Thursday, 2017-09-07

*** wanghao_ has quit IRC00:43
*** harlowja has quit IRC00:48
*** wanghao has joined #openstack-mogan00:49
openstackgerritwanghao proposed openstack/python-moganclient master: Support to getting managebale servers  https://review.openstack.org/49947401:05
*** litao__ has joined #openstack-mogan01:05
wanghaomorning!01:06
wanghaoI saw Sean's patch.  hahaha01:06
zhenguowanghao: lol01:09
wanghaozhenguo: zan01:09
openstackgerritliusheng proposed openstack/mogan master: Remove redundant mock in functional test  https://review.openstack.org/50151201:34
litao__liusheng: do you resolved the bugs in patch 34 in manageable_server part-2?01:37
liushenglitao__: not yet, just want want to check, if you have solution, please pick it again01:38
litao__liusheng: ok, I will try01:38
liushenglitao__: ok01:38
openstackgerritliusheng proposed openstack/mogan master: Remove redundant mock in functional test  https://review.openstack.org/50151201:46
openstackgerritliusheng proposed openstack/python-moganclient master: Don't need features description in README  https://review.openstack.org/50152601:49
zhenguoliusheng: oh, it failed again for mogan docs on RTD01:52
liushengzhenguo: :(01:52
zhenguoliusheng: the same error like before01:53
liushengzhenguo: will check01:53
zhenguoliusheng: thanks01:54
openstackgerritliusheng proposed openstack/python-moganclient master: Add a more detailed README doc  https://review.openstack.org/50152601:57
zhenguohi all, let's move to #openstack-meeting02:00
wanghaook02:00
*** litao__ has quit IRC02:53
*** litao__ has joined #openstack-mogan02:53
openstackgerritMerged openstack/mogan master: Use newer location for iso8601 UTC  https://review.openstack.org/50148003:16
liushengzhenguo: you have moved mogan doc to your own repo, I cannot manage it03:17
openstackgerritMerged openstack/mogan-specs master: Add server group support  https://review.openstack.org/48954103:19
zhenguoliusheng: hah,03:22
zhenguoliusheng: please fee free to remove my repo and add yours03:22
openstackgerritMerged openstack/mogan-specs master: No need flavor and network in managing server  https://review.openstack.org/49498003:23
liushengzhenguo: let me try in a test repo firstly :D03:23
zhenguoliusheng: ok03:23
liushengzhenguo: I found some times, the RTD after building, need wait a while, otherwise there maybe some issues03:24
zhenguoliusheng: seems yes03:25
openstackgerritShangXiao proposed openstack/mogan-specs master: Fix the typo and replace http with https for doc links  https://review.openstack.org/50154703:32
openstackgerritZhenguo Niu proposed openstack/mogan master: Add db migration doc  https://review.openstack.org/50155003:39
openstackgerritliusheng proposed openstack/python-moganclient master: Remove the unsupported python 3.4 identifier  https://review.openstack.org/50155103:39
*** harlowja has joined #openstack-mogan04:06
*** dims has quit IRC04:40
*** harlowja has quit IRC05:11
liushengzhenguo: in my testing, sometimes it may fail because missing oslo_policy, but somtimes don't, strange06:17
zhenguoliusheng: hah, yes06:17
liushengzhenguo: how about just adding oslo_policy in test-requirements.txt ?06:28
zhenguoliusheng: you didn't tried that?06:28
liushengzhenguo: not yet06:29
liushengzhenguo: cannot work ?06:29
zhenguoliusheng: you can try on your own repo06:29
openstackgerritMerged openstack/mogan master: Remove redundant mock in functional test  https://review.openstack.org/50151206:29
zhenguoliusheng: I remember I told you to try that, hah06:29
liushengzhenguo: oh, seems I have tried before06:29
liushengzhenguo: but another error06:29
zhenguoliusheng: yes, seems oslo.cache error06:30
litao__zhenguo: ping06:36
zhenguolitao__: pong06:36
litao__zhenguo: I found an question. When I server.create() is executed, the nic object can't be saved in db06:51
zhenguolitao__: oh, yes, because there's a foreign key06:52
litao__zhenguo: should i call nic.save() ?06:53
zhenguolitao__: no you can't06:53
litao__zhenguo: How to do?06:53
zhenguolitao__: no way06:53
zhenguolitao__: you need to save server first06:54
zhenguolitao__: oh, seems we don't talk about the same thing06:55
zhenguolitao__: sorry, liusheng can you explain why here06:55
litao__liusheng: ping06:55
litao__zhenguo: I know why, I should update the the latest the mogan client codes06:57
zhenguolitao__: why need that06:57
zhenguolitao__: you mean run CLI failed to save db?06:57
litao__zhenguo: I mean I can't get network info from CLI06:58
zhenguolitao__: oh, yes, you need to upgrade the client06:58
litao__zhenguo: The network keyword is change to address, and I didn't pull the mogan client codes06:59
litao__zhenguo: yes, It works06:59
zhenguolitao__: cool06:59
litao__zhenguo: lol07:00
liushenglitao__: still have problem ? :P07:02
openstackgerritTao Li proposed openstack/mogan master: Manage existing BMs: Part-2  https://review.openstack.org/48154407:03
openstackgerritZhenguo Niu proposed openstack/mogan master: Correct db and model inconsistent  https://review.openstack.org/50159707:21
zhenguoliusheng: please have a look at this https://review.openstack.org/#/c/501550/07:29
zhenguoliusheng: and with the instructions there, I find many db and model inconsistent https://review.openstack.org/#/c/501597/07:30
liushengzhenguo: ok, why shaohe -A ? :D07:31
liushengzhenguo: oh07:31
zhenguoliusheng: no -A, he just want to wait for another +2 @shaohe_feng07:32
liushengzhenguo: got it, hah07:32
zhenguoliusheng: how about this one https://review.openstack.org/#/c/499474/ ?07:33
* zhenguo brb07:33
liushengzhenguo: it is better also update the doc07:33
openstackgerritTao Li proposed openstack/mogan master: Manage existing BMs: Part-2  https://review.openstack.org/48154407:36
openstackgerritMerged openstack/mogan master: Add db migration doc  https://review.openstack.org/50155007:36
zhenguoshaohe_feng: please help to look at this https://review.openstack.org/#/c/501551/07:46
zhenguoliusheng: oh, seems we can't specify port_id when creating a server through CLI07:50
openstackgerritXinran WANG proposed openstack/mogan master: uncomment check_port_availability  https://review.openstack.org/50161407:51
openstackgerritXinran WANG proposed openstack/mogan master: Update port['device_id'] when creating server  https://review.openstack.org/50066807:53
liushengzhenguo: I remember ther is a patch added that07:53
zhenguoliusheng: got lost?07:53
zhenguoliusheng: and there's a port_type still there now07:53
liushengzhenguo: oh, seems yes, we don't support specifying port-id when creating07:58
zhenguoliusheng: can you help to add that08:00
liushengzhenguo: sure08:00
zhenguoliusheng: thanks08:00
zhenguoliusheng: I will add partitions support08:00
liushengzhenguo: np08:00
liushengzhenguo: ok08:01
litao__zhenguo,liusheng: For managing servers , should we use a new resource named manageable server or server?08:01
liushenglitao__: where you mean ?08:01
litao__liusheng: In mogan client08:02
zhenguolitao__: seems you can follow up wanghao's patch08:03
litao__zhenguo: OK08:10
litao__zhenguo: https://review.openstack.org/#/c/481544/36 pls review this patch08:28
zhenguolitao__: ok08:28
zhenguolitao__: seems it not pass the gate08:29
litao__zhenguo: yes, I recheck it08:29
zhenguolitao__: ok08:29
litao__zhenguo: I can't find why it failed08:29
openstackgerritZhenguo Niu proposed openstack/python-moganclient master: Add support for specifing patitions when claming a server  https://review.openstack.org/50163408:47
liushengzhenguo: seems our scheduler has been broken :(08:53
zhenguoliusheng: why?08:53
liushengzhenguo: not sure maybe relate with the node-uuid-->node-name change08:54
liushengzhenguo: not sure08:54
openstackgerritZhenguo Niu proposed openstack/mogan master: Correct db and model inconsistent  https://review.openstack.org/50159708:54
zhenguoliusheng: oh08:54
liushengSep 07 16:53:07 szxbzci0007 mogan-scheduler[18308]: 2017-09-07 16:53:07.081 WARNING mogan.scheduler.client.report [req-1f7d453a-027a-4f28-9d64-94f13e6478e2 None None] Unable to submit allocation for server fa9d68a4-30ce-4927-9461-1028fd08:54
liushengSep 07 16:53:07 szxbzci0007 mogan-scheduler[18308]:  <head>08:54
liushengSep 07 16:53:07 szxbzci0007 mogan-scheduler[18308]:   <title>400 Bad Request</title>08:54
liushengSep 07 16:53:07 szxbzci0007 mogan-scheduler[18308]:  </head>08:54
liushengSep 07 16:53:07 szxbzci0007 mogan-scheduler[18308]:  <body>08:55
liushengSep 07 16:53:07 szxbzci0007 mogan-scheduler[18308]:   <h1>400 Bad Request</h1>08:55
liushengSep 07 16:53:07 szxbzci0007 mogan-scheduler[18308]:   The server could not comply with the request since it is either malformed or otherwise incorrect.<br /><br />08:55
liushengSep 07 16:53:07 szxbzci0007 mogan-scheduler[18308]: JSON does not validate: u'node-0' is not a 'uuid'08:55
openstackgerritwanghao proposed openstack/python-moganclient master: Support to getting managebale servers  https://review.openstack.org/49947408:55
liushengSep 07 16:53:07 szxbzci0007 mogan-scheduler[18308]: Failed validating 'format' in schema['properties']['allocations']['items']['properties']['resource_provider']['properties']['uuid']:08:55
liushengSep 07 16:53:07 szxbzci0007 mogan-scheduler[18308]:     {'format': 'uuid', 'type': 'string'}08:55
liushengSep 07 16:53:07 szxbzci0007 mogan-scheduler[18308]: On instance['allocations'][0]['resource_provider']['uuid']:08:55
liushengSep 07 16:53:07 szxbzci0007 mogan-scheduler[18308]:     u'node-0'08:55
zhenguoliusheng: but this pass the gate 20 mins ago https://review.openstack.org/#/c/500668/08:55
liushengzhenguo: yes, it can create a server, but cannot create 2ed server08:55
liushengzhenguo: let me check, not sure08:56
zhenguoliusheng: oh, you mean why put allocations we need the rp uuid instead of name, right?08:57
zhenguo*when08:57
liushengzhenguo: seems yes08:57
zhenguoliusheng: need to change the put_allocations method?08:58
zhenguoliusheng: and seems we can combine some methods in scheduler/report08:58
liushengzhenguo: maybe it is relate with placement api version, let me check08:58
zhenguoliusheng: but we don't bump the API version at all08:59
zhenguoliusheng: we use server.node_uuid before, so it works08:59
zhenguoliusheng: need to find rp uuid from server now08:59
liushengzhenguo: nova/api/openstack/placement/handlers/allocation.py09:00
liushengzhenguo: ALLOCATION_SCHEMA09:00
zhenguoliusheng: yes, I see, it needs rp uuid09:01
liushengzhenguo: yes :(09:01
zhenguoliusheng: but we pass server node now09:01
zhenguoliusheng: we should change09:01
zhenguoliusheng: as they just want rp.uuid instead of rp.name09:01
liushengzhenguo: yes09:01
zhenguoliusheng: is that easy to find the uuid from name?09:02
zhenguoliusheng: leveraging our cache?09:02
liushengzhenguo: scheduler cannot use the cache09:03
zhenguoliusheng: we always put allocations from engine side, right?09:03
liushengzhenguo: not is in scheduler09:03
liushengzhenguo: not sure should we move it to engine09:03
zhenguoliusheng: let me check09:04
zhenguoliusheng: seems you can find the uuid there, as we got the rp object09:05
zhenguoliusheng: filtered_nodes and filtered_affzs_nodes can contain both uuid and name as a workaround09:06
liushengzhenguo: yes, we can, but need to change the process09:06
zhenguoliusheng: sigh, it's tricky09:10
zhenguoliusheng: maybe revert that patch and just set name on engine side before return to API09:11
liushengzhenguo: sigh, I am afraid that may be some conflicts ?09:11
openstackgerritMerged openstack/mogan master: uncomment check_port_availability  https://review.openstack.org/50161409:12
zhenguoliusheng: seems yes09:12
zhenguoliusheng: or maybe add a new patch09:12
zhenguoliusheng: we don't need to change server.node back to server.node_uuid09:12
zhenguoliusheng: just save uuid in db09:12
liushengzhenguo: I am afraid that may cause more inconsistent in the future, how about your first suggestion ?09:13
liushengzhenguo: revert that patch, and convert node_uuid to node_name only in api layer09:14
zhenguoliusheng: you can't, as you also can't use cache on API side09:14
liushengzhenguo: call ironic in client side like others ?09:15
zhenguoliusheng: hah09:15
zhenguoliusheng: so, let's revert it09:19
liushengzhenguo: ok09:19
openstackgerritZhenguo Niu proposed openstack/mogan master: Revert "Return node(name or uuid) with server instead of node_uuid"  https://review.openstack.org/50164209:20
zhenguoliusheng: conflicts happening09:21
liushengzhenguo: hah09:21
zhenguoliusheng: I'll be away for a while, if you got time, please help to address the conflicts, thanks09:22
* zhenguo brb09:23
litao__zhenguo: passed09:27
openstackgerritliusheng proposed openstack/mogan master: Revert "Return node(name or uuid) with server instead of node_uuid"  https://review.openstack.org/50164209:29
openstackgerritliusheng proposed openstack/mogan master: Revert "Return node(name or uuid) with server instead of node_uuid"  https://review.openstack.org/50164209:31
zhenguoliusheng: thanks09:35
liushengzhenguo: np09:36
zhenguolitao__: not sure if the revert above will affect you patch09:36
openstackgerritMerged openstack/python-moganclient master: Remove the unsupported python 3.4 identifier  https://review.openstack.org/50155109:42
zhenguolitao__: hi, I left some nits inline, will test in my env.09:51
openstackgerritliusheng proposed openstack/python-moganclient master: Correct the network parameter when creating server  https://review.openstack.org/50165809:52
liushengzhenguo: plz help to review above patch if you got time ^09:53
zhenguoliusheng: sure09:54
liushengzhenguo: thanks09:54
liushengzhenguo, wanghao can you check the reply of Xinran in the patch https://review.openstack.org/#/c/50066809:58
zhenguoliusheng: ok09:58
liushengzhenguo: not sure if I can +A, but we should respect any negative opinion. hah10:00
zhenguoliusheng: seems wanghao is not online10:00
liushengzhenguo: oh10:00
zhenguoliusheng: oh, seems revert will delelte the port again10:07
* zhenguo brb10:33
Xinranzhenguo,  wanghao  Thanks for your comments on my patch10:59
zhenguoXinran: seems the port will get deleted during the revert process10:59
Xinranzhenguo,  yes I see11:00
zhenguoXinran: not sure if we should delete it11:00
Xinranzhenguo,  not sure why revert will destroy all11:00
Xinranzhenguo,  I think no ...11:01
zhenguoXinran: so we should handle that during revert11:01
Xinranif we pass net_id, will we invoke destroy_network as before?11:04
XinranI think unbind_port and delete_port is enough11:05
Xinranand need we a for loop to filter the requested_networks in revert ?11:06
openstackgerritZhenguo Niu proposed openstack/mogan master: Correct db and model inconsistent  https://review.openstack.org/50159711:07
zhenguoXinran: you mean during revert?11:09
zhenguoXinran: revert will only check the nics object and clean everything11:09
Xinranok,  I think we should just destroy the network whick failed to be built, not all11:10
zhenguoliusheng: maybe we can land this https://review.openstack.org/#/c/501642/ now11:10
zhenguoXinran: we should remove all11:11
Xinranzhenguo,  why, can't understand ...11:14
zhenguoXinran: during the revert we will clean all things, otherwise there will be ports left11:14
zhenguoXinran: I mean what we have created during the server creation except the existing port11:15
Xinranzhenguo,  oh my question is if we should remove all including the sucessfully built network11:16
zhenguoXinran: yes, we just want to remove successfully built network11:17
liushengzhenguo: ok11:17
Xinranzhenguo,  why don't we just remove the error one ?11:19
zhenguoXinran: in fact, error one doesn't need to be removed, as they may just not created11:20
zhenguoXinran: seems I understand what you are asking11:21
zhenguoXinran: our workflow is a list of execute and reverts11:21
zhenguoXiran: when there's an exception happend in buildnetworks, we will revert one by one,11:21
zhenguoXinran: which means we will remove everything, ports and the server11:22
Xinranmeans total execution failed11:22
zhenguoXinran: yes11:22
Xinranzhenguo,  hah ok got it11:22
zhenguoXinran: but revert will just start from the point failed11:22
Xinranif so,  I think we need a for loop in revert11:24
Xinranto check every port11:24
zhenguoXinran: seems there's a look in destroy_networks11:27
zhenguoXinran: maybe we need a separated destroy_networks11:27
zhenguoliusheng: when deleting a server will we remove the specified port/existing neutron port11:27
liushengzhenguo: not sure, let me try11:28
zhenguoliusheng: ok, thanks11:28
zhenguoliusheng: are you testing with nova+ironic or single nova11:29
zhenguoliusheng: not sure if ironic will get rid of the port11:29
liushengzhenguo: oh, you mean in Nova ?11:29
zhenguoliusheng: yes11:29
liushengzhenguo: I don't have a nova+ironic env11:30
zhenguoliusheng: ok11:30
Xinranzhenguo,  you mean server or baremetal server?11:30
zhenguoXinran: mogan server11:31
Xinranzhenguo, yes11:31
Xinranzhenguo, I just tried11:31
Xinranzhenguo,  the associated port will be deleted11:31
liushengzhenguo: just asked, if creating a nova server  with port specified. after deleting server, the port still exist11:31
liushengzhenguo: if specifying network, it will delete the port11:32
zhenguooh, should we make a change11:32
liushengzhenguo: currently, mogan will delete the port ?11:32
zhenguoXinran: yes, we will delete all11:33
zhenguohow can we know it's a existing port when deleting11:33
Xinranin Nova  there is a preexisting_port which save the existing port before do operation11:33
zhenguosave in db?11:34
Xinranzhenguo, seems just a variable hah11:34
zhenguobut when I want to delete a server, how can I know which port is an existing one11:34
zhenguoliusheng: this https://review.openstack.org/#/c/501597/ passed the gate, and fixed a server group policy foreign key error.11:47
*** litao__ has quit IRC11:48
Xinranzhenguo,  https://github.com/openstack/nova/blob/7506ffed5210fc6c67bf89d3b94a0230b3ab5592/nova/network/model.py#L39511:49
Xinranzhenguo,  pls see this code11:50
zhenguoXinran: oh, they have such field with VIF11:50
zhenguoXinran: seems we can also add that to our Nic object11:50
Xinranzhenguo,  yes sure11:51
zhenguoXinran: thanks11:51
Xinranzhenguo,  but my intern report's deadline is coming, also my final presentation11:52
zhenguoXinran: oh, it's ok, I can help to update the patch11:52
zhenguoXinran: :D11:52
Xinranzhenguo,  thanks a lot.11:52
zhenguoXinran: np11:52
Xinranzhenguo,  I will be all time avaible after my presentation11:53
zhenguoXinran: ok11:53
Xinranzhenguo,  thanks  thanks thanks11:53
zhenguoXinran: lol11:53
*** dims has joined #openstack-mogan12:07
openstackgerritOpenStack Proposal Bot proposed openstack/python-moganclient master: Updated from global requirements  https://review.openstack.org/50135813:42
openstackgerritMerged openstack/mogan master: Revert "Return node(name or uuid) with server instead of node_uuid"  https://review.openstack.org/50164213:46
openstackgerritTao Li proposed openstack/mogan master: Manage existing BMs: Part-2  https://review.openstack.org/48154415:25
openstackgerritTao Li proposed openstack/python-moganclient master: Add the CLI for server adoption  https://review.openstack.org/50177715:30
openstackgerritTao Li proposed openstack/python-moganclient master: Add the CLI for managing server  https://review.openstack.org/50177715:31
openstackgerritTao Li proposed openstack/python-moganclient master: Add the CLI for managing server  https://review.openstack.org/50177716:47
*** harlowja has joined #openstack-mogan17:19
*** dims has quit IRC18:34
*** dims has joined #openstack-mogan18:36
*** dims has quit IRC18:41
*** dims has joined #openstack-mogan18:45
*** wanghao_ has joined #openstack-mogan20:54
*** wanghao has quit IRC20:56

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