Tuesday, 2017-06-13

*** wanghao has joined #openstack-mogan00:50
zhenguomorning mogan!00:56
*** litao__ has joined #openstack-mogan01:15
openstackgerritwanghao proposed openstack/mogan master: Specify neutron port when creating servers  https://review.openstack.org/47082401:20
wanghaozhenguo: ping01:20
zhenguowanghao: pong01:20
wanghaozhenguo: I updated the description of port_uuid as you commented, but I think it's better to keep the exlustion explain for port_id and network_id01:21
wanghaozhenguo: that make sense to you?01:21
zhenguowanghao: yes, it should be there, hah01:21
wanghaozhenguo: hah  cool01:21
zhenguowanghao: the patch looks most good, I will test it when my env is ready, thanks for the update!01:22
wanghaozhenguo: sure np.01:22
openstackgerritwanghao proposed openstack/mogan master: Specify neutron port when creating servers  https://review.openstack.org/47082401:24
zhenguoshaohe_feng: when you are online, please remember to review, we got a long review queue :D01:29
* zhenguo really need to get more core reviewers...01:30
openstackgerritZhenguo Niu proposed openstack/mogan-specs master: New flavor for baremetal servers  https://review.openstack.org/45411302:20
zhenguohi all, please help to review the node aggregates spec: https://review.openstack.org/#/c/470927/4/specs/pike/approved/node-aggregate.rst02:22
zhenguoliusheng: hi, should we add a multi-tenant ironic dvsm gate?03:29
liushengzhenguo: what does that for ?03:30
zhenguoliusheng: ironic env with networking-generic-switch03:31
zhenguoliusheng: but not sure if we need it or just ironic need test it03:31
liushengzhenguo: sorry, I don't know much about that :(, but for now, the tempest job can only run in admin context03:34
zhenguoliusheng: hah, that need to be improved03:34
liushengzhenguo: yes, may need to dig more03:35
zhenguoliusheng: yes, please do it03:35
liushengzhenguo: ok, will try :D03:35
zhenguoliusheng: thanks03:36
zhenguoliusheng: and I want to change to use ironic attach/detach vif api instead of listing all ports then choose one by ourselves03:37
zhenguoliusheng: but that means we will lose the ability of specifying port_type03:37
liushengzhenguo: so how about support both ?03:39
zhenguoliusheng: it's hard, as attach/detach vif api doesn't support to pass someting like port03:40
zhenguoliusheng: need to ask ironic support this03:41
zhenguoliusheng: we can discuss this during the meeting to see if there's someone need this feature now or we can wait ironic support it03:45
liushengzhenguo: yes, we can discuss more03:46
zhenguoliusheng: ok03:47
zhenguoliusheng: and please remember to do more reviews :)03:48
liushengzhenguo: ok sure, I am just backed from a internal meeting :D03:48
zhenguoliusheng: hah, ok03:49
*** wanghao has quit IRC06:15
*** wanghao has joined #openstack-mogan06:16
*** wanghao_ has joined #openstack-mogan06:19
*** wanghao has quit IRC06:23
wanghao_zhenguo: ping06:37
wanghao_zhenguo: some comments for node aggregates06:37
zhenguowanghao_: thanks06:38
wanghao_zhenguo: np06:38
* zhenguo brb06:47
litao__zhenguo: ping06:51
*** wanghao_ has quit IRC07:13
*** wanghao has joined #openstack-mogan07:14
zhenguolitao__: pong07:16
zhenguoliusheng: need your opinion on this https://review.openstack.org/#/c/472157/07:56
liushengzhenguo: ok07:56
liushengzhenguo: is this ok with your new flavor model ?07:59
zhenguoliusheng: yes, only change the extra specs part07:59
zhenguoliusheng: after that landed, I will move extra_specs table to flavor, seems no need to use a new table and object if we don't have a specific api endpoint for it08:00
openstackgerritZhenguo Niu proposed openstack/mogan-specs master: New flavor for baremetal servers  https://review.openstack.org/45411308:02
zhenguoliusheng: And with the new flavor model, maybe we can just change to use Placement, hah08:03
liushengzhenguo: ok, got it08:04
liushengzhenguo: we don's user json_schema to validate the patch data ?08:24
liushengzhenguo: s/we/shy08:24
zhenguoliusheng: yes,08:24
liushengzhenguo: why08:24
zhenguoliushegn: as we add a WSME.validator for patch data08:24
liushengzhenguo: others interfaces all use json_schema08:25
liushengzhenguo: any particularity of patch data ?08:26
zhenguoliusheng: patch data is a bit different08:26
zhenguoliusheng: we limited the op and which filed can be update08:26
zhenguoliusheng: and as I understand no difference with json schema check and WSME validator08:27
liushengzhenguo: got it, similar with server's patch api08:28
zhenguoliusheng: yes08:28
zhenguoliusheng: but I find a problem08:28
zhenguoliusheng: we don't have a decorator for json schema check08:28
liushengzhenguo: where need a decorator for json schema check ?08:29
zhenguoliusheng: I mean instead of explicitly call json_schema_check during every API, seems it's better to add a decorator for that, wdyt?08:30
liushengzhenguo: maybe not a good idea, we may need to check 1 input parameter of the api method by json schema check. but we cannot ensure which parameter need to be check for every api methods.08:33
zhenguoliusheng: we only add json schema for request body08:34
zhenguoliusheng: but seems not easy to discovery which one is body, alwasy the last one?08:34
liushengzhenguo: hah, that will add limitation of possible new api interface in the future08:35
zhenguoliusheng: hah, we can leave it as is now08:36
liushengzhenguo: yes, for your patch, I prefer to agree your opinion08:37
zhenguoliusheng: thanks, I will ping shaohe_feng to see it08:37
zhenguozhangyang: are you around?08:39
zhenguozhangyang: do you have a real requirement for specifying port type when claiming a server? I plan to drop it provisionally08:40
*** wanghao_ has joined #openstack-mogan09:04
*** wanghao has quit IRC09:07
* zhenguo brb09:08
*** wanghao_ has quit IRC09:31
openstackgerritliusheng proposed openstack/mogan master: Tempest: add tests for server states interfaces  https://review.openstack.org/47376009:39
openstackgerritZhenguo Niu proposed openstack/mogan master: Change to use server metadata instead of extra  https://review.openstack.org/47318409:58
*** liujiong has quit IRC10:09
liushengzhenguo: I have add a patch about tempest cases of server state APIs10:23
liushengzhenguo: but seems, the rebuild case cannot pass10:23
liushengzhenguo: but it can pass in my local env10:31
zhenguoliusheng: let me check, is there some clue in the logs?10:53
liushengJun 13 10:14:38.168934 ubuntu-xenial-osic-cloud1-s3500-9283043 ironic-conductor[10423]: ERROR ironic.drivers.modules.agent [-] node 240fc407-b298-44c8-a69d-2fdad8cd6d49 command status errored: {u'message': u'Command execution failed: Unable to stat device /dev/vda1 after attempting to verify 3 times.', u'code': 500, u'type': u'CommandExecutionError', u'details': u'Unable to stat device /dev/vda1 after attempting to verify 3 times.'}10:54
liushengzhenguo: above error in ironic-conductor's log10:54
zhenguoliusheng: maybe need to recehck to see10:55
zhenguoliusheng: not sure if it's related to us10:55
liushengzhenguo: ok10:55
zhenguoliusheng: seems all tempest gate failed10:55
liushengzhenguo: yes, just found that :(10:57
zhenguoliusheng: all failed with the same log10:57
liushengzhenguo: but that is not same as my patch10:57
liushengzhenguo: in my patch, it failed because rebuid server failed10:58
zhenguoliusheng: I checked anothe one, is the same as you provided10:58
zhenguoliusheng: in fact, rebuild and create use a same workflow in ironic10:58
liushengzhenguo: let me check, but in https://review.openstack.org/#/c/472157/, I found it failed in cleanup period10:59
zhenguoliusheng: let's recheck first11:00
liushengzhenguo: ok11:00
liushengzhenguo: if the states API patch can merged, we have more job to check the servers action11:02
liushengzhenguo: :D11:02
zhenguoliusheng: thanks for helping improve this :D11:02
liushengzhenguo: np11:03
zhenguoliusheng: btw, are you interested in looking into the placement service11:03
liushengzhenguo: ok, sure, I can try11:04
zhenguoliusheng: thanks11:04
zhenguoliusheng: if it's possible, the node aggregates spec should be changed to be a totally diferent one11:05
liushengzhenguo: seems your patch remove many flavor functional tests, lol11:05
zhenguoliusheng: hah, yes11:05
zhenguoliusheng: as it removed one api endpoint11:05
liushengzhenguo: yes, we didn't have many functional tests11:06
zhenguoliusheng: lol11:06
zhenguoliusheng: I know adding tests is a bit boring, but please focus on that, as we really need it, thanks for sticking on it :D11:07
liushengzhenguo: sure, no problem :D11:10
openstackgerritMerged openstack/mogan master: DOCS: add documentation about testing  https://review.openstack.org/47202811:19
liushengzhenguo: according to the error log, when we delelted a server, the port of the server may still in use and don11:41
liushengzhenguo: and don't be deleted immediatly11:41
liushengzhenguo: that will cause the cleanup process failed because temptest try to clean security group11:42
zhenguoliusheng: how do you judge the node already deleted?11:42
liushengzhenguo: we have disscussed about this, since the deletion api of server is a async call11:46
liushengzhenguo: Details: {u'message': u'Security Group 44270d21-12d3-490b-be66-d66a13e0da87 in use.', u'detail': u'', u'type': u'SecurityGroupInUse'}11:46
zhenguoliusheng: you check it go to deleted status or removed from db?11:46
liushengzhenguo: no, just send a server deletion request and check the http status11:47
zhenguoliusheng: that's not the right approach11:49
zhenguoliusheng: I remember we discussed this before11:50
zhenguoliusheng: the deleting API will return immediately11:50
liushengzhenguo: yes11:50
zhenguoliusheng: you need to check the server notfound11:51
liushengzhenguo: oh, seems I know the reason11:52
zhenguoliusheng: hah11:52
liushengzhenguo: I have do some changes to split the test cases11:52
zhenguoliusheng: ok,11:52
*** litao__ has quit IRC11:55
* zhenguo away12:00
openstackgerritliusheng proposed openstack/mogan master: Ensure the servers deleted completely in tempest's resource_cleanup  https://review.openstack.org/47380612:10
*** openstackgerrit has quit IRC12:18
*** openstackgerrit has joined #openstack-mogan12:19
openstackgerritMerged openstack/mogan master: Use PATCH method for flavor update  https://review.openstack.org/47215712:19
openstackgerritMerged openstack/mogan master: Change the missing node_type to resource_class  https://review.openstack.org/47332712:59

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