13:00:35 #startmeeting senlin 13:00:35 Meeting started Tue Jun 30 13:00:35 2015 UTC and is due to finish in 60 minutes. The chair is Qiming. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:00:36 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:00:38 The meeting name has been set to 'senlin' 13:00:48 okay, no dash 13:00:52 hello 13:00:54 :) 13:00:56 o/ 13:01:10 good morning/evening 13:01:21 hi 13:01:28 please review meeting agenda: https://wiki.openstack.org/wiki/Meetings/SenlinAgenda 13:01:30 hi 13:01:48 if you want to add something, just edit 13:02:31 or you can shoot it during open discussion 13:02:34 #topic test case coverage 13:03:01 in last meeting, we called for filing bugs related to test case coverage 13:03:21 my latest number is 63% coverage, not a bad number 13:04:07 we still need to spend quite some time on writing test cases, especially those related to profiles, actions, policies 13:04:30 https://bugs.launchpad.net/senlin/+bugs?field.searchtext=&orderby=-importance&field.status%3Alist=NEW&field.status%3Alist=TRIAGED&assignee_option=any&field.assignee=&field.bug_reporter=&field.bug_commenter=&field.subscriber=&field.structural_subscriber=&field.tag=&field.tags_combinator=ANY&field.has_cve.used=&field.omit_dupes.used=&field.omit_dupes=on&field.affects_me.used=&field.has_patch.used=&field.has_branches.used=&field.has_branches=on&fie 13:04:31 ld.has_no_branches.used=&field.has_no_branches=on&field.has_blueprints.used=&field.has_blueprints=on&field.has_no_blueprints.used=&field.has_no_blueprints=on&search=Search 13:04:36 i will try to make some time this week 13:04:38 I found some modules like senlin/engine/nodes.py are not well tested 13:04:46 okay, this link is too long, sorry 13:05:04 I will try to start working on scaling and lb policy related test cases after finish related wiki item 13:05:06 haiwei, feel free to file a bug for that then 13:05:25 ok, already filed some 13:05:36 we talked about setting milestones/deadlines for those bugs 13:05:56 however, it is really up to how many resources we can put on it 13:06:10 in case you are not familar with the milestones 13:06:20 https://launchpad.net/senlin/+milestones 13:06:40 we have missed liberty-1, obviously 13:07:04 it doesn't make good sense to release the packages at this stage 13:07:19 next milestone would be 2015-07-29 13:07:41 which means we have 90 days or so to work on this, among other things 13:08:05 29 days? 13:08:17 so ... maybe we should just encourage people to claim bugs and set deadlines themselves 13:08:36 oh, no, 30 days 13:08:53 sounds terrible 13:08:58 maybe a plan but not deadline 13:09:02 aha, time is very tight 13:09:02 I will have a summer vocation from 7/10 though 13:09:24 if we focus on test case, I think we might be able to finish it 13:09:25 wow 13:09:42 we will work hard on this, using all resources we have, and see what happens by liberty-2 13:09:46 anyway, I will try my best in other days 13:10:14 ok, I think at least we should finish test cases for all policies and actions 13:10:16 Qiming and yanyan, do you have vacation plan? 13:10:27 I don't have for this summer 13:10:45 okay 13:10:48 although we have a heavy pressure on this, please don't lower your bar when review, ;) 13:11:06 lixinhui, not yet 13:11:06 sure :) 13:11:18 will let the team know if so, :) 13:11:24 of courese 13:11:31 okay, moving on 13:11:37 #topic documentation 13:11:57 we have been working on some docs for developers during the past weeks 13:12:14 now we have already covered most aspects of the design 13:12:20 it's your jobs, thanks 13:12:30 i have been going through the documentation in order to get caught up on the project. what is there is very good 13:12:37 yeah, my job, :) 13:12:52 otukare 13:13:00 if you have ever checked the output from the python-docs gate, it is pretty cool 13:13:25 what we still need to do is docs for users (user-guide, getting-started ...) 13:13:28 and wiki 13:13:31 yes, saw it :) 13:13:46 will work on this 13:13:54 it will be my high priority 13:14:05 Great! 13:14:11 will also work on it in this week 13:14:20 for some reasons: even a new developer would need to know how to try things out 13:14:46 yes... magnum has a good getting started guide if you need a template to use 13:14:49 yes, you don't need to explain to them one by one 13:14:51 and for others who just heard of the word "senlin", he/she will check openstack wiki first 13:15:09 thanks, I was thinking about that 13:15:16 maybe we should provide a quick start guide at first 13:15:33 I really hope we don't need to write the same thing over and over again, all over different places 13:15:47 I think I have a good news, some people from my company has began to know senlin, not by my introduction 13:15:57 I will check openstack-manuals project, for example, to learn how they structure a guide for users 13:16:09 that means many people may have already known senlin 13:16:20 hopefully, we will only need to copy our writings to that project when senlin gets "integrated" 13:16:58 so if the docs get ready soon, will be very helpful 13:16:59 we have to polish the wiki page as well, it will be the "frontpage" for the project 13:16:59 ah, that makes sense qiming 13:17:22 have just done something: wiki.openstack.org/wiki/Senlin 13:17:36 need to add some details there 13:18:17 any helps on that would be appreciated 13:18:29 #topic object deletion handling 13:18:48 haiwei do you want to talk about it? 13:18:57 https://bugs.launchpad.net/senlin/+bug/1469939 13:18:57 Launchpad bug 1469939 in senlin "dead lock happens when trying to delete a node" [Undecided,New] 13:18:57 yes 13:19:58 in fact I want to talk about another first https://bugs.launchpad.net/senlin/+bug/1469953 13:19:59 Launchpad bug 1469953 in senlin "Node deletion error" [Undecided,In progress] - Assigned to Haiwei Xu (xu-haiwei) 13:20:00 I guess another action might hang there and didn't release node lock correctly 13:20:08 and thus caused this deadlock 13:20:21 yes, yanyanhu 13:20:27 seems to me like the bugs are caused by the same problem? 13:20:45 not exactly i think 13:21:20 1469953 is caused by node deletion when deleting a heat stack will check the heat stack status first 13:21:35 it is not a "dead" lock, it's actually about ensuring locks are released when not needed 13:21:38 see the source http://git.openstack.org/cgit/stackforge/senlin/tree/senlin/profiles/os/heat/stack.py#n117 13:21:50 Qiming, yes 13:22:00 but when senlin sends a DELETE request to heat, heat stack will remain CREATE_COMPLETE for a while before going to DELETE_IN_PROGRESS. so we will get CREATE != DELETE error. 13:22:42 the solution can be add a sleep(1) before checking the status. https://review.openstack.org/#/c/196947/1/senlin/profiles/os/heat/stack.py 13:22:58 sleep(1) is so .... ugly 13:22:58 seems we should provide another status option here 13:23:03 this bug is not related to lock I think 13:23:13 yes, Qiming 13:23:19 need to come up with a better status check 13:23:40 we just need to wait for heat a moment 13:23:40 I think we should change the implementation of _check_action_complete 13:23:57 yanyanhu_, that is a good idea 13:24:05 okay 13:24:42 I will think about it later 13:24:48 let's see what we can do 13:24:57 yes :) 13:25:10 that leads us also to the next topic 13:25:18 #topic SDK exception handling 13:25:32 because _check_action_complete is used by creation, deletion and updating, it should be careful 13:25:56 yes 13:25:58 when Senlin is talking to other OpenStack service, we do that through OpenStackSDK 13:26:32 not only we have to cope with resource status carefully, we have to consider all types of exceptions 13:26:40 yes, currently we are not handling sdk's exception well i think 13:26:42 agree 13:26:57 except sdk.exc.HttpException as ex: 57 raise ex 13:27:05 the sad news is that there is no formal guidelines or even consensus among projects today 13:27:13 we just catch HttpException and raise it again 13:27:17 yes, I guess this is the only exception we catch from openstacksdk 13:27:50 speaking of that, please refer to senlinclient code here 13:27:51 http://git.openstack.org/cgit/stackforge/python-senlinclient/tree/senlinclient/common/exc.py#n200 13:27:59 so what about translating HTTPException to a senlin InternalError 13:28:03 http://git.openstack.org/cgit/stackforge/python-openstacksdk/tree/openstack/exceptions.py 13:28:18 when I was writing the client code, I was trying to consider all possibilities 13:28:20 exception definition in sdk 13:29:02 catch all exception types we can and either kill them inside senlin or return them to users in a meaningful way 13:29:28 we can keep an eye on SDK's exception handling 13:29:48 yes 13:29:58 but we still need to be prepared for the "rare" cases where exceptions are not handled by SDK 13:30:14 so far, I know they have been dealing with 404 very well, ;) 13:30:19 yanyanhu_, we can only get HttpException from SDK? 13:30:33 also some other exceptions 13:30:51 the problem is "some other exceptions" 13:30:58 but I think this is the most common one we will use 13:31:03 we don't know their type, format, message 13:31:27 they are all here? http://git.openstack.org/cgit/stackforge/python-openstacksdk/tree/openstack/exceptions.py 13:31:31 sometimes, maybe, we can only get an error code, in a strange field ... 13:31:35 yes, so I think we don't have much choice now 13:32:14 Qiming: if you need something more out of that exceptions module, perhaps enter a bug on https://bugs.launchpad.net/python-openstacksdk and we can look into it. i dont think we've done much with those exceptions 13:32:17 I did encountered cases where Heat errors were not caught by SDK, but I cannot recall the details 13:32:38 aha, briancurtin, didn't expect you here, ;) 13:32:45 :) 13:32:48 Qiming: my IRC client highligts on "SDK" 13:32:50 it would be pretty cool if we can help improve that 13:33:33 okay, guys, all questions about SDK go to briancurtin, now and in future 13:33:51 maybe we also should enter SDK's meeting to give them some suggestions :) 13:34:06 haiwei, that would be cool 13:34:28 our meeting is today, Tuesdays at 1900 UTC 13:34:30 I really see SDK a foundational service in openstack, it will make a lot things easier 13:34:38 and we can also help them with some implement 13:34:57 1900+0800-2400=0300 13:35:01 ... 13:35:14 midnight in our time zone... 13:35:48 briancurtin, do you have a alternate meeting slot? 13:35:50 hmm, i will look around the calendar and see if there are other times that might work. which timezone are you? 13:36:00 GMT +8 13:36:04 so file a bug to SDK maybe a good idea 13:36:12 haiwei, +1 13:36:16 and help fix it 13:36:17 right now the meeting is in the afternoon for all US people, so we can maybe move it 13:37:04 cool, let us know when there are meetings we can attend, briancurtin 13:37:18 thanks, briancurtin :) 13:37:19 it might be hard to find a time that works since we are 13 hours apart, but maybe we can figure something out 13:37:44 er, 11 13:37:47 okay, talk though code and review then 13:37:49 I am GMT+9 13:38:12 a little easier maybe 13:38:23 comment is also a good way I believe 13:38:33 timezone is really annoying, and you cannot get rid of it 13:39:06 so ... guys, we work with the sdk team to handle exceptions 13:39:42 when we see exceptions from other services, let's talk whether we should move the handling logic to sdk 13:39:51 first we should understand SDK exception's type, i think, I have no idea of HttpException 13:39:58 it seems like a common exception 13:40:16 okay, that's your homework, haiwei, :) 13:40:29 ok 13:40:31 and my homework, obviously 13:40:56 let's move on 13:41:01 #topic open discussions 13:41:18 Qiming, you mean we should try to make most exceptions outside senlin get handled by SDK? 13:41:50 haiwei, we will study case by case, and see if it is common exception to be handled by sdk 13:41:59 hi, Qiming, I think maybe we should give a list of items which have highest priority 13:42:14 list of what? 13:42:28 exception type? 13:42:31 then we can know what is the focus we are woking on recently 13:42:46 nope, all work items of senlin 13:43:10 first things first, test cases 13:43:21 currently test case and exception handle i think 13:43:28 and also some bugs 13:43:30 that will help us find bugs 13:43:54 after unit test cases, we will have funtional tests 13:44:00 and your policy implementation 13:44:16 e.g. test cases, doc, new features we need to support 13:44:17 e.g. in next 30 days, I think the list will include doc, test cases, wiki 13:44:24 yes 13:44:50 okay, we can use etherpad for this 13:45:05 will open one for everyone to input/edit 13:45:08 so we can measure whether these items can be done before the deadline 13:45:31 ok 13:45:47 #action Qiming to open a etherpad for work items and schedule 13:45:53 cool 13:46:18 since we have talked about time zone problems 13:46:42 I really feel sorry for people like lkarm, jruano, ... 13:47:14 when they are online, we are mostly in dreams 13:47:35 the good news: https://review.openstack.org/#/c/196993/ 13:47:49 just figured out how to get our IRC chats logged 13:47:59 its 8am my time, so not too bad for me. i can usually stay online later in the night, but yes timezones are problematic 13:48:26 once the above patch is accepted, we will have IRC log archive 13:48:49 so you guys can check if needed 13:48:52 very good news 13:48:55 or we can use etherpad to ask and answer questions? 13:49:18 yanyanhu_, questions should go to mailinglist I guess 13:49:24 the senlin mailing is not well used now 13:49:31 hmm, right 13:49:55 yes, when we have some general questions, we should leverage that indeed 13:50:12 okay, guys, we still have 10 minutes 13:50:36 I'd like to share with you another "news" 13:50:55 we are inviting haiwei to be a core member of senlin project 13:51:12 yanyanhu_ and I had some discussions on this 13:51:31 haiwei has been a very active contributor since he gets involved in this project 13:51:52 thanks for trusting me, Qiming 13:52:00 he has gained enough knowledge to help the project moving forward 13:52:01 congrats haiwei 13:52:22 it would mean more responsibility and more time dedication though 13:52:23 thanks, it's my honor 13:52:29 so, haiwei, pitty on you 13:52:34 con :) 13:52:49 yes, will work even harder :) 13:52:58 since the team is not big, we are not announcing this in the mailinglist 13:54:12 anything else? 13:54:41 no from me 13:54:43 oh, about the topic proposal, so we have submitted it? 13:54:57 yanyanhu_, still waiting for company review 13:54:59 when the vote will start 13:55:03 ok 13:55:20 need to check out the deadline 13:56:29 okay, let's turn to #senlin if you have any questions, comments, suggestions ... 13:56:33 #endmeeting