Friday, 2016-07-22

*** joehuang has joined #openstack-tricircle00:39
joehuanghello ricky11, first time to meet you here. May I know your name?00:47
*** Yipei_ has joined #openstack-tricircle00:48
*** Yipei_ has quit IRC00:53
*** Yipei_ has joined #openstack-tricircle00:53
*** ronghui has joined #openstack-tricircle00:59
*** hjw has joined #openstack-tricircle01:00
ronghuiwait for a moment, some guys in our team just join the channel01:01
ronghuiI recruited some new friends to join our tricircle01:03
joehuanggood job :)01:04
ronghuiis on the bus01:04
ronghuiand will late few minutes01:05
joehuangok, can you open the spreadsheet in google doc01:05
ronghui  ?????01:05
joehuang]hello yipei01:07
*** yinxiulin has joined #openstack-tricircle01:08
joehuangplease open the spreadsheet:
Yipei_i have already opened it01:09
joehuangdo you have any question on the features01:10
Yipei_for the admin api part, i have no question, the info in the sheet is clear enough01:11
joehuangone more api is added, for the routing table01:12
joehuangthe api to manage the routing table01:12
Yipei_ok, got it01:13
joehuangplease fill the "due date", when the patch could be merged into the trunk01:13
*** huangdongfeng has joined #openstack-tricircle01:14
shinobu_joehuang: it's not easy to decide due date *now*01:15
shinobu_joehuang: let's give more time (but not too much).01:15
joehuangunderstand, but we need an estimated date01:15
shinobu_joehuang: wait for 2 days01:15
shinobu_make sense?01:16
shinobu_joehuang: but we can decide *final* date01:16
joehuangother guys' idea?01:16
shinobu_maybe end of November?01:16
shinobu_let's discuss01:17
joehuango, no, we have to deliver a release alighed with Newton of OpenStack01:17
joehuangit's released on Oct.1501:17
shinobu_any thought more?01:17
ronghuiwe can decide the final date +101:18
joehuangwhat's "final date"01:18
shinobu_this is when we *must* complete regarding to only *api* in defined in spread sheet.01:19
shinobu_make sense?01:19
*** KK has joined #openstack-tricircle01:19
joehuangthis is the same meaning of "due date"01:19
zhiyuani think joe's idea is that we finish the API listed in the table in Newton circle01:20
shinobu_btw when is reasonable for each other?01:20
shinobu_>  it's released on Oct.15?01:21
yinxiulini have finished my plan, please check01:21
joehuangcool :) xiulin01:22
*** lzy has joined #openstack-tricircle01:22
lzyHello everybody morning, my name is Liu Zeyu, I come from Hunan University, and very happy to see you.01:23
joehuangwelcome lzy01:23
joehuangseems like a big family now in tricircle01:23
*** KK is now known as Liu01:23
*** Liu is now known as Liukun01:23
ronghuithey all interested in tricircle01:24
zhiyuanxiongqiu has implemented some cinder APIs, so I think he has sense how long the work will take01:24
ronghuihe just on the bus and will come in few minutes01:26
joehuangfor pod and podbinding, APIs are already there, but no tempest test yet01:26
*** longxiongqiu has joined #openstack-tricircle01:26
ronghuifor cinder  we think we can continue do it and also can try some new features of other APIs01:27
joehuanggreat idea01:28
joehuangfor backup and snapshot, it's mainly forwarding the api according to where the volume is(volume id routing info)01:29
joehuangand volume action also could be forwarded based on volume id routing info01:29
zhiyuanit may take some time to pass tempest test01:30
joehuangagree, implementation is relatively easy, need to spend time to debug the tempest test01:31
joehuangbut for action, it's much more easier01:31
shinobu_joehuang: how can we verify that it must work w/o test?01:33
joehuangthe tempest integration job is configured in the check and gate test, just open corresponding test cases01:34
joehuangWe need to write Admin API tempest test cases by our self01:34
shinobu_writing test cases would take more time?01:35
zhiyuanno, tempest test requires some configuration, we need to properly configure tempest to test our nova and cinder api_gw01:36
zhiyuantempest test of nova and cinder api is already there01:36
joehuangfor Nova API-gateway and Cinder API-GW/Neutron, just reuse Nova/Cinder/Neutron temepst test cases, they are already there01:37
shinobu_what will take a time?01:37
shinobu_i just make sure of something more specifically01:37
shinobu_so that we can schedule properly01:37
joehuang Admin API tempest test cases01:37
joehuangpod/pod binding/routing api01:38
shinobu_what is priority?01:38
joehuangfor all, or for amdin api01:39
joehuangI propose first fill the owner for the items you are interested in01:40
shinobu_sounds good to me01:40
joehuanggood, tricircle neutron plugin and tricircle admin api sheet is being filled very fast01:47
zhiyuanshinobu_: shall we discuss my patch a bit since other members are filling the table01:53
shinobu_zhiyuan: of course01:53
shinobu_what's your thought?01:54
zhiyuanshinobu_: i feel a bit strange to wrap a python builtin function01:55
joehuangthe due date column in Nova API-GW is disapper?01:55
shinobu_joehuang: yinxiulin did something01:56
shinobu_zhiyuan: please free to let me know what you would like to do01:56
joehuangok, move back01:57
joehuanghello, ronghui/xiongqiu, how do you think about the features in Cinder API-GW01:58
*** LLL has joined #openstack-tricircle01:58
*** LLL is now known as leyuquan01:58
zhiyuanshinobu_: python has many builtin functions like "hasattr", "getattr", "isinstance" which the naming convention is different from what we use in the Tricircle. in my opinion, we can just keep them since readers know they are builtin functions01:59
ronghuiFor now there is no problem01:59
ronghuibut we will discuss after the meeting02:00
joehuangso you can fill the owner information first, and due date as much as possible02:00
ronghuiwe will use tempest test first02:00
shinobu_zhiyuan: that's just temporarly solution. so please ignore me. there will be complete solution. thank you for your attention.02:01
zhiyuanshinobu_: :)02:03
ronghui@joehuang  we fill the due date whether we should use tempest test ?02:05
joehuangtempest test case should be there, for cinder-api gw, most reponse is returned directly from the bottom, so it's much easier to pass the tempest test02:06
zhiyuanour work includes implementing the API and passing the tempest test02:07
ronghuii mean we fill the information in the table just now or waiting for tempest test complete?02:07
joehuangno, you need to fill the table now, we are waiting for you :)02:09
joehuang"due date" the date you planned to deliver the feature, i.e. the code is merged in the repository(pass the tempest test)02:10
ronghuisorry  i have no permission to change the table02:16
joehuangyour gmail address? Xiongqiu has been added to the editor list02:16
zhiyuanxiongqiu: you can also assign os-volume_attachments API in nova to yourself, you are already working on it02:25
ronghui@joehuang  is this table complete just now  or today?   some feature of Cinder  we need to distribution02:32
*** Liukun has quit IRC02:33
*** leyuquan has quit IRC02:37
*** hjw has quit IRC02:40
joehuangok, you can complete today02:41
*** Yipei_ has quit IRC02:47
*** LLL_ has joined #openstack-tricircle02:54
*** LLL_ has quit IRC03:01
*** Yipei_ has joined #openstack-tricircle03:24
*** leyuquan has joined #openstack-tricircle03:26
*** lzy has quit IRC03:26
*** liuzeyu has joined #openstack-tricircle03:27
*** liuzeyu has quit IRC03:28
*** leyuquan has quit IRC03:29
shinobu_joehuang: everything looks good for you?03:30
*** ronghui has quit IRC03:31
*** ronghui has joined #openstack-tricircle03:31
*** joehuang has quit IRC03:39
*** Yipei_ has quit IRC03:43
*** Yipei_ has joined #openstack-tricircle04:53
*** longxiongqiu has quit IRC05:07
*** longxiongqiu has joined #openstack-tricircle05:27
*** longxiongqiu has quit IRC05:32
*** longxiongqiu has joined #openstack-tricircle06:08
*** longxiongqiu has quit IRC06:08
*** longxiongqiu has joined #openstack-tricircle06:08
ricky11Hello Tricircle team06:13
ricky11my name is rickyhai from South Korea06:13
ricky11I am interested in what you are doing and just would like have some quick questions as mentioned above06:14
ricky11they are all related to quota and resource management in Tricircle06:14
ricky11Nice to meet you all here :)06:15
*** ronghui has quit IRC06:16
*** Yipei_ has joined #openstack-tricircle06:35
*** ronghui has joined #openstack-tricircle06:40
ricky11<ricky11> Is quota management fully implemented and supported in Tricircle currently ? [00:44] <ricky11> where is the code of this part ? Thank you all.06:40
shinobu_ricky11: please look at tricircle/common/quota.py06:54
*** gongysh has joined #openstack-tricircle06:55
shinobu_ricky11: are you interested in this project?06:55
ricky11ok thanks I will look at that06:55
ricky11yes , interested in :)06:55
shinobu_o.k, thanks06:56
ricky11quota management is fully implemented as described in design doc ?06:57
shinobu_yes, but i would like you to have a look at quota_opt defiend in that file first06:59
*** Yipei_ has quit IRC07:00
*** Yipei_ has joined #openstack-tricircle07:00
shinobu_ricky11: if you are o.k to contribute this project, please let us know at any time07:05
ricky11Sure! thanks for asking me07:08
*** joehuang has joined #openstack-tricircle07:11
joehuanghello, there is some issue to open and combine new test cases in the integration temepest test job, it was fixed in this patch:
joehuangplease review and get it merged soon, so that other one can add new tempest test case to the job, for example, open the volume type tempest test cases07:14
joehuangonly need to add one line in the TESTCASES="$TESTCASES|tempest.api.volume.admin.test_volumes_type"07:14
*** gongysh has quit IRC07:16
*** huangdongfeng has quit IRC07:28
joehuanghello, ricky11, do you have some question on quota?07:37
ricky11Yes Joehuang07:38
joehuangMay I know your name?07:38
ricky11just wondering about quota management and resource management in Tricircle07:38
ricky11my name  is Ricky Hai07:38
joehuangok, for quota management, these is one section in the design doc to talk about the quota management07:39
ricky11how Tricircle can be aware of resource usage or available for a tenant ?07:39
ricky11yes , I looking at that doc now07:40
joehuangTricircle will control the quota usage roughly, and when the quota limit is to be reached07:40
joehuangwill refresh the usage from the bottom openstack instance07:41
joehuangcurrently quota management mechanism has already been ported to the Tricircle07:42
ricky11by polling all bottom openstack instances to get quota usage values then manually calculating by itself ?07:42
ricky11that meant all quota usage values is cached in top part , not in bottom openstack instances07:43
joehuangbut only _process_metadata_quota and _process_injected_file_quota were implemented, nova part, code was writen several month ago, but not tested and submitted07:43
joehuangyou are correct07:44
joehuangonly when the limit reached, then to verify the real usage from the bottom07:44
joehuangin bottom openstack, there is real quota usage only inside this bottom openstack07:45
joehuangone bottom openstack has no knowledge about other openstack07:45
ricky11but only nova that is currently providing real usage, not in cinder and neutron yet07:46
joehuangnot polling all, but polling openstacks where the tenant is spread into07:46
joehuangsome one in another project has raised bugs to Cinder and Neutron07:47
ricky11oh it depends on user's selection from, UX/UX when creating instance07:47
ricky11and then polling action is upon that corresponding bottom opentack instance07:47
joehuangdon't understand this "oh it depends on user's selection from, UX/UX when creating instance07:48
ricky11I meant when a user would like to boot a instance , user should chose AZ for residing07:49
ricky11then polling will be executed based on that AZ right ?07:50
joehuangfor cinder, usage is provided by resftul API, but not supported in cinder client07:52
*** yinxiulin has quit IRC07:52
shinobu_ricky11: i'm not clear about what "UX/UX" means. please clarify for my better understanding.07:52
joehuangAZ is optional, if it's not specified, then will select one bottom OpenStack for the tenant.07:52
joehuangpolling is not based on AZ, but the openstack instances the tenant has been bound07:53
ricky11ok Joe07:54
ricky11@shinobu  I meant UI= User interface like Horizon.that's all07:55
shinobu_o.k, that is not UX standing for User Experience in general?07:55
shinobu_not UX but UI?07:56
ricky11@Joe how does your project deal with these bugs currenty? like bug related to Cinder real time usage and nova usage (provided already but got out of syn somtime)07:57
joehuangI think absolutely accuracy in quota management is almost impossible, even in OpenStack07:59
shinobu_"like bug related to Cinder real time usage and nova usage (provided already but got out of syn somtime)" ?07:59
joehuangconsider that the cloud is mainly used by the way of "pay as you go", we can provide the accuracy in OpenStack level07:59
shinobu_"in OpenStack level" ?08:00
joehuangthis is mainly caused by concurrency in a large pool of machine08:00
ricky11yes, I agreed with you Joe08:01
joehuangthat means the accuracy is at the same level as what OpenStack currently in the quota management08:01
joehuangsometimes there is a little shift08:01
shinobu_o.k, clear08:01
shinobu_so "like bug related to Cinder real time usage and nova usage (provided already but got out of syn somtime)" is not proper description ?08:02
ricky11BTW, currently  in Tricircle for compute resources , it is calculated and syns by polling when limit has been reached08:03
ricky11how about storage in cinder ?08:03
ricky11is it calculated manually by counting ?08:03
ricky11yes I meant sync08:04
shinobu_counting what?08:04
joehuangsame processing mechanism, but code not submitted yet, not calculated manually, Nova-APIGW Cinder APIGW will do that for you08:04
ricky11using restful API for get updated with cinder ?08:06
*** hejiawei has joined #openstack-tricircle08:07
*** hejiawei has quit IRC08:07
joehuangfirst calculated in Cinder-APIGW if a new volume provision request is processed08:07
joehuangwhen the limits is reached, start a job using restful interface to polling the real usage from the bottom cinder08:08
joehuangthe usage in the Tricircle will be refreshed by real usage08:08
ricky11ok I got it08:09
joehuangif the quota is over, then no new volume provision will be allowed08:09
joehuangwould you like to contribute in the Tricircle08:09
ricky11initially how do you know about capacity of each bottom openstack ?08:10
joehuangyou mean the total capacity in each OpenStack pool, not single tenant?08:11
ricky11i meant about initiative resource values at the beginning as these values would be used to identify  if limit is reached or not08:11
joehuangthe capacity of one OpenStack instance will be polled in daily freequency08:11
joehuangok, you are still talking about quota08:12
joehuangif a new tenant account is opened08:12
joehuangthe quota will be applied to this tenant, (use default, or by quota class, and can be different for different tenant)08:13
joehuangthe initial usage is 008:14
joehuangif an already running pod is introduced into the Tricircle08:14
joehuangthen the quota usage should be refreshed to the Tricircle08:15
ricky11Great ! clear now08:17
shinobu_great. please work through code base if you get a chance08:18
ricky11one more question, tenant could be resided in different bottom openstack instances, so global resource view for that tenant will be summed at the end right ?08:18
shinobu_ricky11: btw are you happy to contirbute this project?08:21
joehuangthe global resource view could be retrieved from the Tricircle08:22
joehuangit'll be summed at the Tricircle08:22
ricky11thank you guys for your time , Will let you know whenever it is applicable to contribute to Tricircle08:23
shinobu_ricky11: great. have a good day08:25
joehuanghave a nice weekend08:25
ricky11thank Joe and Shinobu, You too08:26
joehuangto Shinobu and Zhiyuan, your comment has been answered, and the patch is updated too08:35
shinobu_joehuang: please add some *reasonable* comment for post_test_hook.sh08:36
joehuangcomment added for the flavor08:37
shinobu_joehuang: just to clarify that those numbers are dummy.08:38
joehuangyou mean the whole file?08:38
shinobu_joehuang: please refere to
shinobu_joehuang: do you get me?08:39
shinobu_s/me/my point/08:40
joehuangyou are looking at the old patch, the new one is
joehuangthe patchset number is 17 but not 16 now08:42
shinobu_here we go:
*** ronghui has quit IRC08:48
joehuangupdated again: line40 ~ line 4608:53
shinobu_joehuang: what i expected is that "1 1024 10 1" are dummy08:55
shinobu_aren't they?08:55
shinobu_joehuang: Meaning that those numbers are not used08:56
joehuangthese numbers are used in the flavor creation08:56
joehuangnot dummy08:56
shinobu_joehuang: so this is not correct:
shinobu_joehuang: based on Zhiyuan's comment08:57
joehuangthese number are parameters for nova flavor-create command08:58
shinobu_joehuang: i know08:58
shinobu_joehuang: please read carefully Zhiyuan's comment08:59
shinobu_joehuang: " Ram 1024, disk 10, vcpus 1, these three numbers are not used."08:59
joehuangyes, they are not used in the VM creation, but they have to be there to create a flavor09:00
joehuangotherwise, flavor-create command will fail09:00
shinobu_joehuang: "Currently flavor mapping between top and bottom pod is not implemented so this command just creates a record in database but what is actually used is the bottom flavor with the same ID (which is 1). Ram 1024, disk 10, vcpus 1, these three numbers are not used."09:02
shinobu_joehuang: this is not written properly?09:03
joehuangdo you mean I need to add comment as Zhiyuan described?09:03
shinobu_joehuang: no, i want to make sure if that comment is right or not . 1024, 10, 1 those numbers are not used.09:05
joehuangthe comment is right09:06
zhiyuanthose numbers are used to create top flavor, but bottom flavor is not created based on those numbers currently09:14
zhiyuanwe have one top flavor and one bottom flavor with the same ID, but the parameters of these two flavor may be not the same09:15
zhiyuanwhen one user boot a vm via nova_apigw, specifying flavor ID as 1, nova_apigw will pass this request to bottom nova api, so bottom flavor with ID 1 is used09:17
zhiyuanis it clear now?09:19
joehuangthe mapping and sync will be done later09:21
*** joehuang has quit IRC09:53
*** longxiongqiu has quit IRC12:24
*** ricky11 has quit IRC12:28
*** longxiongqiu has joined #openstack-tricircle13:13
*** longxion_ has joined #openstack-tricircle13:14
*** longxiongqiu has quit IRC13:17
*** longxion_ has quit IRC13:20
*** longxiongqiu has joined #openstack-tricircle13:21
*** Yipei_ has quit IRC13:30
*** zhiyuan has quit IRC15:39
*** zhiyuan has joined #openstack-tricircle15:47
*** longxiongqiu has quit IRC17:27
*** longxiongqiu has joined #openstack-tricircle17:40
*** longxiongqiu has quit IRC17:46
*** longxiongqiu has joined #openstack-tricircle20:57
*** longxiongqiu has quit IRC21:01
*** longxiongqiu has joined #openstack-tricircle22:06
*** longxiongqiu has quit IRC22:10
*** longxiongqiu has joined #openstack-tricircle23:04
*** longxiongqiu has quit IRC23:08
*** longxiongqiu has joined #openstack-tricircle23:51
*** longxiongqiu has quit IRC23:56

Generated by 2.14.0 by Marius Gedminas - find it at!