Monday, 2015-12-14

*** xuhaiwei has quit IRC00:06
*** pratikmallya has quit IRC00:10
*** lvdongbing has joined #senlin00:37
*** pratikmallya has joined #senlin00:50
*** xuhaiwei has joined #senlin00:55
*** Qiming has joined #senlin01:07
*** Yanyanhu has joined #senlin01:38
*** zhenguo has joined #senlin01:42
*** elynn__ has joined #senlin02:01
*** pratikmallya has quit IRC02:04
xuhaiweiwhy openstackbot is not invited to Senlin channel?02:05
Qimingseems they are unstable at the moment02:05
Qimingthey may need rests as human02:06
*** elynn__ has quit IRC02:06
xuhaiweiyes, not only senlin02:06
*** elynn has joined #senlin02:06
xuhaiweieveryone does not want to work on Sunday02:07
xuhaiweion Monday02:07
*** cschulz has quit IRC02:16
*** branw has joined #senlin02:47
*** R_lixh has joined #senlin03:26
*** Liuqing has joined #senlin03:39
*** elynn has quit IRC04:02
*** Qiming has quit IRC04:29
*** elynn has joined #senlin05:01
*** elynn_ has joined #senlin05:03
*** elynn has quit IRC05:06
*** elynn_ has quit IRC05:15
*** elynn has joined #senlin05:15
*** Qiming has joined #senlin05:21
elynnHi Qiming, Yanyanhu, I'm working on adding lock breaker functional tests to senlin, but I don't know how to kill a senlin-engine during tests, do you know how ?05:37
elynnOr can I operate db_api directly ?05:37
Yanyanhuyou mean kill the engine process?05:37
Yanyanhuif so you can use 'kill' cmd I think :)05:37
*** Liuqing has quit IRC05:37
elynnYanyanhu: Yes05:37
Yanyanhuor 'kill -1' if you want to restart an engine05:37
elynnCan I use kill cmd in functional tests?05:37
Yanyanhuyes I think05:38
elynnI'm afraid that will affect other tests in parallels. Is there any way to directly operate db_api>05:38
Yanyanhujust like what we did for reloading some configurations before starting the functional test05:38
Yanyanhuoh, this is a problem05:38
Yanyanhuelynn, I'm afraid not... unless there is an interface defined for this purpose05:39
Yanyanhuhi, elynn, can restarting engine meets your requirement?05:40
Yanyanhuit won't influence other functional test cases I think05:40
elynnYanyanhu: How to restart engine?05:41
Yanyanhujust kill -1 will restarted I think05:41
Yanyanhubut I'm not sure whether the engine id will be changed05:41
elynnWhy it won't affect other functional tests?05:41
Yanyanhuhmm, it is based on the assumption that functional test is not run in parallel05:42
Yanyanhuor can we split functional test about lock breaker from other cases?05:43
elynnYanyanhu: split out means put it in a new class?05:43
Yanyanhulet me see05:44
Yanyanhuhmm, maybe this is infeasible05:45
Yanyanhuunless we split them in tox05:45
elynnYes, It's hard to simulate this situlation from functional tests if I can't operate on db api...05:46
Yanyanhuyes, understand05:47
*** Qiming has quit IRC05:47
*** Qiming has joined #senlin05:47
Yanyanhubecause senlin is a process running as service in devstack gate VM, not sure how to touch those interfaces inside the process05:48
Yanyanhuif there is no API for it05:48
Yanyanhucan we control the sequence of test cases?05:49
*** Liuqing has joined #senlin05:49
elynnYanyanhu: I'm not sure about it, still investigating...05:50
YanyanhuBTW, just made a test, using kill -1 to restart senlin service will change the engine id05:51
Yanyanhuthe old engine process will be killed I think05:51
elynnYanyanhu: That's cool!05:51
elynnfunctional test on my local env failed, I'm not sure what's going on.05:52
Yanyanhuyep, so if we can figure out how to avoid influencing other test cases, you can use this way to generate engine failure05:52
Yanyanhuany error log?05:52
elynnSomething like this05:54
Yanyanhuoh, you may need to apply this patch if you did the test in multi-engine env
Yanyanhuok, let me check it05:54
Yanyanhuah, this is a place we may need to promote in future05:55
Yanyanhuthe current exception log in functional test just gave the information where the error happen05:55
Yanyanhubut more useful information is actully in engine log05:56
Yanyanhue.g. why the action failed during test05:56
elynnLet me try to set num_engine to 105:56
Yanyanhuelynn, I think you can check the engine log and locate the error based on the timestamp of exception05:56
Yanyanhuelynn, ok05:57
Yanyanhualso suggest you to apply this patch
elynnYanyanhu: ok, let me try...05:57
*** Qiming_ has joined #senlin06:02
*** Qiming has quit IRC06:03
*** yuanying has joined #senlin06:11
*** elynn has quit IRC06:24
*** elynn has joined #senlin06:31
*** shu-mutou has joined #senlin06:36
*** lvdongbing has quit IRC06:52
*** openstackstatus has joined #senlin07:06
*** ChanServ sets mode: +v openstackstatus07:07
*** openstackgerrit has joined #senlin07:10
*** shu-mutou has quit IRC07:16
openstackgerritMerged openstack/senlin: Engine support for receivers
openstackgerritMerged openstack/senlin: Update TODO list
openstackgerritjunxu proposed openstack/senlin: Fix cluster node updating failed with new image
openstackgerritMerged openstack/senlin: Remove `return None` for functions in file senlin/engine/
Qiming_elynn, there?08:06
elynnHi Qiming_08:06
openstackLaunchpad bug 1514742 in senlin "Failed to acquire lock if engine restarts during a action" [Undecided,New]08:07
Qiming_are you linking back your fix to the bug report?08:07
Qiming_or should I just close it?08:07
elynnJust close it would be ok.08:07
Qiming_got it, thx08:08
openstackgerritMerged openstack/senlin: Fix cluster node updating failed with new image
openstackgerritQiming Teng proposed openstack/senlin: Fix list data type implementation
openstackgerritQiming Teng proposed openstack/senlin: Fix list data type implementation
*** Qiming_ has quit IRC09:14
openstackgerritEthan Lynn proposed openstack/senlin: Use jsonutils from oslo in engine/
openstackgerritEthan Lynn proposed openstack/senlin: Refactor YamlLoader
openstackgerritEthan Lynn proposed openstack/senlin: Add functional tests to engine.parser
elynnUnittests for engine/parser ^09:34
openstackgerritEthan Lynn proposed openstack/senlin: Add unit tests to engine.parser
*** Liuqing has quit IRC09:57
*** zhenguo has quit IRC10:04
*** Yanyanhu has quit IRC10:14
*** Qiming has joined #senlin10:27
*** elynn has quit IRC10:33
openstackgerritQiming Teng proposed openstack/senlin: Add unit tests to engine.parser
openstackgerritMerged openstack/senlin: Use jsonutils from oslo in engine/
openstackgerritMerged openstack/senlin: Revise list field read/write
openstackgerritMerged openstack/senlin: Fix behavior of deletion policy with 'RANDOM' constrain not in random way for selecting candidates.
openstackgerritMerged openstack/senlin: Refactor YamlLoader
openstackgerritzhangguoqing proposed openstack/senlin: delete __context__ that unused
*** pratikmallya has joined #senlin13:02
*** elynn has joined #senlin13:03
*** pratikmallya has quit IRC13:11
*** cschulz has joined #senlin14:18
cschulzQiming I'm having trouble with nova clusters.14:20
cschulzUsing nova_stack example14:20
cschulzSenlin is getting 400 responses back from Nova API14:21
cschulzNova api log says "Invalid keyname"14:21
cschulzIs this a known problem?14:22
Qiminghi, cschulz, never encountered that before14:22
Qimingare you referring to this one?
openstackLaunchpad bug 1517353 in senlin "Node can't be created for KeyError: 'server'" [Undecided,New] - Assigned to Haiwei Xu (xu-haiwei)14:23
cschulzNo that is different.  In our case Nova API is catching an error and giving 400 status code.14:25
cschulzI'll open bug report with details14:25
Qimingbtw, cschulz, what do you mean by 'nova_stack' example?14:26
Qimingare you referring to this line?
Qimingit means you will have to create a nova keypair named 'oskey'14:28
Qimingor if you are using other nova keypair, you will need to revise that line14:29
cschulzGot it.  Sorry I missed that detail.14:29
Qimingsenlin is not returning useful information on this?14:30
cschulzThat is true unfortunately, but it is probably also a fault of nova as it is just responding with 400 bad request.14:38
cschulzworks now thanks for the help.14:39
Qimingabout returning 400, I think nova is following the api-wg guidelines here:
Qimingif you are creating a server in your request, and the server is referencing a resource (keypair here) that cannot be found14:42
Qimingit should return 400 bad request14:42
cschulzYes but for Senlin to be able to issue a helpful error message it need to have some indication of what was wrong with the nova request.14:43
cschulzI seem to remember running into some thing similar with Heat.  I think that Heat actually asks nova if the keypair exists.  Than it can issue an informed error message.14:46
Qimingright, when it is doing a stack validation14:47
cschulzYes your correct.14:47
*** zhenguo has joined #senlin15:17
*** zhenguo has quit IRC15:56
*** pratikmallya has joined #senlin16:30
*** Qiming has quit IRC16:43
*** pratikma_ has joined #senlin17:06
*** pratikmallya has quit IRC17:09
*** elynn has quit IRC17:10
*** openstackstatus has quit IRC17:20
*** openstackstatus has joined #senlin17:23
*** ChanServ sets mode: +v openstackstatus17:23
*** cschulz has quit IRC21:47
*** pratikma_ has quit IRC22:43
*** pratikmallya has joined #senlin23:28

Generated by 2.14.0 by Marius Gedminas - find it at!