08:00:59 <licanwei> #startmeeting watcher
08:01:00 <openstack> Meeting started Wed Jul  3 08:00:59 2019 UTC and is due to finish in 60 minutes.  The chair is licanwei. Information about MeetBot at http://wiki.debian.org/MeetBot.
08:01:01 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
08:01:02 <chenke> Hi
08:01:04 <openstack> The meeting name has been set to 'watcher'
08:01:09 <Dantalion> hi
08:01:15 <licanwei> hi
08:01:15 <chenke> Long time no see.
08:01:40 <licanwei> #topic Announcements
08:02:06 <licanwei> Optimized Nova API calls for data model show between 30 / 40% performance improvement
08:02:17 <licanwei> good news :)
08:02:42 <Dantalion> Yes indeed, with the limit set performance is now much better then before
08:05:51 <Dantalion> Possibly we can get better performance when doing multiple calls in parallel but for now this is fine
08:06:10 <chenke> Yes.
08:06:29 <licanwei> OK, next topic
08:06:38 <licanwei> #topic Review
08:07:16 <licanwei> https://review.opendev.org/667522 improve OptGroup consistency across configuration needs review
08:07:42 <chenke> I think this patch is ok.
08:07:47 <chenke> +1
08:08:10 <licanwei> +1
08:08:32 <chenke> Detailed information is always good in the documentation.
08:08:57 <Dantalion> Okaj, i'll add the workflow after the meeting
08:09:12 <licanwei> https://review.opendev.org/660011 Add reource_name in action input parameter field needs review
08:09:59 <licanwei> maybe need a new patch
08:11:07 <licanwei> I'll review after meeting
08:11:54 <chenke> Ok.
08:12:14 <licanwei> https://review.opendev.org/668005 Improve logging in building of nova data model needs review
08:12:33 <licanwei> +1 for review
08:13:14 <Dantalion> I'll add the space to the logging string today
08:13:53 <chenke> Ok. I will learn this patch tomorrow.
08:14:10 <licanwei> Ok, next
08:14:31 <licanwei> https://review.opendev.org/668602 Remove notifier_driver option in Watcher devstack needs review
08:15:37 <Dantalion> Gerrit is taking a very long time to load
08:15:45 <chenke> yes.
08:15:53 <chenke> gerrit is error now.
08:16:17 <chenke> Code Review - Error
08:16:46 <licanwei> I don't remember we need watcher_message group and notifier_driver option
08:17:17 <licanwei> and I can't find these in watcher source code
08:18:04 <chenke> I will check it. If we don't use this. I think we can remove it.
08:18:21 <licanwei> [https://review.opendev.org/668100 Reduce the query time of the instances when call get_instance_list() needs new patchset
08:18:22 <Dantalion> Yes if these options do not exist we can remove it
08:18:52 <licanwei> need new patch
08:19:07 <chenke> Ok. I am writing unittest for this patch.
08:19:20 <chenke> unittest about limit == -1
08:19:40 <licanwei> https://review.opendev.org/668363 Blacklist sphinx 2.1.0 (autodoc bug) needs review
08:19:51 <licanwei> +1 for review
08:20:21 <licanwei> https://review.opendev.org/668285 Add name field for test data needs review
08:20:59 <licanwei> https://review.opendev.org/668110 Releasenote for grafana datasource needs review
08:21:18 <licanwei> It's OK for me, will +2 after meeting
08:22:16 <Dantalion> Okaj, Grafana patch itself also needs review
08:22:45 <licanwei> OK~
08:23:09 <chenke> Ok.
08:23:13 <licanwei> #topic discussions
08:23:41 <licanwei> Inactive core-reviewer update
08:24:07 <Dantalion> I wonder if core-reviewers replied to the email on the mailing list
08:24:41 <licanwei> I've removed David TARDIVEL and Prudhvi Rao Shedimbi
08:25:10 <licanwei> no one reply
08:25:40 <Dantalion> Okaj, I did see YumengBao en Hidekazu review a patch recently
08:26:06 <licanwei> people inactive over one year will be removed
08:26:33 <Dantalion> I think that is a good strategy
08:26:53 <chenke> We can send it again if they still no one responds.
08:27:01 <chenke> send the email.
08:27:33 <licanwei> Do you plan to attend openinfra summit in Shanghai?
08:28:02 <Dantalion> I do not plan to attend the Shanghai summit
08:28:33 <chenke> When are you going to study for a master's degree
08:29:07 <Dantalion> Starting september will begin so called pre-master the master itself should start in Februari 2020
08:30:11 <licanwei> pre-master means ?
08:30:15 <chenke> It’s coming to September soon.
08:31:10 <Dantalion> pre-master is done when applying for a master degree if the bachelors degree was done on a applied university
08:31:38 <licanwei> ok, thanks
08:31:41 <licanwei> next
08:31:52 <licanwei> about id, uuid and hostname in ComputeNode
08:32:55 <licanwei> we have 3 related fields in ComputeNode
08:33:12 <licanwei> https://github.com/openstack/watcher/blob/master/watcher/decision_engine/model/collector/nova.py#L306
08:33:53 <licanwei> Now id is uuid, refer to https://developer.openstack.org/api-ref/compute/?expanded=list-hypervisors-details-detail#list-hypervisors-details
08:34:15 <chenke> Now uuid is equal to hostname
08:34:25 <licanwei> and as Matt said, service.host is equal to hypervisor name
08:35:00 <Dantalion> I think we need to remove id and only keep uuid but set uuid to node.id since with 2.53 this will be uuid
08:35:02 <chenke> I think we can remove id.
08:35:07 <licanwei> so I think maybe we just need two fields: uuid and hostname
08:35:21 <licanwei> yes, I agree with you
08:35:47 <Dantalion> I would like to make one addition
08:36:00 <licanwei> ?
08:36:08 <Dantalion> I think we should add hypervisor_type to ComputeNode
08:36:29 <Dantalion> since now bare-metal nodes will end up in list of ComputeNodes which might have unexpected results for strategies
08:37:23 <Dantalion> I think we should remove the baremetal nodes from the ComputeNodes. We could either do this during the building of the data model or let strategies decide to ignore them or not
08:37:57 <chenke> What kinds of values does hypervisor_type have
08:38:13 <licanwei> we don't include baremetal nodes in Compute datamodel
08:38:42 <Dantalion> licanwei: when you do hypervisors.list() you also get ironic nodes
08:39:15 <licanwei> They are different hypervisor_type?
08:39:30 <Dantalion> Yes I will make paste and show you
08:39:40 <chenke> Ok
08:39:56 <Dantalion> https://pastebin.com/sxv5dnLZ
08:40:08 <licanwei> If so ,we should remove baremetal nodes
08:40:09 <Dantalion> Here is output of 'openstack hypervisor list' it contains ironic nodes
08:40:26 <licanwei> thanks
08:40:28 <Dantalion> Similarly nova.hypervisors.list() will contain these ironic nodes
08:40:43 <Dantalion> These need to be removed in Compute model I think
08:41:48 <chenke> If this, we should remove it.
08:41:58 <licanwei> ok, we should filter baremetal nodes when building compute data model
08:42:04 <Dantalion> Agree :)
08:43:14 <chenke> +1
08:45:32 <licanwei> I'll report a bug to record this
08:46:30 <chenke> nice
08:47:11 <licanwei> ok, If no more thing to discuss, I'll end the meeting
08:47:43 <Dantalion> I have nothing else, thank you all
08:47:47 <chenke> Very efficient today.
08:48:08 <chenke> thanks everyone.
08:48:38 <licanwei> thank you Corne and chenke
08:48:55 <chenke> Bye~
08:49:00 <licanwei> #endmeeting