Thursday, 2017-06-22

*** wanghao has joined #openstack-mogan00:24
zhenguoliusheng: hi, our physical env is dead?01:09
liushengzhenguo: hmm, I don't know what happend01:09
zhenguoliusheng: hah, seems it restarted01:10
zhenguoliusheng: I just found all services are down01:10
liushengzhenguo: oh, may all the servers have been shutted off last weekend01:10
zhenguoliusheng: seems yes01:10
litao__I found the my devstack is always bad these days01:23
zhenguolitao__: just update it and restack01:23
*** openstackgerrit has joined #openstack-mogan01:25
openstackgerritliusheng proposed openstack/mogan master: Tempest: add tempest test for server floatingip association/disassociation
litao__zhenguo: ok01:25
*** liujiong has joined #openstack-mogan01:31
litao__zhenguo: Do you know this issue when mogan create network01:34
litao__The reason is Remote error: DiscoveryFailure Could not determine a suitable URL for the plugin01:34
zhenguolitao__: no, I didn't see that before01:34
litao__zhenguo: ok01:34
shaohe_fengzhenguo: morning01:37
zhenguoshaohe_feng: o/01:37
* zhenguo brb01:37
zhenguohi all, the weekly meeting will happen soon, please move to #openstack-meeting01:59
*** liujiong_lj has joined #openstack-mogan02:10
*** liujiong has quit IRC02:12
*** liujiong_lj is now known as liujiong02:27
wanghaozhenguo: there is a question left,  that how and who will decide our bug priority in bug review hour?03:03
zhenguowanghao: every one could be03:03
zhenguowanghao: the list there is not by priority03:03
zhenguoeveryone can +1 after a patch03:04
zhenguowhich one got more +1 it need to be reviewed more03:04
wanghaozhenguo: okay  great03:04
zhenguohah, we wil leave a few minutes for everyone to +1 at the beginning of the time03:05
* zhenguo brb03:06
wanghaoliusheng: I didn't find that openstack didn't allow to use six.  So I think it's fine?03:13
wanghaoliusheng: maybe just some functions.03:13
liushengwanghao: yes, it is ok03:14
wanghaoliusheng: I found some patch that replace unicode with six.text_type for py3. haha03:14
wanghaoliusheng: sure03:14
liushengzhenguo: yes, and please confirm my first comment03:15
liushengwanghao: sorry ,wrong person, ^^03:16
wanghaoliusheng: haha03:17
openstackgerritwanghao proposed openstack/mogan master: Refactor exception raise by using mogan exception
wanghaoguys, I'm sorry that I have a internal meeting 16:00-17:30, so I will begin my review hour early~03:24
openstackgerritZhenguo Niu proposed openstack/mogan master: Adds `resources` field for flavors.
zhenguowanghao: hah, maybe we can change the review hour everyday to fit for most of us' time03:29
zhenguohi guys, for the new flavor spec, do you think we need to add a resource_traits filed as well to pass traits information03:31
zhenguomaybe later when we really need it03:33
zhenguoliusheng: hi, for the POC, how about sending the patches out, then we can do it together03:36
liushengzhenguo: ok, sure03:36
zhenguoliusheng: thanks03:36
openstackgerritliusheng proposed openstack/mogan master: WIP: import placement service
openstackgerritliusheng proposed openstack/mogan master: WIP: import placement service
openstackLaunchpad bug 1699662 in Mogan "mogan.tests.unit.db.test_compute_nodes.DbComputeNodeTestCase.test_compute_node_get_all accasionally faield" [Medium,New]04:44
openstackgerritZhenguo Niu proposed openstack/mogan-specs master: New flavor for baremetal servers
zhenguoliusheng: gate is broken?05:10
liushengzhenguo: seems yes06:03
zhenguoliusheng: but seems not related to us06:06
liushengzhenguo: yes06:08
*** liusheng has quit IRC07:36
*** liusheng has joined #openstack-mogan07:37
zhenguoliusheng, shaohe_feng, wanghao, litao__: gates are broken, cancel today's review hour?07:44
liushengzhenguo: we cannot merge any patch :(07:45
zhenguobut specs are fine07:45
liushengzhenguo: hah, yes07:45
zhenguohow about specs review hour07:45
litao__zhenguo: fine07:46
zhenguoliusheng: yes, move az to aggregate seems a good idea07:52
zhenguoliusheng: we will not get az information from ironic node's properties/placement resource providors07:53
openstackgerritZhenguo Niu proposed openstack/mogan-specs master: New flavor for baremetal servers
*** wanghao_ has joined #openstack-mogan08:00
zhenguoliusheng: but placement aggregate doesn't have a notion like metadata08:00
zhenguoliusheng: it only contains a uuid08:00
*** wanghao has quit IRC08:03
liushengzhenguo: I think the placement started not a long time, may they will consider this if they want to meat full Nova's requirement08:08
zhenguoliusheng: so if we move to placement how to handle the aggregate requirement?08:08
liushengzhenguo: you mean aggragate or az ?08:12
zhenguoliusheng: aggregate08:12
zhenguoliusheng: the placement_aggregate is very simple08:12
zhenguoliusheng: but we need a aggregate metadata which include key value pairs to match with flavors'08:12
zhenguoliusheng: I will be away for a few mins, ttyl?08:13
zhenguoliusheng: you can start to review if I don't come back08:13
* zhenguo brb08:13
liushengzhenguo: ok, sure08:13
zhenguoliusheng: I'm back08:39
liushengzhenguo: welcome :), I have left 3 comments of your spec08:40
zhenguoliusheng: hah, thanks08:41
zhenguoliusheng: wrt flavor description, maybe we can just get rid of the max lenght check on JSON chema?08:51
liushengzhenguo: yes, I think so08:53
shaohe_fengwhat's wrong with gate?08:53
zhenguoshaohe_feng: not sure, but seems not related to us08:54
liushengzhenguo: wdyt about another comment of description ?08:54
zhenguoliusheng: you mean allow users to update it?08:54
liushengzhenguo: not sure, but is it possible operator change the hardware of a specific resource class ?08:55
liushengzhenguo: if so ,the description including hardware profile may cannot exactly reflect the hardware info08:55
zhenguoliusheng: you mean after the node deployed?08:56
liushengzhenguo: no, I mean we have a set of nodes with specific hardware standard may have resource class of "gold"08:57
zhenguoliusheng: in a production env, I don't think it make sense to update the hardware spec after it sold to users08:57
liushengzhenguo: and we may change the "gold" meaning08:57
zhenguoliusheng: if you want to change the gold meaning, better to add a new flavor08:58
zhenguoliusheng: as if you change the meaning, users who use the flavor deployed servers before may confuse08:58
liushengzhenguo: yes, that is a way08:58
zhenguoliusheng: as they use the old meaning to require a server, hah08:58
liushengzhenguo: I am just consider the possibilities users will ask why the  hardware info is not equal what I saw in the flavor's profile. lol08:59
zhenguoliusheng: that's need a new flavor09:00
liushengzhenguo: hah09:00
liushengzhenguo: ok, lets operators to ensure that09:01
zhenguoliusheng: yes, hah09:01
zhenguoliusheng: do you think we still need minLength for description field09:05
liushengzhenguo: may don't need09:06
zhenguoliusheng: so when creating flavor they can specify a '' for description?09:07
liushengzhenguo: your required properties only include name and resources. hah09:08
zhenguoliusheng: hah09:09
zhenguoliusheng: I prefer to add description there as well, wdyt?09:09
liushengzhenguo: agree, let force the operater to add flavor with hardware profile. hah09:12
zhenguoliusheng: ok, will update the spec09:12
openstackgerritZhenguo Niu proposed openstack/mogan-specs master: New flavor for baremetal servers
zhenguoliusheng: ok, let's talk about the placement?09:14
liushengzhenguo: ok09:15
zhenguoliusheng: still not very clear about the aggregates09:16
zhenguoin nova host aggregates, we can set any metadata there witch can match with flavor extra spec09:17
liushengzhenguo: may need more knowledge about placement's implementation :)09:17
zhenguoaz also in aggregates metadata09:18
liushengzhenguo: yes, seems az contains host aggregate09:19
zhenguoyes, so if we don't set any az to aggregate, we just got a default az09:21
zhenguowe don't need to set az for every node09:21
zhenguoyou need to create a aggregate with a group of node and set the aggregate metadata with az=xxx to indicate these nodes are in that az09:21
liushengzhenguo: yes, hope placement can provide the capability09:24
liushengzhenguo: maybe we can just wait, may placement will add that support, since Nova may also have the requirement ?>09:25
zhenguoliusheng: nova has a host notion, so they don't need placement09:26
zhenguohost and node a different things in nova09:26
liushengzhenguo: yes09:27
liushengzhenguo: oh, maybe09:27
liushengzhenguo: how about we try to add like traits for aggregate of placemnt ?09:27
zhenguoliusheng: placement only allow set traits to resource providers09:29
zhenguoliusheng: so maybe we have a aggregate of nodes, then set taits like metadata to all resource provider in it. hah09:30
zhenguoliusheng: seems it can work09:30
liushengzhenguo: I mean we can try to implement that in the future09:30
liushengzhenguo: hah09:31
zhenguoliusheng: yes, maybe09:31
zhenguoliusheng: before Pike release we should get rid of things that we are not sure and APIs that node defined09:31
liushengzhenguo: yes, sure09:32
zhenguoliusheng: we can enrich features in next release09:32
zhenguoliusheng: we just need a set of APIs/features that are stable and we are sure it's well defined with Pike release,09:33
*** wanghao_ has quit IRC09:34
liushengzhenguo: hah, that is ture09:34
zhenguoliusheng: so, please feel free to get rid of things you don't like09:34
liushengzhenguo: lol, yes09:34
liushengzhenguo: will look the new spec09:35
zhenguoliusheng: thanks09:35
zhenguoshaohe_feng: please also help to review the specs if you got time09:35
zhenguoshaohe_feng: new flavors, palcement, and adopting servers09:36
*** wanghao has joined #openstack-mogan09:36
*** wanghao has quit IRC09:37
*** liujiong has quit IRC09:59
* zhenguo away11:13
*** litao__ has quit IRC11:54
zhenguoliusheng: how about getting rid of the flavor extra specs and add a resource_traits field instead12:04
liushengzhenguo: what's the relationship between it with traits in placement ?12:05
zhenguoliusheng: reference to that traits12:05
liushengzhenguo: so how to use it? pass it as the query filters to placement api ?12:06
zhenguoliusheng: yes12:06
zhenguoliusheng: extra_specs also aims to do that12:07
zhenguoliusheng: resources: {CUSTOM_BAREMETAL_GOLD: 1}, resource_traits: {CUSTOM_BAREMETAL_GOLD: rack1}12:08
zhenguoliusheng: one is for quantity and the other is for qulitity12:08
liushengzhenguo: oh, I thought you want to user extra_spec to filter the specific requirement of nodes, e.g. node with particular net card12:10
zhenguoliusheng: just wan to change to notion from extra_specs to resource_traits12:11
liushengzhenguo: and also may we will put the az requirement in extra_spec/ resource_traits ?12:11
liushengzhenguo: yes, it is more explict12:11
liushengzhenguo: i like this :D12:12
zhenguoliusheng; like resource_traits: {CUSTOM_BAREMETAL_GOLD: az1}12:12
zhenguoliusheng: everything is ok12:12
liushengzhenguo: hah12:12
zhenguoliusheng: as operators using placement notion to manage resources with traits and resource classes12:13
zhenguoliusheng: they also define flavors, so use traits is more consistent12:13
liushengzhenguo: yes, make sense to me12:13
zhenguoliusheng: hah, so I will update the spec12:13
liushengzhenguo: thanks for your efforts!12:14
liushengzhenguo: our gate still broken ?12:14
zhenguoliusheng: not sure, but seems ironic gate is still broken, you can recheck one to see12:15
liushengzhenguo: let me recheck12:15
zhenguoliusheng: okay12:15
openstackgerritliusheng proposed openstack/mogan master: WIP: import placement service
openstackgerritliusheng proposed openstack/mogan master: WIP: import placement service
openstackgerritliusheng proposed openstack/mogan master: WIP: import placement service
openstackgerritZhenguo Niu proposed openstack/mogan-specs master: New flavor for baremetal servers
*** liujiong has joined #openstack-mogan13:31
*** liujiong has quit IRC14:18

Generated by 2.15.3 by Marius Gedminas - find it at!