Friday, 2017-03-10

hidekazuunfortunately, my session was not chosen.00:02
*** thorst has joined #openstack-watcher00:53
*** zhurong has joined #openstack-watcher00:55
*** yuanying_ has quit IRC01:05
*** yuanying has joined #openstack-watcher01:19
*** thorst has quit IRC01:45
*** sanfern has quit IRC01:53
*** sanfern has joined #openstack-watcher01:54
*** karthikpr has quit IRC01:56
*** karthikpr has joined #openstack-watcher02:04
*** karthikpr has quit IRC02:05
*** yuanying has quit IRC02:20
*** yuanying has joined #openstack-watcher02:21
*** yuanying has quit IRC02:48
*** thorst has joined #openstack-watcher03:13
*** thorst has quit IRC03:18
*** karthikpr has joined #openstack-watcher04:05
openstackgerritlicanwei proposed openstack/watcher master: stale the action plan
*** yuanying has joined #openstack-watcher04:10
*** thorst has joined #openstack-watcher04:14
*** thorst has quit IRC04:18
*** karthikpr has quit IRC04:50
*** hidekazu has quit IRC05:03
*** thorst has joined #openstack-watcher05:15
*** thorst has quit IRC05:19
*** karthikpr has joined #openstack-watcher05:21
*** karthikpr has quit IRC05:43
*** karthikpr has joined #openstack-watcher05:51
*** hidekazu has joined #openstack-watcher05:52
*** karthikpr has quit IRC06:06
*** karthikpr has joined #openstack-watcher06:09
*** karthikpr has quit IRC06:13
*** alexchadin has joined #openstack-watcher06:50
*** thorst has joined #openstack-watcher07:16
*** thorst has quit IRC07:21
openstackgerritMerged openstack/watcher-specs master: Support Description For Dynamic Action
*** alexchadin has quit IRC07:48
*** alexchadin has joined #openstack-watcher07:50
*** thorst has joined #openstack-watcher08:17
*** thorst has quit IRC08:22
*** danpawlik has quit IRC08:33
*** danpawlik has joined #openstack-watcher08:34
*** danpawlik has quit IRC08:41
*** danpawlik has joined #openstack-watcher08:41
*** vincentfrancoise has joined #openstack-watcher08:43
openstackgerritlicanwei proposed openstack/watcher master: stale the action plan
openstackgerritHidekazu Nakamura proposed openstack/watcher master: Added suspended audit state
*** yuanying has quit IRC08:58
*** pksingh has joined #openstack-watcher08:58
sanfernhi alexchadin,09:01
alexchadinhi sanfern09:01
sanfernalexchadin, I was looking at vm_workload_consolidation strategy, period is not an input parameter, is this was intentional09:03
sanfernperiod is always 360009:03
alexchadinlet me see09:08
alexchadinsanfern: yeah, that wasn't updated to latest changes09:09
alexchadinsanfern: I think, it should be fixed09:10
sanfernshall I change to make it to input param09:10
alexchadinsanfern: could you take it out to bug-fix commit?09:10
alexchadinsanfern: we will review it fast09:10
sanfernalexchadin, sure09:11
alexchadinsanfern: period option should be take out to configuration options for this strategy09:11
sanfernfrom all strategy or only vm_workload_stragey09:12
pksinghsanfern: alexchadin i think there is a bug for that09:13
pksinghsanfern: alexchadin let me give you the link09:13
alexchadinsanfern: workload_stabilization is tuned with period09:14
openstackLaunchpad bug 1614021 in watcher "Remove hardcoded strategy attributes and promote them as strategy parameters" [Medium,Triaged] - Assigned to Pradeep Kumar Singh (pradeep-singh-u)09:14
alexchadinpksingh: +109:14
pksinghsanfern: it is assigned to me, you can assign it to you09:14
pksinghalexchadin: :)09:15
*** thorst has joined #openstack-watcher09:18
*** thorst has quit IRC09:23
*** alexchadin has quit IRC09:37
*** Yumeng has quit IRC09:38
*** yuanying has joined #openstack-watcher09:41
*** hidekazu has quit IRC09:54
*** zhurong has quit IRC10:10
*** thorst has joined #openstack-watcher10:19
*** thorst has quit IRC10:23
*** openstackgerrit has quit IRC10:33
*** pksingh has quit IRC10:46
*** necuser__ has quit IRC10:47
*** dtardivel has joined #openstack-watcher10:55
*** necuser__ has joined #openstack-watcher11:01
*** alexchadin has joined #openstack-watcher11:03
*** thorst has joined #openstack-watcher11:20
*** thorst has quit IRC11:24
*** necuser__ has quit IRC11:28
*** openstackgerrit has joined #openstack-watcher11:31
openstackgerritMerged openstack/puppet-watcher master: Inconsistent package_ensure parameter name
*** necuser__ has joined #openstack-watcher11:42
*** alexchadin has quit IRC12:21
*** dtardivel has quit IRC12:23
*** dtardivel has joined #openstack-watcher12:23
sballe_is alexchadin here?12:25
*** thorst has joined #openstack-watcher12:43
*** sanfern has quit IRC12:45
*** sanfern has joined #openstack-watcher12:45
*** necuser__ has quit IRC12:49
*** alexchadin has joined #openstack-watcher12:55
alexchadinhi sballe_12:56
sballe_great your are back12:56
sballe_so we got our talk accepted now we are working hard on making it real and a good demo12:56
sballe_Was other talks accepted12:57
sballe_also you pinged me yesterday12:57
alexchadinsballe_: yeah, cool things. there are 2 presentation with Watcher in the main role ;)12:58
alexchadinHelp telecom operators save energy and reduce cost of data center with Openstack Watcher12:58
sballe_you are on the other talks rigth?12:58
sballe_so you will be presenting as part of the two accepted talks12:58
alexchadinyeap, but if it would be possible I'd like to insert my 5 coins about Watcher in the beginning of yours12:59
sballe_we can add some slides about watcher and I can talk to them12:59
alexchadinyour time is Thu 11  11:00am-11:40am12:59
sballe_I was going ot do that anyway12:59
alexchadinZTE time is Thu 11  11:15am-11:25am12:59
sballe_I can tal to the Watcher slides I have done that many time before13:00
alexchadinit means that I will have only 7-12 minutes to make some announcements and run away to ZTE :D13:00
sballe_oh I would prefer you npt make annoucements during the talk but for me to talk about wtahcer13:00
alexchadinoh, okay!13:01
sballe_We can agree on the content13:01
*** necuser__ has joined #openstack-watcher13:01
sballe_I think it make sense to talk about Watcher and what it is and where it is going in mytalk too13:02
sballe_I will do that13:02
alexchadinto mark context for attendees13:03
sballe_I am very happy with all the momentum we have with Watcher13:03
sballe_100% agree and I was going to do taht13:03
sballe_I'll have you take a look at the slides before they are final just like we have done at past summits with acabot and jwcroppe13:03
alexchadinyes, we will discuss it13:04
sballe_and congrats on getting your talk accepted too13:05
alexchadinsballe_: thank you!13:06
*** edleafe is now known as figleaf13:15
vincentfrancoisealexchadin: so just to clarify the service binding question you told me13:17
vincentfrancoisealexchadin: the problem is about having, say, watcher-applier trigger a decision-engine job which would fail13:18
vincentfrancoisealexchadin: am I correct?13:18
alexchadinwow, new kind of problem :D13:18
alexchadinI meant another thing, when two decision engines are trying to take jobs from DB13:19
vincentfrancoisealexchadin: LOL13:19
alexchadinThey will take jobs twice (if there are 2 DE)13:20
vincentfrancoisealexchadin: where if DE1 sees the job in some sort of a PENDING state, then it would pick it up where DE2 would then see it as ONGOING-ish and it would not handle it13:20
alexchadinWe may bind each job to specified service to make *taking job* process clear13:20
vincentfrancoisealexchadin: just copy/pasting from our conversation ^^13:20
alexchadinyou speaking about apscheduler's states?13:21
vincentfrancoisealexchadin: yes13:21
vincentfrancoisealexchadin: but I don't know how it works in detail13:22
alexchadinhm, we need to know what is inside of job_state13:22
vincentfrancoisealexchadin: so that's something I'm not sure about13:22
vincentfrancoisealexchadin: or have a job "routing" mechanism13:23
alexchadinvincentfrancoise: it is serialized by default, maybe we could use pickle_protocol somehow13:23
alexchadinrouting mechanism sounds horribly13:23
vincentfrancoisealexchadin: job tagging if you prefer :p13:23
alexchadinIt definitely sounds better13:24
vincentfrancoisealexchadin: the job tag would help routing its execution to the correct service13:25
vincentfrancoisealexchadin: hum, this is not an easy thing to solve...13:36
*** alexchad_ has joined #openstack-watcher13:37
*** alexchadin has quit IRC13:38
vincentfrancoisealexchadin: the way I see it would be to create a new jobstore plugin in apscheduler which would enhance the table schema to includes a new "tags" column which would help us filter the job we want to execute13:39
vincentfrancoisealexchad_: for the plugin to override13:39
vincentfrancoisealexchad_: would have to be overriden to actually look like13:40
*** alexchad_ is now known as alexchadin13:41
vincentfrancoise return self._get_jobs(self.jobs_t.c.next_run_time <= timestamp, tags=["decision-engine"])13:41
vincentfrancoisealexchadin: ^13:41
vincentfrancoisealexchadin: the rest would roughly be the same13:42
alexchadinvincentfrancoise: I see13:42
vincentfrancoisealexchadin: but that's a lot more work that I would have expected13:42
alexchadinvincentfrancoise: Don't you think that writing new plugin would be redundant?13:43
vincentfrancoisealexchadin: the new plugin would inherit from  SQLAlchemyJobStore and only override the needed methods13:43
vincentfrancoisealexchadin: so it should remain fairly small13:44
alexchadinvincentfrancoise: and the second point, we don't know whether apscheduler mark *in progress* jobs somehow or not.13:44
alexchadinvincentfrancoise: how would we distribute this plugin?13:45
vincentfrancoisealexchadin: then this would also have to be coded in our custom plugin I guess...13:45
vincentfrancoisealexchadin: for the distribution we simply declare it locally in watcher13:46
vincentfrancoisealexchadin: and apscheduler will see it13:46
vincentfrancoisealexchadin: like any other stevedore plugin we have ;)13:47
alexchadina long work day, you know ;)13:47
vincentfrancoisealexchadin: no worries :p13:47
vincentfrancoisealexchadin: the big drawback is that if we add additional state and tags in the DB schema13:48
vincentfrancoisethat means only the new plugin will be usable by Watcher too13:49
alexchadinyes, it will be watcher-only plugin with ~0 chances that it would be used somewhere else13:50
vincentfrancoisealexchadin: exactly13:50
vincentfrancoisealexchadin: so everyone would have to be okay with that13:50
alexchadinso goes the world :)13:50
alexchadinvincentfrancoise: I propose not to detail the content of plugin in spec cause it can be changed during code reviews.13:53
alexchadinvincentfrancoise: I will mention planned changes regarding plugin13:54
vincentfrancoisealexchadin: IMHO, it's not directly related to your work13:54
alexchadinvincentfrancoise: you mean cron syntax?13:55
vincentfrancoisealexchadin: so IMHO, this is a separate BP that treats the problem of HA13:55
vincentfrancoiseat the decision-engine+background job level13:55
vincentfrancoisealexchadin: it's just that the HA "epic" needs to have an extra BP13:56
alexchadinthen I need to restructure blueprint dependencies13:56
vincentfrancoisealexchadin: maybe yes13:56
vincentfrancoisealexchadin: becuse the problem is actually already there anyway with the interval system13:57
*** alexchadin has quit IRC14:09
*** necuser__ has quit IRC14:55
*** necuser__ has joined #openstack-watcher15:07
*** karthikpr has joined #openstack-watcher15:35
*** karthikpr has quit IRC15:36
*** karthikpr has joined #openstack-watcher15:39
*** karthikpr has quit IRC15:40
*** karthikpr has joined #openstack-watcher15:53
*** karthikpr has quit IRC16:07
*** gabor_antal has quit IRC16:13
*** thorst is now known as thorst_afk16:27
*** jimbaker has quit IRC16:39
*** jimbaker has joined #openstack-watcher16:42
*** jimbaker is now known as Guest6701016:42
*** Guest67010 is now known as jimbaker16:43
*** jimbaker has joined #openstack-watcher16:43
*** logan_ has joined #openstack-watcher16:56
*** logan_ is now known as Guest2958916:57
*** karthikpr has joined #openstack-watcher16:59
*** logan- has quit IRC16:59
*** necuser__ has quit IRC17:03
*** Guest29589 is now known as logan-17:03
*** karthikpr has quit IRC17:05
*** necuser__ has joined #openstack-watcher17:15
openstackgerritSanthosh Fernandes proposed openstack/watcher master: Add period input parameter to vm workload consolidation and outlet temp control strategy.
*** vincentfrancoise has quit IRC17:49
*** karthikpr has joined #openstack-watcher17:51
*** pksingh has joined #openstack-watcher17:58
*** karthikpr has quit IRC18:18
*** wootehfoot has joined #openstack-watcher18:24
*** sanfern has quit IRC18:25
*** pksingh has quit IRC18:34
*** karthikpr has joined #openstack-watcher18:48
*** karthikpr has quit IRC18:53
*** karthikpr has joined #openstack-watcher19:38
*** karthikpr has quit IRC19:57
*** karthikpr has joined #openstack-watcher19:59
*** karthikpr has quit IRC20:06
*** karthikpr has joined #openstack-watcher20:07
*** dtardivel has quit IRC20:30
*** karthikpr has quit IRC20:54
*** thorst_afk has quit IRC21:26
*** thorst_afk has joined #openstack-watcher21:44
*** karthikpr has joined #openstack-watcher22:16
*** karthikpr has quit IRC22:49
*** thorst_afk has quit IRC23:27
*** ss4 has joined #openstack-watcher23:38
*** wootehfoot has quit IRC23:41
*** wootehfoot has joined #openstack-watcher23:43
*** ss4 has quit IRC23:46
*** ss4 has joined #openstack-watcher23:47
*** Metabulation has joined #openstack-watcher23:49
*** wootehfoot has quit IRC23:49
*** ss4 has quit IRC23:52
*** Metabulation has quit IRC23:54

Generated by 2.14.0 by Marius Gedminas - find it at!