Tuesday, 2017-08-29

*** wanghao_ has quit IRC00:44
*** wanghao has joined #openstack-mogan00:44
zhenguomorning mogan!00:56
*** XueFengLiu has joined #openstack-mogan01:07
*** litao__ has joined #openstack-mogan01:27
*** wanghao has quit IRC01:32
Xinranzhenguo,  sorry, I just saw your message... I was not in company yestoday.02:11
Xinranzhenguo,  will abandon now :)02:12
zhenguoXinran: thanks!02:12
Xinranzhenguo,  np02:12
zhenguoliusheng: seems we can't create server via osc after scheduler_hints patch landed02:16
liushengzhenguo: yes02:16
zhenguoliusheng: please rebase you client patch02:17
liushengzhenguo: only can by api request02:17
liushengzhenguo: ok02:17
openstackgerritZhenguo Niu proposed openstack/python-moganclient master: Format server addresses  https://review.openstack.org/49865202:20
openstackgerritliusheng proposed openstack/python-moganclient master: Add support for scheduler hint  https://review.openstack.org/46355302:21
zhenguoliusheng: seems we can't handle server group gracefully,  e,g. if you just create one server with an affinity empty server group, you can't decide which affinity zone it placed, then following up server creating may just failed as there's not avaiallbe node02:27
liushengzhenguo: an affinity empty server group ?, if a server group is empty, scheduler need to select an affinity zone with enough available nodes02:31
zhenguoliusheng: yes, if I just create one server, it will radomly choose an affinity zone right?02:31
liushengzhenguo: yes02:31
zhenguoliusheng: maybe just choose one with only one availble node02:32
zhenguoliusheng: it's also not smart02:32
liushengzhenguo: you mean choose the affinity zone including least nodes ?02:33
zhenguoliusheng: we can't control02:34
liushengzhenguo: if so, we need to get tha total affinity_zone-nodes mapping at the begin of each scheduling like previous02:35
zhenguoliusheng: although you mean get the best affinity zone, but you can't control others choose nodes from it02:36
liushengzhenguo: yes02:36
zhenguoliusheng: nova also face the same problem02:36
liushengzhenguo: yes, the don't have the dedicated host concept02:37
zhenguoliusheng: how about we just providing the same functinalites like nova02:37
zhenguoliusheng: in Pike02:37
zhenguoliusheng: and remove the current get avaialble nodes for multi creation logic02:38
liushengzhenguo: yes, I think so, we can improve if there is performance bottleneck or we have another good idea02:38
liushengzhenguo: why need to remove ?02:38
zhenguoliusheng: it's too complex02:39
liushengzhenguo: not very complex, if you don't use multiple creation, the process is also very simple02:39
zhenguoliusheng:  maybe need others to review that codes02:40
liushengzhenguo: actually, at some places I think current way is similar with Nova02:40
liushengzhenguo: ok02:41
zhenguoliusheng: please help to look at the socat console patch, you can test it with our bare metal env02:42
liushengzhenguo: ok02:42
zhenguoliusheng: I changed the API partten02:42
liushengzhenguo: did you have tested sucessfully ?02:42
zhenguolitao__: for shellinagox, yes02:42
liushengzhenguo: hah02:43
zhenguolitao__: for socat console, I can connect it, but don't print it in the screen02:43
liushengzhenguo: you sent to litao, why ti cannot print02:43
zhenguoliusheng: I don't know how, lol02:44
zhenguoliusheng: need to use horizon02:44
liushengzhenguo: lol02:44
zhenguoliusheng: I only know how to connect to a socket02:44
zhenguoliusheng: not sure how to print it in the web02:44
liushengzhenguo: btw for your client address conversation patch, do you think we need to networ id in "server list" output ?02:44
zhenguoliusheng: seems nova does02:45
zhenguoliusheng: do you have a nova env?02:45
liushengzhenguo: yes, Nova use network name02:45
zhenguoliusheng: oh maybe we can get network name on client side02:45
liushengzhenguo: if a server with many nics, it may look ugly if we print the network id :(02:45
zhenguoliusheng: also for image and flavor02:45
zhenguoliusheng: we can get the name02:46
liushengzhenguo: yes, that is a way02:46
zhenguoliusheng: but if serve with many nics, you will pring the same ip address maybe02:46
liushengzhenguo: or maybe we get it in API side02:46
zhenguoliusheng: if there's no network02:46
liushengzhenguo: yes02:46
zhenguoliusheng: for API, uuid is enough02:47
zhenguoliusheng: client and dashboard can with name02:47
zhenguoliusheng: I find nova also transform flavor and image to name on client side02:47
liushengzhenguo: seems yes, so.. hah02:47
zhenguoliusheng: we need to improve our client02:48
liushengzhenguo: for address, maybe you can do in your patch ?02:48
zhenguoliusheng: yes02:48
liushengzhenguo: ok02:48
* zhenguo brb02:49
zhenguolitao__: please rebase the managing running servers patch when you got time03:24
zhenguoXinran: please remember to upload the server key_name patch03:24
litao__zhenguo: yes, I am rebasing , and update codes03:25
zhenguolitao__: thanks :D03:25
zhenguoliusheng: please help to improve the mogan client, for server group I will test it.03:26
*** wanghao has joined #openstack-mogan03:26
liushengzhenguo: ok, besides the scheduler hints support patch, other functionalities are all supportted03:27
zhenguoliusheng: I remember we discussed about the udpate/set returns03:28
zhenguoliusheng: currently we don't print anything03:28
liushengzhenguo: oh, yes03:28
liushengzhenguo: let me check03:28
zhenguoliusheng: you can help to compare with nova's osc one by one03:28
liushengzhenguo: ok03:28
zhenguoliusheng: maybe we should land all API changes by today, to reserve some time for testing03:29
Xinranzhenguo,  ok, will update today03:31
zhenguoXinran: thanks03:31
zhenguoXinran: both key_name and attaching interface with port_id are changing the APIs, so we should land them soon03:31
Xinranzhenguo,  ok, sorry for the delay03:33
zhenguoXinran: np, hah03:34
liushengzhenguo: ok03:36
zhenguoliusheng: I will go through our APIs to make sure there's no big issues with the format03:37
liushengzhenguo: ok, seems all the set/unset commands in OSC did print anything03:38
liushengzhenguo: I have tested server, flavor, aggregeate03:39
zhenguoliusheng: you mean nova osc?03:39
liushengzhenguo: yes03:39
zhenguoliusheng: if so, it's ok03:40
liushengzhenguo: ok03:40
zhenguoliusheng: should we also format the server show on client?03:52
openstackgerritZhenguo Niu proposed openstack/python-moganclient master: Format server addresses  https://review.openstack.org/49865203:52
* zhenguo brb03:53
liushengzhenguo: currently, it is jsut a raw dict, right ?03:54
liushengzhenguo: maybe it is better to format03:54
zhenguoliusheng: yes04:46
openstackgerritMerged openstack/python-moganclient master: Add support for scheduler hint  https://review.openstack.org/46355304:59
*** wanghao_ has joined #openstack-mogan06:00
*** wanghao has quit IRC06:04
openstackgerritZhenguo Niu proposed openstack/python-moganclient master: Format server addresses  https://review.openstack.org/49865206:19
openstackgerritZhenguo Niu proposed openstack/python-moganclient master: Format server addresses  https://review.openstack.org/49865206:27
zhenguoliusheng: can you try to set html_static to _build/html/_static on our RTD?06:39
liushengzhenguo: o, let me try06:39
zhenguoliusheng: I just run the command according to the builds log manually and find the dir06:39
openstackgerritXinran WANG proposed openstack/mogan master: support specifying port_id when attaching interface  https://review.openstack.org/49412106:40
* zhenguo brb06:41
liushengzhenguo: not sure if the dir is same in RTD06:41
zhenguoliusheng: in fact, you can find that from the build log WARNING06:50
liushengzhenguo: copying static files... WARNING: html_static_path entry '/home/docs/checkouts/readthedocs.org/readthedocs/templates/sphinx/_static' does not exist06:51
zhenguoliusheng: no wanring about the _build/html/_static dir?06:52
liushengzhenguo: WARNING: html_static_path entry '/home/docs/checkouts/readthedocs.org/readthedocs/templates/sphinx/_static' does not exist06:52
liushengzhenguo: above warning06:52
zhenguoliusheng: only this one?06:52
zhenguoliusheng: you can check the previous build log06:52
zhenguoliusheng: the page is still same?06:53
liushengzhenguo: yes06:53
zhenguoliusheng: can you share you priviate test project with me on RTD06:55
liushengzhenguo: ok06:56
liushengzhenguo: sure06:56
liushengzhenguo: done, the my-mogan-api-ref06:56
zhenguoliusheng: thanks, seems they can find the specified dir06:59
liushengzhenguo: do you know is there any tool we can conect to the RDT to see the build dir ?07:00
zhenguoliusheng: seems no07:01
liushengzhenguo: :(07:01
zhenguoliusheng: you can find the log of warning from the previous build log07:01
zhenguoliusheng: the can't find the specified dir before07:01
zhenguoliusheng: two warning log07:01
liushengzhenguo: yes, it just see cannot find the dir we specified. but we don't know what's is it real static file dir07:02
zhenguoliusheng: but from the new build log, you can just find one log warning about the readthedocs dir07:03
zhenguoliusheng: which indicates our specified dir is correct07:03
zhenguoliusheng: maybe I got the reason07:08
liushengzhenguo: what07:09
zhenguoliusheng: because the files we needed not in css dir but just under _static directly07:09
zhenguoliusheng: all files under css, images, js copied correctly07:10
liushengzhenguo: yes07:11
liushengzhenguo: in  local env, it is under _static/07:11
liushengzhenguo: oh, there is css/ dir07:12
zhenguoliusheng: but according to the html it will just find the file in _static/api-site.css07:16
Xinranzhenguo,  I got error when stack my local env,  baremetalcompute flavor create is not an openstack command07:35
zhenguoXinran: you need to update you python-moganclient07:35
Xinranzhenguo,  I have already update it07:36
zhenguoXinran: weird07:36
zhenguoXinran: maybe you can manually setup it07:36
Xinranzhenguo,  let me try :)07:36
zhenguoXinran: ok07:36
zhenguoliusheng: maybe we can just move docs to RTD for now07:37
zhenguoliusheng: and maybe changed to not using openstackdocs theme07:38
Xinranzhenguo,  still the same problem :(07:38
zhenguoXinran: oh07:38
liushengzhenguo: don't use openstackdoc theme it won't have the problem, right ?07:38
zhenguoXinran: after you update the repo and install it?07:38
zhenguoliusheng: api-ref still have problems07:39
zhenguoliusheng: docs we can just use readthecos theme07:39
Xinranzhenguo,  hah it works. I just update another branch of moganclient07:39
Xinranzhenguo, thanks07:40
zhenguoXinran: hah, np07:40
liushengzhenguo: hmm, maybe, and we can try to solve the problem if have enough time, and don't need to update the url07:40
zhenguoliusheng: yes07:41
zhenguoliusheng: for now, we can move docs there, but api-ref also maintained manually07:41
openstackgerritZhenguo Niu proposed openstack/mogan master: Temporarily remove openstackdocs theme  https://review.openstack.org/49815607:41
zhenguoliusheng: can you try above patch for docs?07:41
liushengzhenguo: there is a sample, seems it can use the openstack theme: http://cloudkitty.readthedocs.io/en/latest/07:42
liushengzhenguo: not sure why07:42
zhenguoliusheng: we can also use openstacktheme for docs, now warnings and errors on the page07:44
liushengzhenguo: oh, our problem only existed in api-ref07:44
zhenguoliusheng: yes, because os-api-ref extension07:44
zhenguoliusheng: we can maintain that before we are official07:45
zhenguoliusheng: maybe soon, hah07:45
liushengzhenguo: hah, yes07:46
zhenguoliusheng: we don't need to waste time to figure out why it doesn't work07:46
liushengzhenguo: hah07:46
zhenguoliusheng: but do you think it's better to not using openstackdocs theme before we are official?07:47
liushengzhenguo: http://my-mogan-api-ref.readthedocs.io/en/latest/07:47
liushengzhenguo: maybe it is better to provide a public api doc07:47
liushengzhenguo: in case we need to show some times07:47
zhenguoliusheng: do you find the theme is also ok although soem button is not work07:49
zhenguoliusheng: but at lest the REST verb can show now07:49
liushengzhenguo: yes07:49
zhenguoliusheng: ok, so let's change the theme07:50
liushengzhenguo: the detail button cannot work..07:50
zhenguoliusheng: it's ok07:50
openstackgerritZhenguo Niu proposed openstack/mogan master: Temporarily remove openstackdocs theme  https://review.openstack.org/49815607:52
liushengzhenguo: and just found, if we host our doc in RTD, it can be search by google. lol07:53
zhenguoliusheng: so we can land the above patch, and you need to rebase your patch to update the docs link07:53
zhenguoliusheng: hah07:53
zhenguoliusheng: does our doc links changed on RTD07:54
liushengzhenguo: ok07:54
liushengzhenguo: no, do yo prefer the subproject link or a saperated link for api-ref ?07:55
zhenguoliusheng: both are ok07:55
liushengzhenguo: if subproject link is ok, I don't need to update this patch: https://review.openstack.org/#/c/49345607:55
zhenguoliusheng: ok07:57
zhenguoliusheng: so let's just land them07:58
liushengzhenguo: ok07:59
zhenguoliusheng: on my test, this is ok https://review.openstack.org/#/c/498652/08:00
liushengzhenguo: ok, I am testing08:01
liushengzhenguo: don't format in "server show" ?08:03
zhenguoliusheng: not sure format to what, also like that?08:03
zhenguoliusheng: but maybe we need another patch to format all server stuff like image uuid to name and flavor uuid to name08:04
liushengzhenguo: seems for Nova vm, the OSC server show is same with "server list"08:04
liushengzhenguo: ok08:04
zhenguoliusheng: oh, if so, I can format the server show as well08:05
liushengzhenguo: ok08:05
zhenguoliusheng: maybe you can help to format the image and flavor08:05
liushengzhenguo: sure08:05
zhenguoliusheng: thanks08:05
liushengzhenguo: np :D08:06
zhenguoliusheng: for socat console, maybe you can just test the shellinabox part, as the api is same, we also need to add change the client08:06
zhenguowanghao: do you have time to write the client code for listing manageable servers?08:06
zhenguoliusheng, Xinran: seems luyao helped to update the attaching/detaching interface client patch08:07
liushengzhenguo: ok, will review08:08
zhenguoshaohe_feng: if you are around, please have a look at the two up patches which already got +208:08
*** Xinran has quit IRC08:08
*** Xinran has joined #openstack-mogan08:15
*** shaohe_feng has quit IRC08:16
*** shaohe_feng has joined #openstack-mogan08:19
liushengzhenguo: seems the luyao's patch have some issues, and may need the patch https://review.openstack.org/#/c/494121/ landed firstly, if luyao don't have time I can help to update the patch later08:19
zhenguoliusheng: ok, thanks08:20
openstackgerritZhenguo Niu proposed openstack/python-moganclient master: Format server addresses  https://review.openstack.org/49865208:41
* zhenguo brb08:45
openstackgerritMerged openstack/python-moganclient master: Format server addresses  https://review.openstack.org/49865208:55
zhenguolitao__: how's the managing running server's patch?09:00
openstackgerritZhenguo Niu proposed openstack/mogan master: Add socat console support  https://review.openstack.org/49383609:07
zhenguoliusheng: do we have a default affinity zone for servers not in any?09:10
liushengzhenguo: no09:10
zhenguoliusheng: do we need that?09:10
liushengzhenguo: maybe don't, if need, the default affinity zone including all nodes ?09:11
zhenguoliusheng: no, just for nodes not in any affinity zone09:11
zhenguoliusheng: seems not easy to control ha09:12
liushengzhenguo: but if we don't set any affinity zone, it will include all the node, right ? and if we add a node to a affinity zone, we need to remove it from the default one firstly09:12
liushengzhenguo: hah, yes09:13
zhenguoliusheng: will we support to specify same-affinity-zone or different-affinity-zone with a existing server id in scheduler hints when claiming a server?09:14
liushengzhenguo: is that different with affinity and anti-affinity ?09:15
zhenguoliusheng: that doens't need a server group09:15
liushengzhenguo: oh, seems like server creating with host specified in Nova. hah09:16
zhenguoliusheng: yes09:16
zhenguoliusheng: seems that's useful09:16
zhenguoliusheng: we don't always use server group09:16
liushengzhenguo: maybe we can consider in the future09:16
liushengzhenguo: yes09:17
zhenguoliusheng: if we have time left, we can do it, seems not complex09:17
zhenguoliusheng: we can check whether the specified server with an affinity zone, if not we just deny it09:17
liushengzhenguo: and do you think should we expose affinity APIs? or provide a api to show the relationship mapping with affinity zone and nodes09:17
zhenguoliusheng: seems they can use aggregate to show09:18
zhenguoliusheng: if we provide list aggregate by metadata key or just for affinity zone  and availability zone09:19
liushengzhenguo: seems it is not very convenient09:19
liushengzhenguo: yes, I mean that09:19
wanghao_zhenguo: I think I can do it, but maybe until next week.09:19
liushengzhenguo: like availability zone09:19
zhenguowanghao_: hah, if so we can help :D09:19
zhenguoliusheng: can we list which nodes in availability zone?09:19
zhenguowanghao_: we plan to cut release this week09:20
liushengzhenguo: no, that is what I think we should to improve09:20
zhenguoliusheng: maybe admins need that info09:20
liushengzhenguo: I think it is help to admins09:20
liushengzhenguo: hah, yes09:20
zhenguoliusheng: I will review the server group patch again, lol09:22
liushengzhenguo: hah, thanks09:22
zhenguoliusheng: hah09:22
*** wanghao_ has quit IRC09:23
*** wanghao has joined #openstack-mogan09:24
openstackgerritTao Li proposed openstack/mogan master: Manage existing BMs: Part-2  https://review.openstack.org/48154409:24
*** wanghao has quit IRC09:28
zhenguoliusheng: some comments inline09:32
liushengzhenguo: ok09:33
openstackgerritTao Li proposed openstack/mogan master: Manage existing BMs: Part-2  https://review.openstack.org/48154409:35
liushengzhenguo: thanks, replied inline, and will update the patch09:38
zhenguoliusheng: ok, thanks!09:38
zhenguoliusheng: Pike will be formally released tomorrow before 1500utc09:48
liushengzhenguo: so urgent!09:49
zhenguoliusheng: it's just for official, right?09:50
liushengzhenguo: not sure :(09:50
zhenguoliusheng: for us, we just create a branch09:50
liushengzhenguo: yes, it is09:50
openstackgerritXinran WANG proposed openstack/mogan master: support specifying port_id when attaching interface  https://review.openstack.org/49412109:51
* zhenguo brb10:18
openstackgerritliusheng proposed openstack/mogan master: Use server group in scheduler  https://review.openstack.org/49615110:18
openstackgerritMerged openstack/mogan master: Present Mogan docs on readthedocs  https://review.openstack.org/49345610:57
openstackgerritMerged openstack/mogan master: Temporarily remove openstackdocs theme  https://review.openstack.org/49815611:17
zhenguoliusheng: seems we need to make a new build for docs on RTD11:19
liushengzhenguo: why ?11:19
zhenguoliusheng: it can not automatically trigger the rebuild with new commits11:19
liushengzhenguo: oh, that need to click the button11:20
zhenguoliusheng: ok, I will try11:20
liushengzhenguo: but according to the RDT docs, it should be able to config a httphook11:20
zhenguoliusheng: seems we are not the owner of our repo, hah11:21
liushengzhenguo: hah, yes, it is belong to openstack11:22
zhenguoliusheng: hah11:22
liushengzhenguo: does that matter ?11:22
zhenguoliusheng: not sure if we can add the hook11:22
liushengzhenguo: let me try11:22
zhenguoliusheng: ok, thanks11:22
openstackgerritliusheng proposed openstack/mogan master: Use server group in scheduler  https://review.openstack.org/49615111:22
zhenguoliusheng: oh, build failed11:24
zhenguoliusheng: we use this http://mogan-api-ref.readthedocs.io/en/latest/ for api now11:44
zhenguoliusheng: seems as I removed api-ref as a subproject of mogan on RTD11:44
liushengzhenguo: so need to add the subproject ? or update the link ?11:46
zhenguoliusheng: you can add the subproject11:46
liushengzhenguo: ok11:46
zhenguoliusheng: I don't know why mogan docs build failed11:46
liushengzhenguo: any hint ?11:46
zhenguoliusheng: you can see the build logs11:47
openstackgerritXinran WANG proposed openstack/mogan master: support specifying port_id when attaching interface  https://review.openstack.org/49412111:49
liushengzhenguo: werid, it is ok previously11:50
openstackgerritZhenguo Niu proposed openstack/python-moganclient master: Change to use new console URI  https://review.openstack.org/49877511:50
zhenguoliusheng: yes11:50
liushengzhenguo: seems need to install the missed packages, but it is ok previously...11:55
zhenguoliusheng: you can try on you private repo with my change to comment the oenstackdocs theme11:56
zhenguoliusheng: not sure if it's related to that11:56
zhenguoXinran: I mean you can specify port_id when claiming a server11:58
zhenguoXinran: just like what you are doing for attaching interface11:59
*** litao__ has quit IRC11:59
Xinranzhenguo,  oh...12:00
Xinranzhenguo, create server with port_id in place of network ?12:00
zhenguoXinran: yes, we also support that12:01
zhenguoXinran: we already supported12:01
Xinranzhenguo,  ah..  didn't find it in code12:02
zhenguoXinran: http://mogan-api-ref.readthedocs.io/en/latest/v1/index.html#request12:03
zhenguoXinran: you mean the schema will check port_id and net_id not specified at the same time?12:07
Xinranzhenguo,  ok I see, in create_server.py12:07
zhenguoXinran: yes12:08
Xinranzhenguo,  I mean the schma will check if our in put is one of port_id and net_id12:08
Xinranzhenguo,  I will add another patch to fix this12:08
zhenguoXinran: ok, thanks12:09
zhenguoXinran:  have you tested with the schema when specified both?12:09
zhenguoXinran: seems the OneOf there just for required fields12:10
Xinranzhenguo,  yes I have tested.12:10
zhenguoXinran: oh, if so it's ok12:10
Xinranzhenguo,  it will cause error12:10
liushengzhenguo: seems another problem occured if I uncomment the http theme option12:10
liushengzhenguo: sigh :(12:10
zhenguoliusheng: :(12:11
zhenguoXinran: got it12:11
liushengzhenguo: https://readthedocs.org/projects/my-mogan-api-ref/builds/5898843/12:11
zhenguoliusheng: maybe just rebase my patch?12:12
liushengzhenguo: which one ?12:13
zhenguoliusheng: openstackdocs theme12:13
liushengzhenguo: I have forked the latest code of mogan, and uncomment the line12:13
zhenguoliusheng: oh, sorry, I mean revert12:14
liushengzhenguo: it is equal revert :(12:14
zhenguoliusheng: uncomment also cause problems?12:14
liushengzhenguo: another problem12:15
liushengzhenguo: ^^^12:15
liushengzhenguo: https://readthedocs.org/projects/my-mogan-api-ref/builds/5898843/12:15
zhenguoliusheng: if so, our gate should be broken?12:16
liushengzhenguo: no, seems only RTD cannot build12:16
zhenguoliusheng: oh12:17
*** XueFengLiu has quit IRC16:16
*** XueFengLiu has joined #openstack-mogan16:19

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