Tuesday, 2017-07-18

*** wanghao has joined #openstack-mogan00:29
*** litao__ has joined #openstack-mogan00:52
zhenguomorning mogan!00:54
zhenguoshaohe_feng: please have a look at this https://review.openstack.org/#/c/478406/01:11
zhenguowanghao, litao__: would be appreciated if you can add more quotas support before Pike:D seems every resource should have a quota, like keypair, metadata, network, etc.01:16
wanghaozhenguo: yes,  I have create a branch yesterday01:17
zhenguowanghao: thanks01:17
wanghaozhenguo: I plan to support flavor first,  do you think it's okay.01:17
zhenguowanghao: sure01:17
wanghaozhenguo: thanks,  I think this week can commit the code patch.01:18
zhenguowanghao: cool01:18
zhenguowanghao: btw, does nova have a quota for flavors?01:19
wanghaozhenguo: well not sure yet,  let me check it.01:20
zhenguowanghao: ok, as flavor is for admins, so maybe not need a quota, need to check nova01:20
wanghaozhenguo: sure01:20
wanghaozhenguo: yeah, nova didn't check quota for flavor.01:21
zhenguowanghao: hah01:21
wanghaozhenguo: so we also didn't need it.01:21
zhenguowanghao: yes01:22
wanghaozhenguo: well,  then it should be keypair and something else.01:22
zhenguowanghao: yes, keypair, and maybe just something else like metadata and injected files01:22
zhenguowanghao: network related stuff seems have it's own quota from neutron01:23
wanghaozhenguo: yes, neutron have quota for network01:23
wanghaozhenguo: okay, I got it.01:23
zhenguowanghao: hah, so it's easy for us, not much to do01:23
wanghaozhenguo: :)  haha01:24
*** harlowja has quit IRC01:34
litao__zhenguo: If i have spare time , I will assist wanghao01:44
zhenguolitao__: thanks :D01:45
*** litao___ has joined #openstack-mogan02:06
*** mattcburns_ has joined #openstack-mogan02:06
*** wanghao_ has joined #openstack-mogan02:10
*** litao__ has quit IRC02:13
*** wanghao has quit IRC02:13
*** mattcburns has quit IRC02:13
*** litao___ is now known as litao__02:13
*** mattcburns_ is now known as mattcburns02:13
liushengzhenguo: hi, for this:  https://review.openstack.org/#/c/483203, I just given a -1 for the strange mac address, please correct me if there is a reason :)03:01
zhenguoliusheng: hah, also not sure why I added that :D03:03
liushengzhenguo: hah03:03
openstackgerritZhenguo Niu proposed openstack/mogan master: Add indexes to tables  https://review.openstack.org/48320303:13
*** harlowja has joined #openstack-mogan03:40
*** harlowja has quit IRC04:54
openstackgerritZhenguo Niu proposed openstack/mogan master: Adds aggregates DB model and API  https://review.openstack.org/48278605:22
openstackgerritMerged openstack/mogan master: Clean the compute_node and compute_port objects and db interfaces  https://review.openstack.org/47840605:42
openstackgerritZhenguo Niu proposed openstack/mogan master: [WIP]Add aggregate object  https://review.openstack.org/48463005:50
openstackgerritZhenguo Niu proposed openstack/mogan master: Add indexes to tables  https://review.openstack.org/48320305:51
openstackgerritMerged openstack/mogan master: Updated from global requirements  https://review.openstack.org/48455905:51
zhenguoliusheng: with the last placement patch landed, we have totally migrated to placement :D06:10
liushengzhenguo: hah06:11
zhenguoliusheng: maybe you can +A this again https://review.openstack.org/#/c/483203/ :D06:48
liushengzhenguo: done06:49
zhenguoliusheng: thanks06:49
liushengzhenguo: np :)06:49
liushengzhenguo: please take a look https://review.openstack.org/#/c/482057/ if you have time :)06:50
zhenguoliusheng: sure06:50
*** liusheng has quit IRC06:59
*** liusheng has joined #openstack-mogan06:59
zhenguoliusheng: just left a couple of comments on the uwsgi patch although I'm not quite familiar with that :D07:06
liushengzhenguo: thanks, let me check07:07
openstackgerritZhenguo Niu proposed openstack/mogan master: [WIP]Add aggregate object  https://review.openstack.org/48463007:13
openstackgerritMerged openstack/mogan master: Add indexes to tables  https://review.openstack.org/48320307:28
liushengzhenguo: hi, I have replied you about your last comment inline07:31
zhenguoliusheng: ok, I'm fine with it :D07:46
liushengzhenguo: thanks07:46
zhenguoliusheng: hah07:47
openstackgerritliusheng proposed openstack/mogan master: Support running api under uWSGI  https://review.openstack.org/48205707:47
litao__liusheng: Does the mogan client support creating flavor with resource classes?07:55
liushenglitao__: yes07:55
liushengliusheng: --resources07:56
liushenglitao__: --resources07:56
litao__liusheng: Ok, maybe I need to pull the latest codes.07:56
liushenglitao__: hah07:57
zhenguoliusheng: does placement support an API to create aggregates08:29
liushengzhenguo: yes08:29
zhenguoliusheng: just POST /aggregates?08:29
liushengzhenguo: sorry, let me confirm08:30
zhenguoliusheng: I can't find 'aggregates' in ROUTE08:32
liushengzhenguo: sorry, seems there is only get aggregates by rp and put aggregates for a rp interfaces08:36
zhenguoliusheng: so, how do we know the aggregate id to put :(08:37
liushengzhenguo: GET/PUT /resource_providers/{uuid}/aggregates08:37
zhenguoliusheng: or anything is ok?08:37
liushengzhenguo: let me test.08:37
zhenguoliusheng: ok ,thanks08:37
openstackgerritZhenguo Niu proposed openstack/mogan master: Support running api under uWSGI  https://review.openstack.org/48205708:38
zhenguoliushen: just asked alex, the uuid can be created automatically08:45
zhenguoliusheng: if the specified uuid doens't exist in placement08:45
liushengzhenguo: it is strange in my testing, the id is a integer than uuid08:46
zhenguoliusheng: ?08:47
zhenguoliusheng: you specify a uuid then get an integer08:47
liushengzhenguo: curl -g -i -X PUT -H "Accept: application/json" -H "Content-Type: application/json" -H "X-Auth-Token: $(openstack token issue -f value -c  id)" -H "OpenStack-API-Version: placement latest" -d '["42896e0d-205d-4fe3-bd1e-100924931787","5e08ea53-c4c6-448e-9334-ac4953de3cfa"]'08:47
zhenguoliusheng: the placement_aggregate table will save them?08:48
liushengzhenguo: no the in the resource_provider_aggregates table, there are two items, each one have a column named resoruce_provider_id, and another is "id"08:49
liushengzhenguo: the "id" column is a integer08:49
zhenguoliusheng: that's ok08:50
zhenguoliusheng: nothing in placement_aggregates table?08:50
zhenguoliusheng: the resource_provider_aggregates just save the id field instead of uuid08:51
liushengzhenguo: oh, yes, it is08:51
liushengzhenguo: thanks for reminding me, hah08:51
zhenguoliusheng: so, it's easy for us, we can add our node aggregates, not need to sync with placement aggregate uuid :D08:52
liushengzhenguo: but the uuid isn't generated automatically08:52
liushengzhenguo: it is the inputted two items08:52
zhenguoliusheng: we create aggregate object on mogan side then got the uuid08:53
zhenguoliusheng: when add nodes to the aggregates, we call PUT rps/uuid/aggregates with the aggregate uuid08:53
liushengzhenguo: yes, maybe a aggregate uuid is enough. the aggregate object including nothing else.hah08:54
zhenguoliusheng: name and metadata as well, hah08:54
liushengzhenguo: oh, yes, for us08:54
zhenguoliusheng: yes, seems placement aggregate is just designed for our use case08:55
liushengzhenguo: hah08:55
liusheng zhenguo for the PUT interface, the inputted items must be uuid format08:55
zhenguoliusheng: hah, so things become clear, I will update the spec, and add patches soon08:55
zhenguoliusheng: that's ok08:56
zhenguoliusheng: another thing is the  rp cache in report client, if it's singleton, I think we can leverage that08:56
liushengzhenguo: yes, actually in Nova, there is anothor related cache mechanism, for the resource class, usaully, it exposed as name, but for internal usages, it used by id. there is a exchanging cache08:58
zhenguoliusheng: yes, I also noticed tthat08:58
zhenguoliusheng: if you like, you can work on the aggregate cache part, hah08:59
liushengzhenguo: ok, will try. hah09:01
zhenguoliusheng: hah, thanks09:01
zhenguoliusheng: so, before adding nodes to aggregates, the mogan aggregate will not interact with placement09:02
liushengzhenguo: got it09:02
zhenguoliusheng: hah, now we really learned how to use placement aggregates, so it's totally different with nova host aggregates09:03
liushengzhenguo: yes09:03
liushengzhenguo: actually nova also has aggregates and aggregate_metadata09:04
zhenguoliusheng: yes,09:04
openstackgerritZhenguo Niu proposed openstack/mogan master: Adds aggregates DB model and API  https://review.openstack.org/48278609:25
openstackgerritZhenguo Niu proposed openstack/mogan master: [WIP]Add aggregate object  https://review.openstack.org/48463009:25
openstackgerritZhenguo Niu proposed openstack/mogan master: [WIP]Add aggregate API  https://review.openstack.org/48469009:25
openstackgerritZhenguo Niu proposed openstack/mogan-specs master: Node aggregates  https://review.openstack.org/47092709:32
*** wanghao_ has quit IRC09:37
openstackgerritMerged openstack/mogan master: Support running api under uWSGI  https://review.openstack.org/48205709:55
*** zhuli_ has quit IRC11:42
*** zhuli has joined #openstack-mogan11:42
*** zhuli has quit IRC11:43
*** zhuli has joined #openstack-mogan11:43
*** litao__ has quit IRC12:06
zhenguoliusheng: how to restart service with the current devstack12:16
liushengzhenguo: same way, service devstack@mogan-api restart12:16
zhenguoliusheng: ok, thanks12:16
liushengzhenguo: np12:18
*** harlowja has joined #openstack-mogan17:24

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!