Monday, 2016-02-29

*** Qiming_ has quit IRC00:14
openstackgerritMerged openstack/senlin-dashboard: Add detach policy in manage-policies forms  https://review.openstack.org/28578100:22
*** Qiming_ has joined #senlin01:14
*** idonotknow_ has joined #senlin01:23
openstackgerritMerged openstack/senlin: Refactor cluster detach policy operation  https://review.openstack.org/28580101:32
openstackgerritQiming Teng proposed openstack/senlin: Refactor node create operation in cluster action  https://review.openstack.org/28586201:42
*** zzxwill has joined #senlin01:48
*** Yanyanhu has joined #senlin01:56
openstackgerritMerged openstack/senlin: Refactor cluster update policy operation  https://review.openstack.org/28580201:58
openstackgerritQiming Teng proposed openstack/senlin: Refactor node update operation in cluster action  https://review.openstack.org/28586401:59
*** elynn has joined #senlin02:00
openstackgerritMerged openstack/senlin: Remove event logs in lbaas driver  https://review.openstack.org/28519402:03
*** elynn has quit IRC02:04
*** elynn has joined #senlin02:05
*** Qiming_ has quit IRC02:07
*** Qiming has joined #senlin02:08
openstackgerritMerged openstack/senlin: Enable cluster registry and health check in Health Manager  https://review.openstack.org/28229902:11
openstackgerritMerged openstack/senlin: Fix cluster status_reason setting when actions failed  https://review.openstack.org/28561902:11
openstackgerritMerged openstack/senlin: Refactor action create operation  https://review.openstack.org/28580302:15
elynnMorning02:16
openstackgerritMerged openstack/senlin: Refactor webhook trigger operation  https://review.openstack.org/28580902:17
openstackgerritMerged openstack/senlin: Refactor node create operation in cluster action  https://review.openstack.org/28586202:25
openstackgerritMerged openstack/senlin: Refactor node update operation in cluster action  https://review.openstack.org/28586402:25
openstackgerritMerged openstack/senlin: Remove unused `paramiko`  https://review.openstack.org/28369002:35
openstackgerritQiming Teng proposed openstack/senlin: Refactor node delete operation in cluster action  https://review.openstack.org/28586602:35
*** Yanyanhu has quit IRC02:44
*** Yanyanhu has joined #senlin02:45
Yanyanhuhi, guys, does senlin-engine works correctly in your env?02:46
YanyanhuI got an error when starting senlin-engine02:47
elynnHaven't try it, let me have a try.02:47
Yanyanhuwhich complains context obj doesn't have session property02:47
Yanyanhuelynn, thanks :)02:47
elynnI get this error ... AttributeError: 'dict' object has no attribute 'session'02:49
elynnI get this error AttributeError: 'dict' object has no attribute 'session'02:49
Yanyanhuyes, the same error02:49
Yanyanhulet me check the reason. Guess context is not used correctly in some cases02:49
elynnCause by this patch ? https://review.openstack.org/#/c/282299/02:54
Yanyanhuah, possible02:56
lixinhuielynn, yanyna02:56
lixinhuiyes´╝îI may commit into a patch quickly to remove the block then provide fix soon02:56
Yanyanhulixinhui, thanks a lot :)02:57
Yanyanhuin line 6702:57
Yanyanhuyou may need to convert the context from dict to context obj :)02:58
Yanyanhuor use admin_context I think02:59
lixinhuiOkay, Thanks, Yanyan. I will check and tune it son03:00
lixinhuis/son/soon03:00
openstackgerritQiming Teng proposed openstack/senlin: Refactor add nodes operation in cluster action  https://review.openstack.org/28587003:01
openstackgerritDi XiaoLi proposed openstack/python-senlinclient: Update OSC metavars  https://review.openstack.org/28587103:04
openstackgerritCindia-blue proposed openstack/senlin: Quick Fix to Remove the block when Engine Start  https://review.openstack.org/28587203:14
openstackgerritMerged openstack/senlin: Quick Fix to Remove the block when Engine Start  https://review.openstack.org/28587203:29
*** Yanyanhu has quit IRC03:33
*** Yanyanhu has joined #senlin03:33
*** zzxwill has quit IRC03:42
openstackgerritMerged openstack/senlin: Refactor node delete operation in cluster action  https://review.openstack.org/28586603:46
openstackgerritMerged openstack/senlin: Refactor add nodes operation in cluster action  https://review.openstack.org/28587003:50
*** elynn has quit IRC03:52
*** zzxwill has joined #senlin03:52
*** Qiming_ has joined #senlin04:35
*** Qiming has quit IRC04:35
*** zzxwill has quit IRC04:36
*** elynn has joined #senlin04:42
*** elynn has quit IRC04:47
*** elynn has joined #senlin04:48
*** zzxwill has joined #senlin05:08
*** yuanying has quit IRC05:24
*** yuanying_ has joined #senlin05:24
haiwei_hi all , back to office now, long time no see05:34
Qiming_welcome05:41
*** sridhar_ram has joined #senlin05:42
*** sridhar_ram has quit IRC05:42
haiwei_it seems a lot of jobs have been done in the last week05:42
openstackgerritQiming Teng proposed openstack/senlin: Refactor cluster recover action  https://review.openstack.org/28589205:44
Qiming_haiwei_, everything is okay with you?05:49
haiwei_yes, a little terribel at first, got much better these two days,05:51
haiwei_and broke out 2 days in advance05:51
openstackgerritQiming Teng proposed openstack/senlin: Add unit test for cluster check action  https://review.openstack.org/28589606:21
Qiming_glad to know you are recovering06:23
haiwei_thanks06:23
*** idonotknow__ has joined #senlin06:31
*** idonotknow_ has quit IRC06:32
*** idonotknow__ has left #senlin06:35
*** idonotknow__ has joined #senlin06:35
Qiming_hi, need some eyes on this: https://review.openstack.org/#/c/285635/06:46
Qiming_the goal is to generalizing it into a generic affinity prolicy since server_group resource is finally supported by sdk06:47
Qiming_however, the policy is still making use of some DRS specific calls which are not exposed from openstacksdk06:47
Qiming_if we cannot get such support 'simulated' in certain way, we have to drop DRS support completely06:49
Qiming_idonotknow__, please help review06:49
Qiming_thx06:49
* Qiming_ going out for starbucks06:49
*** elynn_ has joined #senlin06:53
*** elynn has quit IRC06:55
lixinhuiQiming_, I just reviewed that. Seems server_group create/get/delete are supported these days06:56
lixinhuiwhat specfic calls your patch can not today?06:57
*** Ayush_ has joined #senlin07:03
Yanyanhuhi, lixinhui, this patch works now https://review.openstack.org/28517207:24
YanyanhuI think you can apply it for healthmonitor test in lbaas v2 environemnt07:24
lixinhuiSuper, thanks, Yanyanhu :)07:25
Yanyanhuthe healthmonitor can be created successfully and it will check member using specified method. But I didn't test it using real workload since I don't have a complete test bed in my local env :)07:25
Yanyanhuno problem. If you meet any issue, plz just ping me07:25
Yanyanhuoh, BTW, the path_arg patch is still needed for pool_member created07:26
Yanyanhuit is still hung there at sdk side...07:26
lixinhuiOkay, I will keep local sdk for this with the patch and 0.7.407:27
Qiming_lixinhui, by DRS specific calls, I mean the get_hypervisor_by_id() and similar calls07:38
lixinhui...07:39
lixinhuiBran once have some patch about this07:39
Qiming_get_hypervisors() is not supported on SDK either07:39
lixinhuiI do not know if we can get it into sdk07:39
lixinhuiI mean timeline07:39
lixinhuiI can keep work on that07:39
Qiming_without those patches landed somewhere, we won't get the policy working07:40
lixinhuiI see07:43
lixinhuiJust forgot this thing07:43
lixinhuiI will check Byane's comments on that patch then return to you07:43
Qiming_ok07:54
Qiming_dixiaoli, there?07:59
dixiaoliyes07:59
Qiming_just trying the osc plugin, it works great!08:00
Qiming_got a question regarding the object show command08:00
Qiming_ openstack cluster node show 4b508:00
Qiming_+---------------+--------------------------------------+08:00
Qiming_| Field         | Value                                |08:00
Qiming_+---------------+--------------------------------------+08:00
Qiming_| status        | ACTIVE                               |08:00
Qiming_| domain        | None                                 |08:00
Qiming_| tags          | {}                                   |08:00
Qiming_| profile_id    | 22ec1344-dc3e-44b9-8713-469ca9fd9fac |08:00
Qiming_| updated_at    | None                                 |08:00
Qiming_| profile_name  | pstack                               |08:00
Qiming_| cluster_id    | cc39fc17-df6f-46eb-a6ca-8e84bca3c632 |08:00
Qiming_| user          | 5e5bf8027826429c96af157f68dc9072     |08:00
Qiming_| physical_id   | 7845e8da-81a8-4ca5-b644-0fc348ba3a86 |08:00
Qiming_| status_reason | Creation succeeded                   |08:00
Qiming_| id            | 4b578c22-fd50-481f-8947-543ea907f028 |08:00
Qiming_| index         | 1                                    |08:00
Qiming_| name          | node-cc39fc17-001                    |08:01
Qiming_| data          | {}                                   |08:01
Qiming_| created_at    | 2016-02-28T10:01:08                  |08:01
Qiming_| project       | 42d9e9663331431f97b75e25136307ff     |08:01
Qiming_| init_at       | 2016-02-28T10:01:05                  |08:01
Qiming_| role          |                                      |08:01
Qiming_| metadata      | {}                                   |08:01
Qiming_+---------------+--------------------------------------+08:01
Qiming_I'm seeing that the Field column is not ordered08:01
dixiaoliQiming_, let me have a try08:03
Qiming_okay08:03
*** idonotknow_ has joined #senlin08:12
*** lixinhui_ has joined #senlin08:13
*** idonotknow__ has quit IRC08:14
dixiaoliQiming_, yes,it is my mistake.08:15
dixiaoliUse "return zip(*sorted(six.iteritems(node)))"  in _show_node will return the sorted columns.08:16
Qiming_dixiaoli, not a mistake, just something we can improve08:16
Qiming_will that induce a lot overhead?08:16
lixinhui_Qiming, Yanyanhu08:16
dixiaoliQiming_, okay, will update it08:16
lixinhui_I got this error with context08:16
lixinhui_http://paste.openstack.org/show/488506/08:16
lixinhui_Qiming_08:17
Yanyanhuyes08:17
lixinhui_I once asked tis question before08:17
Yanyanhuneed a patch here08:17
Yanyanhujust as we discussed this morning08:17
lixinhui_what is the proper way to get context at the very beginning08:17
lixinhui_of service creation08:17
lixinhui_Okay, Yanyanhu08:18
lixinhui_admin_context?08:18
Yanyanhuhmm, I think maybe we should build a context of the owner of cluster08:18
Yanyanhufor this health check08:18
Qiming_dixiaoli, I just tried 'columns = sorted(list(six.iterkeys(cluster)))08:18
Qiming_'08:18
Yanyanhulet me think about it08:18
Qiming_dixiaoli, event this 'columns = sorted(six.iterkeys(cluster))' works for me08:19
Yanyanhuthe simplest way is using admin context08:19
lixinhui_Okay, I will try this08:20
Yanyanhuoh, maybe we should use the context of the health policy owner08:20
Yanyanhusince he/she is the one that make senlin check cluster status periodically08:20
Qiming_cannot read the pasted content08:21
Yanyanhulixinhui_, I think you can try that context :)08:21
lixinhui_:)08:21
Qiming_an exception thrown in health_manager.py ?08:22
Qiming_line 135, when registering cluster?08:22
Qiming_which version of code ?08:22
YanyanhuQiming_, not at the time registering08:22
Yanyanhuthe context used in health_manager.py is actually a dict08:23
Yanyanhunot context obj08:23
Qiming_the contexts says this: File "/opt/stack/senlin/senlin/engine/health_manager.py", line 135, in register_cluster\n08:23
Yanyanhuso db api complains no session attribute in context08:23
Qiming_I'm not seeing the code line 135 in this way ...08:23
Yanyanhuoh, right, my mistake08:23
dixiaoliQiming_, good. we need sort the columns.I will submit a patch to update it.08:23
Yanyanhuit happened in that case as well08:23
Qiming_yes, you are right, it seems the context was not correctly intializied08:24
Qiming_but without looking at the source code, tihs log is not helpful at all08:25
YanyanhuQiming_, yes, that's right08:25
Yanyanhuthis should be an internal error I think08:25
YanyanhuI mean not a user-faced error, just a bug of service08:26
Qiming_my line 135 of health_manager is this:08:27
Qiming_135     def unregister_cluster(self, cluster_id):08:27
Yanyanhume too08:28
Qiming_okay, this line is wrong08:28
Qiming_http://git.openstack.org/cgit/openstack/senlin/tree/senlin/engine/health_manager.py#n12808:29
Qiming_it is abusing the return value from get_service_context()08:29
Qiming_get_service_context() returns a dict, not a RequestContext object08:29
Yanyanhuyes08:29
lixinhui_I have change it by enlin_context.RequestContext.from_dict(params)08:30
lixinhui_Them DBError 'RequestContext' object has no attribute 'translate'08:31
lixinhui_Then DBError 'RequestContext' object has no attribute 'translate'08:31
lixinhui_I am trying with ctx = senlin_context.get_admin_context()08:32
Yanyanhuadmin context is ok I think08:32
Qiming_http://git.openstack.org/cgit/openstack/senlin/tree/senlin/engine/health_manager.py#n2808:33
Qiming_that is something we need to clean as well08:33
Qiming_we are importing context twice08:33
Yanyanhusight08:33
Yanyanhusigh08:33
lixinhui_Clean it08:35
lixinhui_cleaned it08:35
*** lixinhui_ has quit IRC08:37
openstackgerritDi XiaoLi proposed openstack/python-senlinclient: Sort resource show columns  https://review.openstack.org/28593908:44
Qiming_lixinhui, reformated your trace log08:54
Qiming_it looks like your environment setting has triggered some language specific unicode handling logic08:54
Qiming_I have never met that before08:54
lixinhuicool08:57
lixinhuiToday, who vmware office is super slow08:57
Qiming_it could be cause by this ...08:58
lixinhuivmware network condition is very slow08:58
Qiming_http://git.openstack.org/cgit/openstack/senlin/tree/senlin/db/sqlalchemy/api.py#n123808:58
Qiming_don't understand why are we not using session directly there ...08:59
Yanyanhulixinhui, our wifi is broken for the whole day :)08:59
Yanyanhunot much better09:00
lixinhuiI think admin_conext work well09:00
lixinhuiwill check it again09:00
lixinhuifeel better, Yanyanhu09:00
lixinhuiQiming_, I will rebuilt the env and find why09:01
openstackgerritMerged openstack/senlin: Documentation for deletion policy  https://review.openstack.org/28355809:04
Qiming_lixinhui, there are something not correct in the db api09:07
lixinhuioh?09:09
lixinhuiafter change to admin_context, start, attach and detach works now09:10
lixinhuibut need full tests by this time09:11
Qiming_it is not about which context09:11
*** Yanyanhu has quit IRC09:12
lixinhuigo ahead09:15
*** zzxwill has quit IRC09:18
openstackgerritMerged openstack/python-senlinclient: Update OSC metavars  https://review.openstack.org/28587109:30
openstackgerritMerged openstack/senlin: Revise vSphere DRS policy  https://review.openstack.org/28563509:31
openstackgerritQiming Teng proposed openstack/senlin: Refactor cluster recover action  https://review.openstack.org/28589209:39
*** elynn_ has quit IRC09:49
openstackgerritQiming Teng proposed openstack/senlin: Refactor cluster recover action  https://review.openstack.org/28589209:50
Qiming_this is insane09:51
Qiming_cannot trigger ci at all: https://review.openstack.org/#/c/285892/09:51
Qiming_the code was submitted 4 hours ago09:51
*** Qiming_ has quit IRC10:20
*** zzxwill has joined #senlin10:46
*** idonotknow_ has quit IRC10:59
*** Qiming has joined #senlin11:16
*** Ayush_ has quit IRC11:24
*** zzxwill has quit IRC11:27
*** elynn_ has joined #senlin11:55
*** elynn_ has quit IRC12:00
-openstackstatus- NOTICE: Infra currently has a long backlog. Please be patient and where possible avoid rechecks while it catches up.12:07
lixinhuiQiming, I found the inccorrect with db13:27
lixinhuiI am preparing the patch13:27
lixinhuiactually that is caused by illegal format13:28
openstackgerritQiming Teng proposed openstack/senlin: Optimize the registry claim function  https://review.openstack.org/28604313:35
Qimingwhat format?13:35
Qimingyou are always talking about something without giving me any context, !!!13:35
*** zzxwill has joined #senlin13:41
lixinhuiQiming, about this problem http://paste.openstack.org/show/488557/13:48
Qimingokay, I know that is a problem13:50
Qimingyou were talking about something's format is wrong13:50
Qimingbut I have no idea what you are referring to13:51
lixinhuiSign13:51
*** elynn has joined #senlin13:51
lixinhuiJust a fake alarm13:51
lixinhuiI once think the root cause for this is wrong format by augments so the expecption is thown out13:52
lixinhuibut find it will happend since the first time to use the DB13:52
lixinhuiin health_manager13:52
Qimingokay13:54
Qimingmy belief is that I don't trust any test cases13:55
QimingI trust the code itself13:55
openstackgerritQiming Teng proposed openstack/senlin: Fix various problems in health manager  https://review.openstack.org/28605814:00
Qiminglixinhui, ^14:01
Qimingpartial fix14:01
Qimingstill reading the code14:01
elynnHi Qiming14:02
Qimingyes, elynn14:02
elynnabout https://review.openstack.org/#/c/28347214:03
elynnI'm thinking about fixing it in senlin side.14:03
Qimingokay?14:03
Qimingnot quite sure I understand your intent14:04
elynnoh, I update the bug description.14:05
Qimingthe proposal looks interesting at first sight14:05
Qimingget_file function is poison14:06
elynnyes, but can't get rid of it ...14:06
Qimingsince heat already provides a get_file function14:07
Qimingwhey don't we try get it dirtier?14:07
Qimingget_file: today accepts a filename14:07
Qimingit can be improved to accept an optional parameter, e.g. 'type'14:08
Qimingwhich defaults to 'raw'14:08
Qimingthat means the file content will be dumped into the template directly14:08
Qimingyou can have 'type' set to 'yaml'14:09
Qimingthen you will treat the file as a YAML source and dump the content there as a dict14:09
Qiming'str_json', as a new function, doesn't look very appealing, to be honest14:09
elynnHmm, Interesting point.14:10
elynnI'm not so sure I can convince community to make a change to get_file...14:11
Qimingusers won't like one more function which is not very frequently used14:11
Qimingactually, it promotes better template composibility14:11
Qimingwhat do you think14:11
Qimingyou can have snippets scattered here and there14:11
Qimingthen in a master template, you can combine them together into a working template14:12
Qimingwho would hate that? :)14:12
Qimingbefore proposing that, you can discuss with shardy and see if he has a better idea, regarding str_replace14:13
elynnQiming, yes, seems modify get_file could do that.14:13
elynnQiming, yes, I will. If propose it, I think it will only get merged in next release...14:14
elynnjust ping shardy, seems he is not available now. Still not figure out how to use str_replace.14:14
Qimingokay14:16
elynnI'm thinking about changing https://github.com/openstack/senlin/blob/master/senlin/profiles/os/heat/stack.py#L49-L53 to string as a workaround?14:16
Qimingwell it could be14:18
QimingI'm thinking of a different format:14:18
Qimingfor example14:18
Qimingheat_template_1: &TEMPLATE114:18
Qiming  heat_template_version: 2013-05-2314:18
Qiming  resources:14:18
Qiming    ran:14:18
Qiming      type: OS::Heat::RandomString14:18
Qimingtype: os.heat.stack14:18
Qimingversion: 1.014:18
Qimingproperties:14:18
*** Qiming has quit IRC14:19
*** Qiming has joined #senlin14:19
Qimingthis is effectly laying out the heat raw template side by side with senlin profile content14:20
elynnwelcome back Qiming14:20
elynnwhat do you mean &TEMPLATE1 ?14:21
Qimingit is builtin cross-reference supported by YAML14:21
elynnlike alias?14:22
Qiming&SOMETHING defines a location14:22
Qiming*SOMETHING references that content14:22
Qimingafter the yaml is parsed, you can reference the template content via data['properties']['template'] as if the content is there14:23
elynnSeems still need to put template content into the same yaml file?14:24
Qimingyep14:27
Qimingbut the readability is greatly improved14:27
Qimingthe philosophy of putting all data into a single file is for ease of version control14:28
Qimingwe don't separate parameters, environment, template into many files14:28
Qimingwhen one or two files changed14:28
Qimingwe don't know we are creating the same thing as before14:28
Qimingheat was getting into a big trouble because of this14:29
Qimingall things were scattered everywhere14:29
Qimingconvergence is really a headache14:29
Qiminghowever, a different philosophy is to separate things into many composable units14:30
Qimingthen you will need a good version control system to manage those pieces14:30
elynnI'm would like the later separate things philosophy :) it will make things easy to compose.14:31
elynnLet me think about it tomorrow, need to try the ways you said and find a better one.14:32
Qimingokay14:32
Qimingthe "senlin way" is way too restrictive14:32
Qimingwe can become a little bit "open" in future14:33
Qimingbut if we start very "open" today, we won't get a chance to become "restrictive"14:33
*** lixinhui_ has joined #senlin14:34
elynnyou always have so many solutions in your mind :) it's enlighten to chat with you.14:35
Qiming...14:35
elynn;)14:36
*** elynn has quit IRC14:38
Qiminggate is terrible today14:38
*** lixinhui_ has quit IRC14:41
*** lixinhui_ has joined #senlin14:45
openstackgerritMerged openstack/senlin: Refactor cluster recover action  https://review.openstack.org/28589215:37
*** zzxwill has quit IRC15:43
*** Qiming has quit IRC15:52
*** lixinhui_ has quit IRC16:18
*** openstackgerrit has quit IRC17:18
*** openstackgerrit has joined #senlin17:19
openstackgerritMerged openstack/senlin: Fix various problems in health manager  https://review.openstack.org/28605818:05
*** Qiming has joined #senlin23:30
openstackgerritQiming Teng proposed openstack/senlin: Add unit test for cluster check action  https://review.openstack.org/28589623:35

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!