Monday, 2017-03-20

*** pshedimb_ has joined #openstack-watcher01:11
*** pshedimb_ has quit IRC02:34
*** sanfern has joined #openstack-watcher02:51
*** Yumeng has joined #openstack-watcher03:24
*** zhurong has joined #openstack-watcher03:25
openstackgerritlicanwei proposed openstack/watcher master: exception when running 'watcher service list'  https://review.openstack.org/44735403:41
*** zhurong has quit IRC04:16
*** zhurong has joined #openstack-watcher04:59
*** zhurong has quit IRC05:14
*** adisky_ has joined #openstack-watcher05:23
openstackgerritChangBo Guo(gcb) proposed openstack/watcher master: Remove old oslo.messaging transport aliases  https://review.openstack.org/44736905:25
*** karthikpr has quit IRC05:47
*** karthikpr has joined #openstack-watcher06:25
*** karthikpr has quit IRC06:29
*** alexchadin has joined #openstack-watcher07:19
*** Yumeng has quit IRC08:45
*** vincentfrancois1 has joined #openstack-watcher08:51
*** openstackgerrit has quit IRC09:02
*** vincentfrancois1 has quit IRC09:11
*** alexchadin has quit IRC09:27
*** alexchadin has joined #openstack-watcher09:58
*** sanfern has quit IRC10:49
*** Yumeng has joined #openstack-watcher10:49
*** karthikpr has joined #openstack-watcher10:57
*** karthikpr has quit IRC11:01
*** alexchadin has quit IRC11:04
*** vmahe has joined #openstack-watcher11:10
*** zhurong has joined #openstack-watcher11:18
*** alexchadin has joined #openstack-watcher11:33
*** zhurong has quit IRC11:37
alexchadinhi vincentfrancoise11:41
*** alexchadin has quit IRC12:07
*** sanfern has joined #openstack-watcher12:10
*** alexchadin has joined #openstack-watcher12:12
Yumengalexchadin:hi12:33
alexchadinhi Yumeng12:34
Yumengalexchadin:Based on what has been agreed in our last irc meeting, I have spilted add-power-on-off (https://review.openstack.org/#/c/435436)  into three sub-features, so there are two more specs which need your approval.12:34
YumengCan you please have a look and give any feedback so that I can start update related specs?12:35
Yumenghere are the links12:35
Yumenghttps://blueprints.launchpad.net/watcher/+spec/build-baremetal-data-model-in-watcher12:35
Yumenghttps://blueprints.launchpad.net/watcher/+spec/strategy-to-trigger-power-on-and-power-off-actions12:35
Yumengonly these two needs approval12:35
Yumenghttps://blueprints.launchpad.net/watcher/+spec/add-power-on-off       this one is  already approved and I just updated12:36
alexchadinYumeng: let me take a look12:37
Yumengalexchadin: Thanks! you can do it any time at your convenience. if any questions email me at bao.yumeng@zte.com.cn or ping me tomorrow?12:39
alexchadinYumeng: it looks good, but you need to link all these three BP with dependencies12:39
alexchadinYumeng: Strategy depends on Action, Action depends on Ironic Data Model12:40
Yumengalexchadin: OK. Done! you can see the dependencies now12:42
alexchadinYumeng: now it looks perfectly ;)12:43
*** andreaf has joined #openstack-watcher12:44
Yumengalexchadin::Thanks Alex! I have updated them as approved.12:44
Yumengalexchadin: see you then12:45
andreafhello folks - could I have reviews on https://review.openstack.org/#/c/442756/ please?12:45
alexchadinYumeng: have a nice day12:45
alexchadinhi andreaf12:45
andreafalexchadin: hi12:45
andreafalexchadin: see http://lists.openstack.org/pipermail/openstack-dev/2017-February/112938.html12:46
*** Yumeng has left #openstack-watcher12:46
andreafalexchadin: we want refactor scenario tests on Tempest side, so I would ask you to keep what you need of it the watcher tempest plugin - at least until we are able to offer a stable interface around that12:47
alexchadinandreaf: I see, thank you for your work, I will review it now12:47
andreafalexchadin: cool, thank you!12:48
alexchadinandreaf: is it just a copy of tempest.scenario.manager?12:49
andreafalexchadin: yeah pretty much12:49
andreafthe original scenario.manager module includes 4 classes but you only use the first one, so I removed the 3 unused ones12:49
alexchadinandreaf: Have you tried to launch functional tests with it?12:50
andreafalexchadin: I only relied on what runs in the gate12:50
andreafalexchadin: there's a failure on gate-watcher-dsvm-multinode-ubuntu-xenial-nv, but it seems definitely unrelated12:51
andreafthe job is non voting, because of failures I suppose?12:51
alexchadinandreaf: yeah, it fails time by time12:51
alexchadinandreaf: +112:51
alexchadinandreaf: I'm ok with this patch12:53
andreafalexchadin: cool, thanks for reviewing12:53
alexchadinandreaf: +2 from me12:53
alexchadinhi adisky_12:54
adisky_hi alexchadin12:54
alexchadinI've had a discussion with dims about Debian-related packages12:55
alexchadinadisky_: ^12:55
alexchadinadisky_: as you can see, there is a thread that reveals the problem: http://lists.openstack.org/pipermail/openstack-dev/2017-February/112220.html12:56
alexchadinadisky_: no one have taken this work12:56
adisky_alexchadin: ohh12:57
adisky_alexchadin:  so as of now, i will write only for rpm distros13:01
*** adisky_ has left #openstack-watcher13:01
alexchadinadisky_: I will contact zigo to gather info about managing packages in Debian. I hope we can do it by ourselves13:02
vincentfrancoiseandreaf: Hi, I just merged your PS13:05
alexchadinvincentfrancoise: hi13:08
*** openstackgerrit has joined #openstack-watcher13:10
openstackgerritMerged openstack/watcher master: Local copy of scenario test base class  https://review.openstack.org/44275613:10
vincentfrancoisealexchadin: hi13:11
vincentfrancoisealexchadin: since you are here, I want to discuss with you about the ironic integration with Watcher13:12
alexchadinvincentfrancoise: I wanted to discuss Py3.5, but ok13:12
alexchadinvincentfrancoise: let's discuss Ironic firstly ;)13:13
vincentfrancoisealexchadin: as you wish13:13
vincentfrancoisealexchadin: did you see the discussion I had with licanwei at the end of last week?13:13
alexchadinvincentfrancoise: yeap13:13
vincentfrancoisealexchadin: cool so no need for me to re-explain it then :)13:14
alexchadinvincentfrancoise: you propose to extend current CDM?13:15
vincentfrancoisealexchadin: what option do you prefer ? Separate CDM for ironic or integrate ironic into the current CDM?13:15
vincentfrancoisealexchadin: I personally don't mind either way since there's no silver bullet I guess13:16
vincentfrancoisealexchadin: dtardivel if in favour of the separate CDM and I can't remember which one licanwei was preferring13:17
vincentfrancoisealexchadin: yumeng I mean*13:17
alexchadinvincentfrancoise: I re-read your conversation13:18
vincentfrancoisealexchadin: but she was about to carry on with option 2 and add it to the current CDM13:18
vincentfrancoisealexchadin: you have the link? I can give it to you13:18
alexchadinvincentfrancoise: on the one hand, if we add Ironic to the current CDM, we will extremely raise count of objects in graph13:19
*** gabor_antal_km is now known as gabor_antal13:20
alexchadinvincentfrancoise: it may bring a lot of problems with debugging it and maintaining13:20
*** karthikpr has joined #openstack-watcher13:22
alexchadinvincentfrancoise: on the other hand, having two CDM (or more?) will affect on background mechanism13:23
vincentfrancoisealexchadin: yes that's about it13:23
vincentfrancoisealexchadin: the more I think about it, the more I reckon we should go for a separate CDM but that's only a gut feeling13:24
*** karthikpr has quit IRC13:26
alexchadinvincentfrancoise: the easier we make it, the easier our life will be13:26
vincentfrancoisealexchadin: if you don't have an answer yet it's not a problem but maybe that's something we need to discuss in priority on wednesday13:26
vincentfrancoisealexchadin: I agree but both will make our life harder13:27
alexchadinvincentfrancoise: we have two ways to update CDM, right? Using notifications and background job13:27
vincentfrancoisealexchadin: but with N different CDMs, if one goes wrong, it doesn't break everything13:27
vincentfrancoisealexchadin: which is why I would probably go for option 2213:27
vincentfrancoise2*13:28
vincentfrancoisealexchadin: yes13:28
alexchadinvincentfrancoise: actually, I would prefer to leave notifications only way, cause it allows Watcher to be as synced as possible13:29
alexchadinvincentfrancoise: in this case, I would vote for separating CDM13:30
vincentfrancoisealexchadin: the background job can be useful if the notifications do not contain and/or cover all data13:31
alexchadinvincentfrancoise: Doesn't Nova notifications provide all necessary data?13:32
vincentfrancoisealexchadin: but if all the needed info are here (which is the case for nova), then we just need the job once to create the initial CDM and let the notifications handle the rest13:32
alexchadinvincentfrancoise: most of primary OpenStack projects are going to support full-notifications actions.13:33
alexchadinvincentfrancoise: in case of problems, we can contact project developers to ask them for adding some specified infos if it is possible13:34
vincentfrancoisealexchadin: in that case, maybe we can make the background job optional13:35
vincentfrancoisealexchadin: and disable it by default13:35
alexchadinvincentfrancoise: +113:35
vincentfrancoisealexchadin: then create as many CDMs as we want (such as Ironic)13:36
vincentfrancoisealexchadin: the other issue we have with having many CDMs is that the audit scope wouldn't work properly I think13:36
alexchadinvincentfrancoise: what's the issue?13:37
vincentfrancoisealexchadin: well imagine someone creating a new nova model with a slightly different data structure13:38
vincentfrancoisealexchadin: then the scope will not be able to filter a data structure it doesn't know13:38
vincentfrancoisealexchadin: so it would break I think13:38
vincentfrancoisealexchadin: the get_scoped_model() method is strongly related to the CDM we want to scope13:40
vincentfrancoisealexchadin: so maybe we should make it a requirement to implement a "CDM Scoper" for every new CDM we create13:41
vincentfrancoisealexchadin: and say that the one you did is specific to the compute CDM13:41
alexchadinvincentfrancoise: our current scope is based on Nova CDM. Why can't developer write its own Scope for new CDM?13:42
*** karthikpr has joined #openstack-watcher13:42
alexchadinvincentfrancoise: you want to bind scope to CDM?13:43
vincentfrancoisealexchadin: +1, we just need to say it in the docs and refactor the codebase to make it simple to implement a new one13:43
vincentfrancoisealexchadin: yes13:43
vincentfrancoisealexchadin: but the scope schema should always be the same13:44
alexchadinvincentfrancoise: why can't developer extend it in CDM boundaries?13:45
vincentfrancoisealexchadin: A single audit scope schema structure for all CDMs, but one scoping implementation per CDM13:45
*** edleafe- is now known as edleafe13:45
vincentfrancoisealexchadin: ah sorry I wasn't precise enough13:46
vincentfrancoisealexchadin: The developer should be able to add new options13:46
vincentfrancoisealexchadin: but the operator should only see a unified schema that collected all the extra options13:46
vincentfrancoisealexchadin: or something like this13:46
alexchadinvincentfrancoise: one more BP, please ;D13:50
vincentfrancoisealexchadin: i.e. a core schema structure onto which we add extra sections13:50
vincentfrancoisealexchadin: :D13:51
alexchadinWe will rejoice Yumeng, once she returns13:51
*** exploreshaifali has joined #openstack-watcher13:57
vincentfrancoisealexchadin: yes :p13:57
vincentfrancoisealexchadin: what about python 3.5 now?13:57
*** vmahe has left #openstack-watcher13:57
alexchadinvincentfrancoise: Supporting of Py3.5 in OS projects is one of the main goals for Pike cycle13:58
alexchadinvincentfrancoise: I've seen this PS this morning: https://review.openstack.org/#/c/446189/13:58
alexchadinvincentfrancoise: and here is Completion Criteria: https://governance.openstack.org/tc/goals/pike/python35.html13:59
alexchadinvincentfrancoise: I propose to test our functional tests and try to make Py3.5 default interpreter for Watcher14:00
vincentfrancoisealexchadin: do we have to keep python 2.7?14:01
alexchadinvincentfrancoise: for now yes14:01
vincentfrancoisealexchadin: because if so we should have 2 separate tempest jobs I'd say14:01
sballe_morning14:02
alexchadinsballe_: morning14:02
alexchadinvincentfrancoise: is there version-related issues?14:02
vincentfrancoisealexchadin: but the py27 one could run only a subset of the lot14:02
vincentfrancoisealexchadin: no, it's just that most companies still prefer py27 so we shouldn't make sure we don't break anything because we switched to py3514:04
vincentfrancoisesballe_: hi14:04
vincentfrancoisealexchadin: but the py27 one would not have to be a multinode job I'd say14:04
alexchadinvincentfrancoise: EOL of 2.7 is 202014:04
alexchadinvincentfrancoise: I will add it to open discussions14:06
vincentfrancoisealexchadin: I agree, we can remove it in 1 or 2 years then14:06
alexchadinvincentfrancoise: since there are a lot of things to discuss on Wednesday, I will take 30 minutes for Open Discussions14:07
vincentfrancoisealexchadin: IMHO, move the open discussion before the code reviews14:07
alexchadinvincentfrancoise: good proposal :)14:08
andreafvincentfrancoise: thank you!14:27
*** alexchadin has quit IRC15:04
*** exploreshaifali has quit IRC15:08
*** exploreshaifali has joined #openstack-watcher15:23
*** pshedimb_ has joined #openstack-watcher15:37
*** shaifali has joined #openstack-watcher16:45
*** exploreshaifali has quit IRC16:48
openstackgerritVincent Fran├žoise proposed openstack/watcher master: Added tempest test for workload_stabilization  https://review.openstack.org/39335516:59
*** danpawlik has quit IRC17:07
openstackgerritVincent Fran├žoise proposed openstack/watcher master: Added Actuator Strategy  https://review.openstack.org/42511017:15
*** pshedimb_ has quit IRC17:18
*** shaifali has quit IRC17:34
*** brunograz has quit IRC17:37
*** brunograz has joined #openstack-watcher17:39
*** shaifali has joined #openstack-watcher17:49
*** vincentfrancoise has quit IRC17:50
*** sanfern has quit IRC18:03
*** shaifali has quit IRC18:19
*** pshedimb_ has joined #openstack-watcher18:22
*** shaifali has joined #openstack-watcher18:27
openstackgerritChris Spencer proposed openstack/watcher master: Adding tempest test that executes outlet_temp_control strategy.  https://review.openstack.org/43918418:28
*** wootehfoot has joined #openstack-watcher18:42
*** shaifali has quit IRC18:44
*** shaifali has joined #openstack-watcher18:46
*** pshedimb_ has quit IRC20:07
*** pshedimb_ has joined #openstack-watcher20:20
openstackgerritChris Spencer proposed openstack/watcher master: Adding tempest test that executes outlet_temp_control strategy.  https://review.openstack.org/43918420:29
*** shaifali has quit IRC20:32
*** pshedimb_ has quit IRC21:14
*** pshedimb_ has joined #openstack-watcher21:21
openstackgerritChris Spencer proposed openstack/watcher master: Adding tempest test that executes outlet_temp_control strategy.  https://review.openstack.org/43918422:14
*** wootehfoot has quit IRC23:14
*** hidekazu has joined #openstack-watcher23:24

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