| *** Qiming has quit IRC | 00:34 | |
| *** yanyanhu has joined #senlin | 01:04 | |
| *** zzxwill has joined #senlin | 01:23 | |
| *** yanyanhu has quit IRC | 01:26 | |
| *** yanyanhu has joined #senlin | 01:27 | |
| *** Qiming has joined #senlin | 01:31 | |
| openstackgerrit | Qiming Teng proposed openstack/python-senlinclient: Revert "Remove senlin CLI commands" https://review.openstack.org/300281 | 02:01 |
|---|---|---|
| zzxwill | We will keep the CLI for a while until openstacksdk is ready? | 02:04 |
| Qiming | yes, zzxwill | 02:19 |
| Qiming | it is not about openstacksdk, it is about this one: https://review.openstack.org/#/c/298553 | 02:19 |
| yanyanhu | hi, Qiming, round? | 02:26 |
| yanyanhu | around | 02:26 |
| yanyanhu | looks like we can't get action id from the cluster_delete senlinclient interface? | 02:27 |
| yanyanhu | I found by default, Rally relies on checking resource status switching e.g. 'DELETE_COMPLETE' to decide whether the resource has been deleted completely | 02:28 |
| yanyanhu | so we may need to replacing it by checking action status switching to 'SUCCEEDED' | 02:29 |
| yanyanhu | since the db item of cluster will be removed after cluster is deleted | 02:30 |
| yanyanhu | so is that possible to expose action_id to enduser from senlinclient when performing resource deletion operations | 02:32 |
| yanyanhu | my fault, I noticed that resource call in cluster resource of sdk has included location into the return value | 02:44 |
| yanyanhu | so I think we can get action_id from that attire | 02:45 |
| *** yuanying has quit IRC | 02:51 | |
| Qiming | okay | 02:56 |
| *** yanyanhu has quit IRC | 03:40 | |
| *** yanyanhu has joined #senlin | 03:41 | |
| *** yuanying has joined #senlin | 03:49 | |
| *** zzxwill has quit IRC | 04:00 | |
| *** zzxwill has joined #senlin | 04:20 | |
| *** zzxwill has quit IRC | 04:38 | |
| *** zzxwill has joined #senlin | 04:54 | |
| yanyanhu | hi, Qiming, around? | 05:22 |
| Qiming | yes | 05:25 |
| yanyanhu | hi, Qiming, I'm wondering whether we need to refactor the parameter definition in senlinclient interfaces | 05:28 |
| yanyanhu | http://git.openstack.org/cgit/openstack/python-senlinclient/tree/senlinclient/v1/client.py | 05:28 |
| yanyanhu | it's very difficult to just refer to the code to know how to invoke those interfaces | 05:29 |
| yanyanhu | all input parameters are now included in a single dictionary attrs | 05:29 |
| Qiming | yes, that is the way we make the interface consistent and stable | 05:30 |
| yanyanhu | I check the code of senlin proxy interfaces in sdk side, also the same | 05:30 |
| yanyanhu | yes, if so I think we may need some docstring to help user understand it | 05:30 |
| Qiming | they are the same so that in future people can avoid depending on senlinclient | 05:30 |
| yanyanhu | e.g. which parameters are optional and which are not | 05:30 |
| Qiming | for developers invoking senlin api, they are going to explore the API doc | 05:31 |
| yanyanhu | understand. I'm now invoking senlin client in Rally and found I need to check code of Senlin API to know which parameters I have to provide :) | 05:31 |
| Qiming | or you can check api doc? | 05:31 |
| yanyanhu | yes, there is no problem for API invoking since the description is clear in API doc | 05:32 |
| Qiming | reading source code for developers is a pain, we know that ... | 05:32 |
| yanyanhu | I see. so maybe we need give user a hint that the parameter definition in senlinclient interface is completely consistent with the API interfaces | 05:33 |
| Qiming | that is implied | 05:33 |
| Qiming | we are always striving to make things consistent | 05:33 |
| Qiming | if there are inconsistency, we should treat them as bugs | 05:33 |
| Qiming | maybe I can find a better reference for you | 05:33 |
| yanyanhu | yes. We know that. Just concern maybe a new user don't know they are same :) | 05:33 |
| yanyanhu | since senlinclient interface is actually different from other xxxclient | 05:34 |
| Qiming | why there is such a concern? | 05:35 |
| Qiming | you mean the code are different from the doc? | 05:36 |
| yanyanhu | because usually when a guy first tries to invoke senlinclient interface to talk with senlin, the first place they try to find reference is here: http://git.openstack.org/cgit/openstack/python-senlinclient/tree/senlinclient/v1/client.py | 05:37 |
| Qiming | okay | 05:38 |
| yanyanhu | but they can't find any doc there. Then they need to check the code of sdk to understand the 'attrs' are exactly consistent with those properties that described in senlin API doc | 05:38 |
| Qiming | so what you are talking about is to better documenting the client api | 05:38 |
| Qiming | we have done that | 05:38 |
| Qiming | in sdk side | 05:38 |
| yanyanhu | exactly | 05:38 |
| Qiming | but I cannot find the generated code at the moment | 05:38 |
| Qiming | give me 5 minutes okay? | 05:38 |
| yanyanhu | ok | 05:38 |
| Qiming | http://developer.openstack.org/sdks/python/openstacksdk/users/proxies/cluster.html | 05:42 |
| yanyanhu | yes, this is generated from the docstring of proxy calls of senlin service | 05:44 |
| Qiming | that is about the interface calls | 05:44 |
| Qiming | if we create another set of docs at senlinclient side | 05:45 |
| Qiming | how are we planning to maintain them? | 05:45 |
| yanyanhu | maybe just some parameter description embedded inline? | 05:45 |
| yanyanhu | actually not necessary | 05:45 |
| Qiming | each time some parameter changed at server side, we will need to modify the server api, bump the api version, update the api doc and update the sdk proxy then senlinclient docstring? | 05:46 |
| yanyanhu | just letting user know they can directly refer the API doc is ok | 05:46 |
| Qiming | I think they should know that | 05:46 |
| yanyanhu | ok, I see. | 05:46 |
| yanyanhu | not very diffult | 05:47 |
| Qiming | we have done an excellent job synchronizing the api doc with the code | 05:47 |
| yanyanhu | yep | 05:47 |
| Qiming | if there is something we can improve and we should do | 05:47 |
| Qiming | it would be add some pointers in the senlinclient.v1.client module to the location of the api doc | 05:47 |
| Qiming | we should not duplicate the docs here and there | 05:48 |
| yanyanhu | that would be the best | 05:48 |
| Qiming | it would be a huge headache moving forward | 05:48 |
| yanyanhu | right | 05:48 |
| Qiming | you have to remember that there are other docs out of sync | 05:48 |
| Qiming | for example, for the cluster_create method | 05:50 |
| Qiming | we can add a docstring: http://developer.openstack.org/api-ref-clustering-v1.html#createCluster | 05:50 |
| Qiming | in the API doc, we have documented almost everything, the parameters needed (optional), data type, meanings, default values (if any) and even sample bodies | 05:52 |
| yanyanhu | Yes | 05:52 |
| openstackgerrit | Yanyan Hu proposed openstack/python-senlinclient: Add a link to API doc in client module https://review.openstack.org/300315 | 06:01 |
| yanyanhu | hi, Qiming, just propose a patch to add that description. | 06:01 |
| yanyanhu | but feel it's not that noticeable... | 06:02 |
| Qiming | yes | 06:03 |
| Qiming | it is not achieving the goal we talked above | 06:04 |
| yanyanhu | so maybe a line for each methods | 06:05 |
| Qiming | I'm afraid so | 06:06 |
| yanyanhu | I see. Let me add them | 06:07 |
| Qiming | btw, I'm a little bit concerned about the deprecation of senlinclient cli now | 06:10 |
| Qiming | I was very optimistic about it before and I believed that is the right direction to go | 06:10 |
| yanyanhu | since some users have used it? | 06:10 |
| Qiming | however, seeing that a single openstack command seemingly integrated all other client commands | 06:11 |
| Qiming | I'm doubting about it now | 06:11 |
| Qiming | whether that is something that will eventually happen | 06:12 |
| Qiming | consider we have all service clients with OSC plugins implemented | 06:12 |
| Qiming | how can anybody avoid conflicts between commands and/or options if they are all implemented in their own client code base | 06:13 |
| yanyanhu | this is a problem... | 06:13 |
| Qiming | if the thing is to be integrated, it has better be done within openstackclient project | 06:13 |
| Qiming | we are blocked by the --profile parameter there | 06:14 |
| Qiming | I don't see a good chance to rename that to --os-profile as suggested, BEAUSE IT HAS BEEN RELEASED | 06:14 |
| Qiming | we cannot make change either, BECAUSE WE HAVE HAD SENLINCLIENT RELEASED | 06:14 |
| yanyanhu | ... | 06:15 |
| Qiming | imagine this, when magnum, mistral, trove, zaqar, heat, sahara, barbican, ... all pouring into the single 'openstack' command | 06:15 |
| Qiming | what will happen? | 06:15 |
| openstackgerrit | EldonZhao proposed openstack/python-senlinclient: Add action-id to output of node-delete https://review.openstack.org/300320 | 06:16 |
| yanyanhu | Qiming, that's true. Unless openstackclient can help to coordinate | 06:23 |
| openstackgerrit | EldonZhao proposed openstack/python-senlinclient: Add action-id to output of node-delete https://review.openstack.org/300320 | 06:26 |
| *** yanyanhu has quit IRC | 06:28 | |
| *** yanyanhu has joined #senlin | 06:31 | |
| openstackgerrit | EldonZhao proposed openstack/python-senlinclient: Add action-id to output of node-delete https://review.openstack.org/300320 | 06:43 |
| openstackgerrit | EldonZhao proposed openstack/python-senlinclient: Add action-id to output of node-delete https://review.openstack.org/300320 | 06:58 |
| openstackgerrit | EldonZhao proposed openstack/python-senlinclient: Add action-id to output of node-delete https://review.openstack.org/300320 | 07:06 |
| *** zzxwill has quit IRC | 07:34 | |
| *** zzxwill has joined #senlin | 07:38 | |
| *** yanyanhu has quit IRC | 07:59 | |
| *** yanyanhu has joined #senlin | 08:00 | |
| *** zzxwill has quit IRC | 08:31 | |
| openstackgerrit | Yanyan Hu proposed openstack/python-senlinclient: Add link to API doc in client module https://review.openstack.org/300315 | 08:37 |
| openstackgerrit | EldonZhao proposed openstack/senlin: Protect node form deleting when node status is creating https://review.openstack.org/300359 | 08:37 |
| openstackgerrit | EldonZhao proposed openstack/python-senlinclient: Add action-id to output of node-delete https://review.openstack.org/300320 | 08:41 |
| yanyanhu | hi, Qiming, the API doc of profile creating is incorrect? | 08:49 |
| yanyanhu | http://developer.openstack.org/api-ref-clustering-v1.html#createProfile | 08:49 |
| yanyanhu | there is no 'profile' parameter for createProfile API call? | 08:49 |
| yanyanhu | should be context I think | 08:50 |
| Qiming | ? | 08:51 |
| yanyanhu | I mean the request parameters | 08:51 |
| yanyanhu | there is a parameter named as 'profile' | 08:52 |
| Qiming | yes | 08:53 |
| yanyanhu | maybe that should be context? | 08:53 |
| Qiming | it should be context | 08:53 |
| yanyanhu | ok, will try to fix it | 08:53 |
| yanyanhu | so this should be a patch to doc project? | 08:53 |
| Qiming | wait a minute | 08:53 |
| yanyanhu | ok | 08:54 |
| Qiming | the api is not accurate | 08:54 |
| Qiming | for profile create request | 08:54 |
| Qiming | the only parameter allowed is a dict named 'profile' | 08:54 |
| Qiming | inside that profile dict, there are up to 4 keys: name, spec, metadata, context | 08:55 |
| Qiming | so ... it is a difference between REST API and Python API | 08:55 |
| Qiming | for REST API, we are accepting a {'profile': {...}} JSON as body | 08:56 |
| Qiming | for Python API, we are accepting (name=, spec=, context=, metadata=) | 08:56 |
| yanyanhu | I see | 08:57 |
| Qiming | and ... | 08:57 |
| yanyanhu | so just need to replace profile with context | 08:57 |
| Qiming | even context is not part of profile | 08:57 |
| yanyanhu | the example request is correct | 08:57 |
| Qiming | s/not/now/ | 08:57 |
| Qiming | SDK interface is correct | 08:58 |
| Qiming | doc is broken | 08:58 |
| Qiming | server side implementation is broken | 08:58 |
| yanyanhu | you mean context is actually part of spec | 08:59 |
| yanyanhu | we never used it | 08:59 |
| Qiming | yes | 08:59 |
| yanyanhu | so didn't find this error I think... | 08:59 |
| Qiming | check heat stack sample ... | 09:00 |
| Qiming | context is part of spec | 09:00 |
| yanyanhu | yes, that's the design | 09:00 |
| Qiming | this has to be fixed | 09:00 |
| yanyanhu | I see, will propose a bug report for it | 09:01 |
| Qiming | no wonder why eldon was complaining | 09:01 |
| Qiming | thanks | 09:01 |
| yanyanhu | no problem | 09:01 |
| openstackgerrit | EldonZhao proposed openstack/senlin: Protect node form deleting when node status is creating https://review.openstack.org/300359 | 09:09 |
| openstackgerrit | Yanyan Hu proposed openstack/senlin: Fix create profile API interface https://review.openstack.org/300379 | 09:17 |
| * Qiming is hitting his head against the wall ... | 09:24 | |
| Qiming | SecondaryIPAddresses | 09:25 |
| Qiming | this is what docker_inspect tells me: | 09:26 |
| Qiming | "State": { | 09:26 |
| Qiming | "Dead": false, | 09:26 |
| Qiming | "Error": "", | 09:26 |
| Qiming | "ExitCode": 0, | 09:26 |
| Qiming | "FinishedAt": "2016-03-17T10:04:11.051160162Z", | 09:26 |
| Qiming | "OOMKilled": false, | 09:26 |
| Qiming | "Paused": false, | 09:26 |
| Qiming | "Pid": 0, | 09:26 |
| Qiming | "Restarting": false, | 09:26 |
| Qiming | "Running": false, | 09:26 |
| Qiming | "StartedAt": "2016-03-17T10:04:11.036002407Z" | 09:26 |
| Qiming | } | 09:26 |
| Qiming | how would I explain that? | 09:26 |
| yanyanhu | ... | 09:26 |
| yanyanhu | never noticed that before | 09:26 |
| Qiming | it is a mess | 09:26 |
| Qiming | this is the complete output from docker_inspect: http://paste.openstack.org/show/492707/ | 09:29 |
| Qiming | when I'm doing senlin node-show docker-1 --details | 09:29 |
| Qiming | Am I supposed to dump the whole shit? | 09:29 |
| yanyanhu | it's mainly about the process info I think | 09:30 |
| yanyanhu | most of them are useless | 09:30 |
| yanyanhu | really a mess | 09:30 |
| Qiming | because I'm not doing a lot network/storage customization yet | 09:31 |
| openstackgerrit | Merged openstack/senlin: Fix create profile API interface https://review.openstack.org/300379 | 09:52 |
| yanyanhu | hi, Qiming, has proposed fix to api-site and added you to reviewer list | 09:57 |
| yanyanhu | https://review.openstack.org/300390 | 09:57 |
| *** Qiming has quit IRC | 10:15 | |
| *** yanyanhu has quit IRC | 10:15 | |
| *** zzxwill has joined #senlin | 10:32 | |
| *** zzxwill has quit IRC | 10:34 | |
| *** cschulz has joined #senlin | 11:36 | |
| *** Qiming has joined #senlin | 12:46 | |
| *** openstackgerrit has quit IRC | 13:33 | |
| *** openstackgerrit has joined #senlin | 13:34 | |
| *** Liuqing has joined #senlin | 13:48 | |
| *** zzxwill has joined #senlin | 13:50 | |
| Liuqing | HELLO, Qiming, now the obj_id of event-list is not cluster or node short_id, so how could we filter events of cluster/node, filter by obj_name? | 13:50 |
| *** openstack has quit IRC | 13:58 | |
| *** openstack has joined #senlin | 14:02 | |
| ddeja | Hi, is it possible to use Senlin for VM creation when using Nova Network? | 14:07 |
| openstackgerrit | Yanyan Hu proposed openstack/python-senlinclient: Add link to API doc in client module https://review.openstack.org/300315 | 14:11 |
| openstackgerrit | Liuqing Jing proposed openstack/senlin-dashboard: Remove useless variable assignment https://review.openstack.org/300506 | 14:12 |
| openstackgerrit | Liuqing Jing proposed openstack/senlin-dashboard: Remove policies which had attached to cluster while attaching policy https://review.openstack.org/300512 | 14:20 |
| *** Liuqing has quit IRC | 14:47 | |
| openstackgerrit | Liuqing Jing proposed openstack/senlin-dashboard: Add cluster policies in cluster detail page https://review.openstack.org/300536 | 14:47 |
| *** Liuqing has joined #senlin | 14:48 | |
| *** Liuqing has quit IRC | 15:07 | |
| *** Liuqing has joined #senlin | 15:09 | |
| *** zzxwill has quit IRC | 15:09 | |
| *** Liuqing has quit IRC | 15:12 | |
| *** zzxwill has joined #senlin | 15:37 | |
| *** zzxwill has quit IRC | 15:44 | |
| *** zzxwill has joined #senlin | 15:54 | |
| *** zzxwill has quit IRC | 16:04 | |
| *** Qiming has quit IRC | 16:26 | |
| *** openstackgerrit has quit IRC | 16:31 | |
| *** openstackgerrit has joined #senlin | 16:46 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!