| *** zzxwill has joined #senlin | 00:40 | |
| *** shu-mutou-AFK is now known as shu-mutou | 00:42 | |
| *** zzxwill has quit IRC | 00:50 | |
| *** zzxwill has joined #senlin | 00:55 | |
| *** Qiming has quit IRC | 01:20 | |
| *** openstackgerrit has quit IRC | 01:47 | |
| *** openstackgerrit has joined #senlin | 01:48 | |
| *** Yanyanhu has joined #senlin | 01:59 | |
| *** Qiming has joined #senlin | 02:07 | |
| Qiming | Yanyanhu, lixinhui | 02:08 |
|---|---|---|
| Yanyanhu | hi | 02:08 |
| Qiming | I believe the patches we need from openstacksdk are all in | 02:08 |
| Qiming | the last one is the pool_id thing | 02:09 |
| Qiming | it is being verified | 02:09 |
| Yanyanhu | ok | 02:09 |
| Yanyanhu | cool | 02:09 |
| Qiming | had to send terry an email to get that unblocked | 02:09 |
| Yanyanhu | we can start to move to 0.8.1 | 02:09 |
| Yanyanhu | great | 02:09 |
| Qiming | yes | 02:09 |
| Yanyanhu | will start to work on driver rework today | 02:09 |
| Qiming | please help verify if there are still pending patches we need | 02:09 |
| Yanyanhu | ok | 02:10 |
| Qiming | https://review.openstack.org/#/q/project:openstack/python-openstacksdk+status:open | 02:10 |
| Yanyanhu | ah, I saw terry has removed that block | 02:11 |
| Qiming | ya | 02:11 |
| Yanyanhu | cool, looks everything is ready now :) | 02:11 |
| lixinhui | Cool!!! | 02:13 |
| *** yuanying_ has joined #senlin | 02:15 | |
| Qiming | lixinhui, please check this one: https://review.openstack.org/286388 | 02:17 |
| Qiming | I'm expecting we are a little bit stricter at parameter passing | 02:17 |
| Qiming | if we are passing keyword arguments, we explicitly spell them out if possible | 02:18 |
| Qiming | passing additional **kwargs would make the interface checking very difficult if possible at all | 02:18 |
| Qiming | lixinhui, about this: https://review.openstack.org/#/c/286575/ | 02:19 |
| Qiming | I'd suggest we fix it in health manager | 02:20 |
| Qiming | maybe it is caused by the context usage | 02:20 |
| Qiming | we should not reuse the context created in the __init__() method | 02:20 |
| Qiming | so ... a workaround is to create an admin context each time calling db apis | 02:21 |
| Qiming | session merge is very dangerous .. | 02:21 |
| *** yuanying has quit IRC | 02:23 | |
| *** zzxwill has quit IRC | 02:25 | |
| *** zzxwill has joined #senlin | 02:26 | |
| openstackgerrit | Merged openstack/senlin: Use assert(Not)Equal/Less/Greater https://review.openstack.org/286477 | 02:27 |
| openstackgerrit | Merged openstack/senlin: Remove WritableLogger from wsgi https://review.openstack.org/284044 | 02:27 |
| openstackgerrit | Merged openstack/senlin: Make map schema resolve json serialized map value https://review.openstack.org/286447 | 02:28 |
| lixinhui | Thanks for these guidance, Qiming. I will act accordingly | 02:28 |
| openstackgerrit | Merged openstack/senlin: Add functional test for cluster update https://review.openstack.org/286987 | 02:29 |
| Yanyanhu | hi, Qiming, could you plz help to workflow +1 this patch. I will clone the lastest sdk code and start some tests. Thanks | 02:29 |
| Qiming | which one? | 02:30 |
| Yanyanhu | the path_arg one | 02:30 |
| Yanyanhu | looks like it needs w+1 again | 02:30 |
| Yanyanhu | https://review.openstack.org/#/c/187934/ | 02:30 |
| Qiming | checking | 02:30 |
| Yanyanhu | sorry, omit the link | 02:30 |
| openstackgerrit | Merged openstack/senlin: Revise the logic of cluster/profile metadata update https://review.openstack.org/286998 | 02:30 |
| openstackgerrit | Merged openstack/senlin: Add functional test for cluster update with new profile https://review.openstack.org/287091 | 02:30 |
| Qiming | okay, done | 02:31 |
| Yanyanhu | thanks :) | 02:31 |
| Qiming | didn't find the patch being checked on gate | 02:31 |
| Yanyanhu | yes | 02:31 |
| Qiming | hint | 02:31 |
| Qiming | to check if your patch is queued at gate, go: http://zuul.openstack.org/ | 02:32 |
| Qiming | you can filter the pipeline there by the project you want to focus on | 02:32 |
| Yanyanhu | yes, this page is cool :) | 02:32 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Revise DRS policy to be a generic affinity policy https://review.openstack.org/286461 | 02:46 |
| *** zzxwill has quit IRC | 03:08 | |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Support to affinity policy https://review.openstack.org/287570 | 03:09 |
| *** zzxwill has joined #senlin | 03:13 | |
| *** yuanying has joined #senlin | 03:21 | |
| *** elynn has joined #senlin | 03:23 | |
| *** yuanying_ has quit IRC | 03:26 | |
| lixinhui | Qiming, bug related to 286575 is not caused by context. I have revised HealthManger arguments besides params for right mapping through notify method. | 03:27 |
| openstackgerrit | Merged openstack/python-senlinclient: Sort resource show columns https://review.openstack.org/285939 | 03:28 |
| Qiming | lixinhui, it doesn't make any sense to do a session join | 03:28 |
| Qiming | if that is required operation, we will need to do that everywhere | 03:28 |
| Qiming | 286575 is about context usage for sure | 03:29 |
| Qiming | session is a property persisted into a request context | 03:29 |
| lixinhui | I think we can keep self.ctx in ini as you suggestted to patch2 of 286388 | 03:30 |
| openstackgerrit | Merged openstack/python-senlinclient: Fix node update receive a assertion error https://review.openstack.org/286391 | 03:31 |
| lixinhui | since it is not the root cause of the bug 286575 | 03:31 |
| openstack | bug 286575 in linux (Ubuntu) "usb-automount not working - workaround "modprobe -r ehci_hcd"" [Undecided,Fix released] https://launchpad.net/bugs/286575 | 03:31 |
| Qiming | lixinhui, I suspect that is the root cause | 03:32 |
| Qiming | we can roll back to initialize a context before making db api calls | 03:32 |
| lixinhui | since DB write/delete is using request context | 03:32 |
| lixinhui | from notify | 03:32 |
| lixinhui | that is irrelavant to admin_context | 03:33 |
| Qiming | I mean we don't use that context | 03:33 |
| lixinhui | I am sure we need to use request context instead of admin context | 03:33 |
| lixinhui | remember 'tranlator' atrribute mssing problem? | 03:34 |
| lixinhui | that is caused by wrong context | 03:34 |
| Qiming | that is a different problem | 03:34 |
| lixinhui | that is the reason why I commit into 266388 | 03:34 |
| lixinhui | s/266388/286388 | 03:34 |
| Qiming | the reason of that problem is that you are passing a dict instead of a context instance to db call | 03:34 |
| lixinhui | no | 03:35 |
| lixinhui | I tried admin_contxt firstly | 03:35 |
| Qiming | ok ... | 03:35 |
| Qiming | explain to me the magic of an admin context then? | 03:35 |
| lixinhui | http://paste.openstack.org/ | 03:36 |
| lixinhui | that is the content difference between the two | 03:36 |
| Qiming | where? | 03:37 |
| Qiming | the url above is incomplete | 03:37 |
| lixinhui | wait, office network is very poor these days... | 03:38 |
| Qiming | maybe we should do this step by step | 03:39 |
| Qiming | sometimes we are talking about things on different sources | 03:39 |
| lixinhui | http://paste.openstack.org/show/489080/ | 03:39 |
| lixinhui | yes, please do not roll back | 03:40 |
| Qiming | let's focus on the first patch, the parameter passing one | 03:40 |
| Qiming | the second one is all about garbage | 03:41 |
| Qiming | we are not using that context to talk to any other services | 03:41 |
| Qiming | we are only using that to talk to database | 03:41 |
| Qiming | when talking to database, the only thing that matters is the context.session property: http://git.openstack.org/cgit/openstack/senlin/tree/senlin/common/context.py#n77 | 03:42 |
| lixinhui | it is tightly related to context | 03:42 |
| lixinhui | to fix the translate missing problem | 03:42 |
| Qiming | that 'translate' problem is not caused by this, believe me | 03:43 |
| lixinhui | I change the registry method arguments from admin_contxt to request context | 03:43 |
| lixinhui | then it never comes again | 03:43 |
| Qiming | we are talking about things on different code base | 03:43 |
| lixinhui | please calrify | 03:43 |
| Qiming | it never come back again because you are using a real RequestContext instance | 03:44 |
| lixinhui | what is it | 03:44 |
| lixinhui | ? | 03:44 |
| Qiming | the previous bug was caused by this: http://git.openstack.org/cgit/openstack/senlin/tree/senlin/common/context.py#n112 | 03:44 |
| Qiming | you were using the return value from get_service_context() | 03:44 |
| Qiming | which is only a dict | 03:44 |
| Qiming | it is meant to be used for the initialization of a RequestContext instance | 03:45 |
| Qiming | another evidence | 03:45 |
| Qiming | when you use get_admin_context | 03:45 |
| lixinhui | when http://git.openstack.org/cgit/openstack/senlin/tree/senlin/common/context.py#n112 is commit in | 03:45 |
| lixinhui | again | 03:45 |
| lixinhui | and agin | 03:45 |
| Qiming | you are using a RequestContext instance with nothing initalized, just the is_admin field for convenience | 03:45 |
| lixinhui | I have changed it to admin_ctx | 03:46 |
| lixinhui | for all my tests | 03:46 |
| lixinhui | from the very beginning | 03:46 |
| Qiming | I don't know on which code base you are doing the test | 03:46 |
| Qiming | obviously you are not working on the master branch | 03:46 |
| lixinhui | let me check now | 03:46 |
| Qiming | this one: https://review.openstack.org/#/c/286388/ | 03:47 |
| *** zzxwill has quit IRC | 03:47 | |
| Qiming | let's focus on this one first | 03:47 |
| Qiming | this one is not merged | 03:47 |
| Qiming | are you testing against it? | 03:47 |
| lixinhui | i am using the new code | 03:47 |
| lixinhui | I think we are talking about 286388 | 03:48 |
| Qiming | with or without that patch: https://review.openstack.org/#/c/286388/ | 03:48 |
| lixinhui | the problem of transltor missing will jump out | 03:48 |
| Qiming | but you are talking about this one: https://review.openstack.org/#/c/286575/ | 03:48 |
| lixinhui | no | 03:48 |
| lixinhui | you are the one trying to relsove bug mentioned by 286575 in health manager | 03:49 |
| Qiming | I have no objections to https://review.openstack.org/#/c/286388/ | 03:49 |
| Qiming | just need to get it fixed and merged | 03:49 |
| Qiming | then we can proceed to talk about whether 286575 makes sense | 03:49 |
| lixinhui | okay | 03:50 |
| Qiming | with 286388 fixed and merged | 03:50 |
| Qiming | we continue the discussion | 03:50 |
| Qiming | whether we should use the ctx passed to register_cluster method | 03:51 |
| Qiming | if we are using the 'ctx' method, do we still need 286575 | 03:51 |
| lixinhui | we need | 03:51 |
| Qiming | 286575 is not based on 286388, that was where I was confused | 03:51 |
| Qiming | so you are talking about some problems you want to solve, based on a patch that is not yet in | 03:52 |
| lixinhui | no | 03:52 |
| Qiming | and the two patches don't have dependencies among them | 03:52 |
| lixinhui | 286388 is the first patch I submut to resolve DB calling problem | 03:52 |
| Qiming | yes | 03:52 |
| lixinhui | after that, I met new problem | 03:53 |
| Qiming | then please focus on 286388 | 03:53 |
| Qiming | without 286388, how could we review 286575? | 03:53 |
| Qiming | to reproduce the problem with 286575, we need 286388 right? | 03:53 |
| lixinhui | yes | 03:54 |
| Qiming | or is 286575 independent? | 03:54 |
| Qiming | then you are talking about some problems not based on the 'current' code at master, you are talking about your local copy | 03:54 |
| Qiming | that would confuse any reviewers | 03:54 |
| lixinhui | not really | 03:54 |
| lixinhui | if that is DB problem | 03:54 |
| *** elynn has quit IRC | 03:55 | |
| lixinhui | even no 286388 | 03:55 |
| lixinhui | just non one tried that | 03:55 |
| Qiming | you fix in 286575 makes totally no sense | 03:55 |
| lixinhui | since without 286388, DBAPi even can not be called correctly | 03:55 |
| Qiming | please get back to 286388 | 03:55 |
| Qiming | we are doing this step by step | 03:55 |
| Qiming | and 286388 is not merged | 03:56 |
| Qiming | if 286575 needs 286388, you are supposed to add a dependency between the two patches | 03:56 |
| Qiming | or else we may accidentally get 286575 merged before 286388 | 03:56 |
| lixinhui | why not | 03:57 |
| lixinhui | since my fix is irrelavant to health manager | 03:57 |
| lixinhui | and as I said | 03:57 |
| lixinhui | The problem is there | 03:57 |
| lixinhui | Just easy or not to find | 03:57 |
| lixinhui | it | 03:57 |
| lixinhui | The thing in my mind is clear | 03:58 |
| Qiming | that patch basically means that all db sessions should do a merge() call | 03:58 |
| Qiming | which is insane | 03:58 |
| lixinhui | till you try to reolsve the bug of 286575 | 03:58 |
| lixinhui | by health manager | 03:58 |
| lixinhui | not really | 03:58 |
| lixinhui | just if the object needed is not in current session | 03:59 |
| Qiming | 286575 was caused by the usage of sessions | 03:59 |
| lixinhui | we can add them back | 03:59 |
| Qiming | and sessions are property of context | 03:59 |
| Qiming | that means an incorrect usage of contexts | 03:59 |
| lixinhui | the verification is done by context match or not | 03:59 |
| lixinhui | if not the ruight context, the | 03:59 |
| lixinhui | merge will fail | 03:59 |
| Qiming | how could an object falls out of the current session? | 03:59 |
| lixinhui | just timeout | 03:59 |
| lixinhui | some reason | 03:59 |
| Qiming | if an object can fall out of current session, do we need to add guard to all other DB api calls> | 04:00 |
| Qiming | ? | 04:00 |
| lixinhui | i think other methods need this need | 04:00 |
| lixinhui | to ooperate on the oject by pair operations as attach and deatch | 04:01 |
| lixinhui | since the two operations can be casued with long interval | 04:01 |
| Qiming | it is database | 04:01 |
| lixinhui | s/caused/called | 04:01 |
| lixinhui | but session just memory object' | 04:02 |
| lixinhui | of transaction | 04:02 |
| Qiming | we create a new session each time | 04:02 |
| Qiming | each session is an indenpendent call to the DB layer | 04:02 |
| Qiming | session A created an object | 04:02 |
| Qiming | session B can delete it | 04:02 |
| Qiming | we can never guarantee that an object is created and deleted by the same session | 04:03 |
| lixinhui | we do not | 04:03 |
| lixinhui | we get session from context every time | 04:03 |
| Qiming | which context? | 04:03 |
| lixinhui | just we can not avoid the object is not in the memory all the time | 04:04 |
| lixinhui | it can merge back if context right | 04:04 |
| Qiming | which context? | 04:04 |
| lixinhui | request context | 04:04 |
| Qiming | where is it initialized? | 04:04 |
| lixinhui | if not right, the merge act will fail | 04:04 |
| lixinhui | you can find the root | 04:05 |
| lixinhui | code location | 04:05 |
| lixinhui | I just can try this by practice | 04:05 |
| lixinhui | for lack of experence | 04:05 |
| Qiming | I know the root | 04:05 |
| lixinhui | What I discussed here is the resilt I got from the real system | 04:05 |
| Qiming | so I'm suggesting you to fix it in a different way, step by step | 04:05 |
| lixinhui | instead of my deriation | 04:05 |
| lixinhui | derivation | 04:06 |
| lixinhui | or assumtion | 04:06 |
| lixinhui | assumption | 04:06 |
| lixinhui | how about this | 04:06 |
| lixinhui | we leave the problem open | 04:06 |
| Yanyanhu | hi, guys, did you meet the following error when creating nova server profile using the built-in example spec? | 04:07 |
| Yanyanhu | The specified file is not a valid YAML file: mapping values are not allowed here | 04:07 |
| Yanyanhu | in "senlin/profiles/os/nova/server.py", line 29, column 34 | 04:07 |
| lixinhui | but merge into 286388 firstly | 04:07 |
| Qiming | could you please focus on 286388 | 04:07 |
| lixinhui | i want | 04:07 |
| Qiming | then I'll propose a fix to the context/session problem and see if it works | 04:08 |
| lixinhui | it is not me to induce this distraction | 04:08 |
| lixinhui | what context seesion patch | 04:08 |
| Qiming | there is no distraction | 04:08 |
| lixinhui | could you let me know what you want to do | 04:09 |
| Qiming | let's get 286388 in first | 04:09 |
| lixinhui | or you can get in your patch direcly | 04:09 |
| lixinhui | I can abandon | 04:09 |
| lixinhui | 286388 | 04:09 |
| Qiming | not overriding the 286388 one | 04:09 |
| *** zzxwill has joined #senlin | 04:09 | |
| Qiming | that one is almost correct | 04:09 |
| lixinhui | and will test them all in next week | 04:09 |
| lixinhui | okay | 04:10 |
| Qiming | saw my comments on 286388? | 04:10 |
| lixinhui | saw it | 04:10 |
| lixinhui | and tune it a little | 04:10 |
| Qiming | okay, let's get it fixed and merge it | 04:10 |
| lixinhui | I have revised HealthManger arguments besides params for right mapping through notify method. | 04:11 |
| Qiming | then we are talking about the same version of code | 04:11 |
| Qiming | "I have revised HealthManger arguments besides params for right mapping through notify method." | 04:11 |
| Qiming | checked in already? | 04:11 |
| lixinhui | Will do it right now. | 04:11 |
| Qiming | if it is not checked in, I have no idea what changes you have made | 04:11 |
| Qiming | and ... we will still be talking about different things | 04:12 |
| lixinhui | ... I see ... | 04:12 |
| Qiming | you are running too fast ahead of everyone else | 04:12 |
| Qiming | while I'm playing a guard's role regarding the so called 'code quality' | 04:13 |
| lixinhui | I am trying to cooeprate with you to get everything qualified | 04:14 |
| Qiming | yes, very appreciated | 04:14 |
| Qiming | the problem is something like this | 04:14 |
| Qiming | I'm still on the 1st floor | 04:14 |
| Qiming | you are already on the 2nd, maybe the 3rd floort | 04:15 |
| lixinhui | try my best to reach the bar | 04:15 |
| Qiming | and you have found some problem in some room at the 2nd floor | 04:15 |
| Qiming | which I have no idea yet | 04:15 |
| Qiming | I'm focusing on the things we need to do on the 1st floor | 04:15 |
| Qiming | after the work is done at this level, we may choose a different way upstairs, so we can avoid getting trapped to the 'problematic room' on the 2nd floor | 04:16 |
| lixinhui | I see and will do that | 04:17 |
| Qiming | sorry for not being able to catch up with every single thread | 04:17 |
| Qiming | everyone got a good reason to hate the 'blocker', well... I am one | 04:18 |
| lixinhui | no, you are really good leader and tutor | 04:18 |
| lixinhui | Just missing the days we sit togetehr and work together | 04:19 |
| Qiming | so, please help, get back to 286388 | 04:19 |
| Qiming | we work out a solution based on that commit | 04:20 |
| Qiming | okay? | 04:20 |
| lixinhui | okay | 04:20 |
| Qiming | great, thansk | 04:20 |
| lixinhui | :) | 04:20 |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Fix Health Manager Problems https://review.openstack.org/286388 | 04:21 |
| *** elynn has joined #senlin | 04:22 | |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Fix Health Manager Problems https://review.openstack.org/286388 | 04:22 |
| *** elynn has quit IRC | 04:26 | |
| *** elynn has joined #senlin | 04:26 | |
| *** zzxwill has quit IRC | 04:36 | |
| *** zzxwill has joined #senlin | 04:52 | |
| *** zzxwill has quit IRC | 04:53 | |
| *** zzxwill has joined #senlin | 04:55 | |
| Qiming | lixinhui, there? | 05:28 |
| openstackgerrit | Merged openstack/senlin: Add healthmonitor support for lb_policy https://review.openstack.org/285172 | 05:38 |
| *** zzxwill has quit IRC | 05:44 | |
| openstackgerrit | Merged openstack/senlin: Revise action dependency check https://review.openstack.org/287538 | 05:47 |
| lixinhui | No. Qiming, I am not here | 05:50 |
| Qiming | okay, keep waiting | 05:52 |
| lixinhui | just Joking | 05:52 |
| lixinhui | I saw your comments | 05:52 |
| lixinhui | in your example, in DB registry, you hope to keep 'params':{ 'params': {'max_interval': 50} }? | 05:54 |
| lixinhui | in a more complete dict: (Pdb) registry.__dict__ | 05:55 |
| lixinhui | {'_sa_instance_state': <sqlalchemy.orm.state.InstanceState object at 0x7fb9db703510>, u'check_type': u'NODE_STATUS_POLLING', u'engine_id': '2c21af16-460c-447c-86c3-0f7b3e6dc5ca', 'interval': 60, 'params':{ 'params': {'max_interval': 50} }, 'cluster_id': u'1c51a727-0151-47a9-8fe8-fe864834d1cc'} | 05:55 |
| Qiming | in db, it will become params: {'some-key': 'some-value'} | 05:55 |
| lixinhui | then please use **params | 05:56 |
| Qiming | no | 05:56 |
| Qiming | in line 171, the internals of the params key is extracted | 05:56 |
| Qiming | using **params is polluting the call | 05:56 |
| lixinhui | please see interface of notify... | 05:57 |
| lixinhui | and register | 05:57 |
| Qiming | when determining the function interface, we should forget that params is a dict | 05:57 |
| Qiming | just pass it as an integer | 05:57 |
| Qiming | if it is an integer, **params would be wrong | 05:57 |
| Qiming | and **params leaves no room for future extension | 05:57 |
| lixinhui | how that can be an interger | 05:57 |
| lixinhui | it must be key: value | 05:57 |
| Qiming | sigh ... | 05:58 |
| Qiming | I spend one 30 minutes to find a reference to you but I cannot find the article | 05:58 |
| Qiming | I know you will have questions like this | 05:58 |
| lixinhui | {'_sa_instance_state': <sqlalchemy.orm.state.InstanceState object at 0x7fb9db703510>, u'check_type': u'NODE_STATUS_POLLING', u'engine_id': '2c21af16-460c-447c-86c3-0f7b3e6dc5ca', 'interval': 60, 'params':{ 'params': {} }, 'cluster_id': u'1c51a727-0151-47a9-8fe8-fe864834d1cc' | 05:58 |
| Qiming | I cannot convince you on anything | 05:58 |
| Qiming | I have to refer to some standards | 05:59 |
| lixinhui | that is the pdb result I got with params=params as your suggested | 05:59 |
| Qiming | but I cannot find the standard right now | 05:59 |
| Qiming | that is because something is wrong in the call path somewhere | 05:59 |
| Qiming | my last comment to the patch is this: it works doesn't mean the code is right | 06:00 |
| lixinhui | too abstract to understand | 06:00 |
| lixinhui | I do not like such kind of comment | 06:00 |
| Qiming | as a convention, when you are passing a parameter, you should forget that parameter is a dict | 06:00 |
| lixinhui | I only like things work well | 06:00 |
| lixinhui | if the design is good, that is best | 06:01 |
| lixinhui | if not, | 06:01 |
| lixinhui | work well is the most important thing | 06:01 |
| Qiming | sorry | 06:01 |
| lixinhui | in your example | 06:01 |
| Qiming | cannot waste more time on this | 06:01 |
| Qiming | I'm really sorry | 06:01 |
| lixinhui | me eitehr | 06:01 |
| Qiming | this kind of code is not acceptable | 06:01 |
| Qiming | I have been telling you the simple truth | 06:02 |
| Qiming | when passing a parameter, forget that parameter is a dict | 06:02 |
| lixinhui | then you should define the argment as args | 06:04 |
| lixinhui | instead of kwargs | 06:04 |
| Qiming | we had to make it kwargs due to the oslo_messing call interface | 06:05 |
| Qiming | no matter how many data you want to pass through the 'params' parameter, they are all an integral part of a single parameter | 06:05 |
| Qiming | side by side, we may need to add other parameters in future | 06:05 |
| Qiming | if we use **params here today | 06:06 |
| Qiming | all those future parameters we be swallowed in to this 'params' | 06:06 |
| Qiming | that is totally not acceptable | 06:06 |
| Qiming | i really don't care it works or not, if the code looks like this | 06:06 |
| Qiming | especially when we already see problems in it | 06:07 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/senlin-dashboard: Imported Translations from Zanata https://review.openstack.org/286397 | 06:18 |
| Qiming | hi, guys, please input your ideas reagarding the design summit sessions | 06:32 |
| Yanyanhu | oh, right, need finish it by friday | 06:33 |
| Qiming | yes, then get back to ttx | 06:33 |
| Yanyanhu | ok | 06:33 |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Fix Health Manager Problems https://review.openstack.org/286388 | 06:44 |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Fix Health Manager Problems https://review.openstack.org/286388 | 06:55 |
| Qiming | dixiaoli, elynn | 06:56 |
| Qiming | are you aware of any changes we need before cutting a release of python-senlinclient today or tomorrow? | 06:57 |
| elynn | No from me, I will check with dixiaoli after she come back from meeting. | 06:59 |
| Qiming | thanks, elynn | 07:01 |
| Qiming | another job is about sdk 0.8.1 | 07:01 |
| Qiming | previously we have been locking to sdk 0.7.4 | 07:01 |
| Yanyanhu | hi, Qiming, just tested nova server, heat stack profile and also lb policy with health_monitor enabled, they all works well! | 07:02 |
| Yanyanhu | I'm surprised | 07:02 |
| Yanyanhu | just a little change need to be made in neutron driver | 07:02 |
| Yanyanhu | sdk version is dev0.8.1 | 07:02 |
| Qiming | obviously we will need some patches which get merged between 0.7.4 and 0.8.1 | 07:02 |
| Yanyanhu | nova server and heat stack update were also tested. | 07:02 |
| Qiming | but I'm not sure those changes are affecting the senlinclient package in any way | 07:03 |
| Yanyanhu | yes, I think the influence should be different in client side | 07:03 |
| Yanyanhu | BTW, I didn't test placement policy related part. Hi, xinhui, if possible, could you please help to test this part, thanks :) | 07:04 |
| Qiming | we'd better get prepared since I'm pushing brian to release a 0.8.1 today, then we cut senlinclient release and test senlin server (Yanyan is already doing that) | 07:04 |
| Yanyanhu | about the availability_zone param I think | 07:04 |
| Qiming | Yanyanhu, you mean zone_placement policy? | 07:04 |
| Yanyanhu | yes | 07:04 |
| Yanyanhu | since I didn't have related configuration in my local env | 07:05 |
| Yanyanhu | multi-zone and region | 07:05 |
| Yanyanhu | so didn't test that | 07:05 |
| Qiming | I still believe the best test case is your eyes | 07:05 |
| Yanyanhu | the code looks good I think :) | 07:05 |
| Qiming | the code may pass some test cases, but it doesn't mean we have coverage for all corner cases | 07:05 |
| Yanyanhu | yep | 07:06 |
| Qiming | I'm looking at it | 07:06 |
| Yanyanhu | I think I may have some misunderstanding about sdk's property | 07:06 |
| Qiming | if the code is not referencing any particular property of an availability zone object, it should not break | 07:06 |
| Yanyanhu | I see | 07:06 |
| Qiming | Yanyanhu, I myself was a little too much concerned about those changes as well | 07:07 |
| Qiming | Yanyanhu, zone_placement policy should work | 07:08 |
| Yanyanhu | I think I need to spend sometime to read the code of resource class again | 07:08 |
| Qiming | no particulare reference to any properties | 07:08 |
| Yanyanhu | nice | 07:08 |
| Yanyanhu | I will submit that small patch for neutron driver | 07:08 |
| Yanyanhu | actually, it is for the change done by richard for pool_member create/update proxy call | 07:09 |
| Qiming | the only thing we are referring to AZs happened in 'validate_azs()' function | 07:10 |
| Qiming | there, we are referring to the 'zoneName' property | 07:11 |
| Yanyanhu | yes | 07:11 |
| Yanyanhu | let me check whether related property has been changed | 07:11 |
| Yanyanhu | it wasn't | 07:12 |
| Qiming | yep | 07:13 |
| Qiming | 44 azs = conn.compute.availability_zones() | 07:14 |
| Qiming | 45 for s in azs: | 07:14 |
| Qiming | 46 print(s['zoneName']) | 07:14 |
| Qiming | 47 print(s.zoneName) | 07:14 |
| Qiming | 48 print(s.name) | 07:14 |
| Qiming | 49 # print(s['name']) <-- this doesn't work | 07:14 |
| Qiming | the first three ways of property references all worked, the last one fails | 07:14 |
| Yanyanhu | name property is still there | 07:15 |
| Yanyanhu | weird | 07:15 |
| * Qiming is feeling very much relieved | 07:15 | |
| Qiming | if you want to reference 'name', you will do az.name | 07:16 |
| Qiming | not az['name'] | 07:16 |
| Yanyanhu | oh | 07:16 |
| Yanyanhu | but it shoud be an object of availability_zone.AvailabilityZone type | 07:17 |
| Yanyanhu | I think | 07:17 |
| Qiming | yes, <class 'openstack.compute.v2.availability_zone.AvailabilityZone'> | 07:19 |
| Qiming | http://git.openstack.org/cgit/openstack/python-openstacksdk/tree/openstack/resource.py#n359 | 07:19 |
| Yanyanhu | ah | 07:20 |
| Yanyanhu | a bug? | 07:20 |
| elynn | Hi Qiming , dixiaoli said there is no more patches from her for senlinclient. | 07:23 |
| Qiming | thanks, elynn and dixiaoli | 07:23 |
| *** zzxwill has joined #senlin | 07:23 | |
| Qiming | Yanyanhu, why is that a bug? | 07:23 |
| openstackgerrit | Yanyan Hu proposed openstack/senlin: Revise neutron driver for change in sdk 0.8.1 release https://review.openstack.org/287627 | 07:24 |
| Yanyanhu | I mean is there a bug here? | 07:24 |
| Qiming | no | 07:24 |
| Yanyanhu | not there is a bug :) | 07:24 |
| Qiming | it is designed that way | 07:24 |
| Yanyanhu | ok | 07:24 |
| Qiming | openstack.Resource can be used as a Map | 07:24 |
| Yanyanhu | yes, that's is flexible, just a little complicated in some cases | 07:24 |
| Qiming | because it implemented __getitem__, __setitem__ methods | 07:24 |
| Yanyanhu | yea | 07:24 |
| Qiming | as SDK, convenience, ease-of-use is the top concern | 07:25 |
| Qiming | if you don't like this way, there is another choice | 07:25 |
| Qiming | that is a good design, IMHO | 07:25 |
| Yanyanhu | I see | 07:26 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Fix action status update https://review.openstack.org/287630 | 07:30 |
| openstackgerrit | Merged openstack/senlin: Fix Health Manager Problems https://review.openstack.org/286388 | 07:33 |
| Qiming | jus tested senlinclient | 07:35 |
| Qiming | the output fields are based on the response json property keys, not the resource property name | 07:36 |
| *** dixiaoli has quit IRC | 07:36 | |
| Qiming | for example, the 'user' property of a cluster has been changed to 'user_id' | 07:36 |
| Yanyanhu | yes, it is I think | 07:37 |
| Yanyanhu | also notice that | 07:37 |
| Qiming | however the openstack cluster show c1 still prints 'user', not 'user_id' | 07:37 |
| Qiming | that is great | 07:37 |
| Yanyanhu | so I guess there is automatic converting between a key and an attr | 07:38 |
| Qiming | yii, that is magic | 07:39 |
| Qiming | I hate magics, :) | 07:39 |
| Yanyanhu | haha, me too | 07:39 |
| Qiming | I have to know how things work, :( | 07:39 |
| Yanyanhu | :) I will go to read the code as well | 07:41 |
| Yanyanhu | but I can sleep well without knowing the truth, haha | 07:42 |
| openstackgerrit | Yanyan Hu proposed openstack/senlin: Fix an error in test lbaas driver https://review.openstack.org/287634 | 07:43 |
| openstackgerrit | Yanyan Hu proposed openstack/senlin: Fix an error in test lbaas driver https://review.openstack.org/287634 | 07:44 |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Argument Twist to Health Policy https://review.openstack.org/287635 | 07:45 |
| elynn | Hi Yanyanhu , I just approve your patch https://review.openstack.org/#/c/287627/ , but I realize that should we wait for the openstacksdk 0.8.1 ? | 07:46 |
| Yanyanhu | elynn, yes, it is... | 07:47 |
| elynn | I thought it was released... | 07:47 |
| Qiming | oh, boy, not yet | 07:47 |
| Qiming | but it doesnt matter that much | 07:48 |
| Yanyanhu | but maybe not that bad since it just change the driver method about pool_member creating | 07:48 |
| Yanyanhu | we can revert it | 07:48 |
| Qiming | we used to develop against the master branch of openstacksdk | 07:48 |
| Yanyanhu | yes. When will 0.8.1 come | 07:48 |
| Yanyanhu | by the end of this week? | 07:48 |
| Qiming | not sure, just sent Brian an email this morning | 07:49 |
| Yanyanhu | ok | 07:49 |
| Qiming | after I saw the pool id fix merged | 07:49 |
| Qiming | that is the last patch we need from them | 07:49 |
| elynn | I remove workflow, not sure it will merge or not. | 07:50 |
| Yanyanhu | let me check the zuul page | 07:52 |
| *** dixiaoli has joined #senlin | 07:52 | |
| Yanyanhu | elynn, now it is in a very odd status :) | 07:59 |
| Yanyanhu | https://review.openstack.org/#/c/287627/ | 07:59 |
| Qiming | Yanyanhu, https://review.openstack.org/#/c/287634/2/senlin/tests/functional/drivers/openstack/lbaas.py | 08:08 |
| Qiming | teach me why we need a 'vm' as parameter when you have time | 08:08 |
| Yanyanhu | faint, it should be hm | 08:08 |
| Yanyanhu | my fault | 08:09 |
| * Qiming faints too | 08:09 | |
| Yanyanhu | stupid mistake | 08:09 |
| openstackgerrit | Yanyan Hu proposed openstack/senlin: Fix an error in test lbaas driver https://review.openstack.org/287634 | 08:10 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Revision to FEATURES.rst https://review.openstack.org/287642 | 08:13 |
| openstackgerrit | Qiming Teng proposed openstack/senlin: Fix action status update https://review.openstack.org/287630 | 08:16 |
| Qiming | lixinhui, not online? | 08:17 |
| lixinhui | yes, I am here | 08:18 |
| Qiming | do you still have problems/questions with ceilometer? | 08:19 |
| lixinhui | no | 08:19 |
| Qiming | okay, just noticed that lianhao lu is on ceilometer channel | 08:47 |
| Qiming | guys | 08:53 |
| Qiming | since we don't have other patches hanging regarding senlinclient | 08:54 |
| Qiming | I'm gonna generate a release now | 08:54 |
| Yanyanhu | ok | 08:54 |
| Qiming | it will be version 0.4.0 | 08:56 |
| Qiming | it will be a milestone for senlin client | 08:56 |
| Yanyanhu | nice | 08:57 |
| Yanyanhu | so this will be the final release of senlinclient in Mitaka if no exception happen? | 08:57 |
| Qiming | yes | 08:57 |
| Yanyanhu | I see | 08:57 |
| Qiming | we still have a lot patches hanging at server side: https://review.openstack.org/#/q/project:openstack/senlin+status:open | 08:58 |
| Yanyanhu | yes | 08:58 |
| Qiming | we are cutting a m3 release for server side | 09:00 |
| openstackgerrit | Merged openstack/senlin: Minor tweak setup-service tool https://review.openstack.org/283020 | 09:03 |
| Qiming | lixinhui, I'm trying to reproduce the problems you reported about deleting health_registry entries | 09:13 |
| lixinhui | Okay | 09:20 |
| Qiming | for i in {1..100}; do senlin cluster-policy-attach -p hp1 c1; senlin cluster-policy-detach -p hp1 c1; done | 09:21 |
| *** Yanyanhu has quit IRC | 09:26 | |
| lixinhui | no need that way | 09:26 |
| lixinhui | just attach then deatach | 09:26 |
| lixinhui | detach | 09:26 |
| Qiming | okay, mysql is sometimes unpredictable | 09:26 |
| Qiming | when working on the action dependency | 09:26 |
| Qiming | some problems are very sporadic, completely unpredictable | 09:27 |
| openstackgerrit | Merged openstack/senlin: Fix action status update https://review.openstack.org/287630 | 09:30 |
| lixinhui | Qiming, so we shoud claim it as "known issue" and "wont fix in senlin"? | 09:33 |
| Qiming | cannot reproduce it | 09:34 |
| lixinhui | ai | 09:34 |
| lixinhui | I can | 09:34 |
| Qiming | if it is still there, we'd better fix it, we still have time | 09:34 |
| Qiming | let me find the bug report and see if there are hints | 09:35 |
| lixinhui | How for me to send you the log | 09:37 |
| lixinhui | 09:37 | |
| lixinhui | or email | 09:37 |
| lixinhui | or some better way with Openstack infrastrature | 09:37 |
| Qiming | maybe QQ | 09:37 |
| *** openstackgerrit_ has joined #senlin | 09:54 | |
| *** openstackgerrit_ has quit IRC | 10:02 | |
| *** Qiming has quit IRC | 10:11 | |
| *** elynn has quit IRC | 10:15 | |
| openstackgerrit | Cindia-blue proposed openstack/senlin: Revise Cache of Registry of Health Manager https://review.openstack.org/286575 | 10:46 |
| *** shu-mutou is now known as shu-mutou-OFF | 11:01 | |
| *** openstackgerrit has quit IRC | 12:03 | |
| *** openstackgerrit has joined #senlin | 12:04 | |
| *** branw has joined #senlin | 12:21 | |
| *** elynn has joined #senlin | 14:06 | |
| *** elynn has quit IRC | 14:06 | |
| openstackgerrit | Merged openstack/senlin: Fix an error in test lbaas driver https://review.openstack.org/287634 | 14:24 |
| *** Qiming has joined #senlin | 14:32 | |
| *** Qiming has quit IRC | 16:02 | |
| *** zzxwill has quit IRC | 16:32 | |
| *** dhellmann has joined #senlin | 20:36 | |
| dhellmann | any cores around? I'm looking for Qiming or someone else to talk about the milestone release. | 20:37 |
| *** shu-mutou-OFF has quit IRC | 22:10 | |
| *** xuhaiwei has joined #senlin | 23:39 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!