthikpr has quit IRC21:28
*** pshedimb_ has quit IRC22:23
*** pshedimb_ has joined #openstack-watcher22:35
*** wootehfoot has joined #openstack-watcher22:57
*** pshedimb_ has quit IRC23:08
*** hidekazu has joined #openstack-watcher23:23
*** karthikpr has joined #openstack-watcher23:30
*** pshedimb_ has joined #openstack-watcher23:36
*** pshedimb_ has quit IRC23:42

Generated by 2.14.0 by Marius Gedminas - find it at!