Tuesday, 2017-11-14

« Monday, 2017-11-13 Index Wednesday, 2017-11-15 »
*** jmlowe has quit IRC00:03
*** liyi has joined #senlin00:36
*** liyi has quit IRC01:34
*** liyi has joined #senlin01:38
*** liyi has quit IRC02:51
*** liyi has joined #senlin02:54
*** liyi has quit IRC03:32
*** tman has joined #senlin03:35
*** liyi has joined #senlin03:36
tmanhi guys03:38
tmanI'm testing the affinity policy03:39
tmanFollowing this guide, I understand that when it's not needed to create server_group in nova beforehand03:39
tmanhttps://docs.openstack.org/senlin/latest/contributor/policies/affinity_v1.html03:40
tmanI tried but failed to scale-out the cluster with an affinity policy attached03:41
tmanHere is the senlin-engine.log03:42
tmanhttp://paste.ubuntu.com/25958257/03:43
tmanaffinity policy spec:03:43
tmanhttp://paste.ubuntu.com/25958266/03:44
tmanthe profile spec doesn't also include the "scheduler_hints:     group:" property03:50
tmanprofile spec:03:51
tmanhttp://paste.ubuntu.com/25958294/03:51
chenyb4tman, you can use "senlin profile-type-show os.nova.server-1.0" get about scheduler_hints04:06
openstackgerritOpenStack Proposal Bot proposed openstack/python-senlinclient master: Updated from global requirements  https://review.openstack.org/51917104:20
tman@chenyb4: but the guide said that it's not needed to declare the server_group name in both of profile and policy specs04:33
tmanquoted: "If the profile spec doesn’t contain a scheduler_hints property or the scheduler_hints property doesn’t have a group value, the Senlin engine will use the servergroup.name value from the affinity policy spec, if provided. If the policy spec also failed to provide a group name, the Senlin engine will try to create a server group with a random name, e.g. “server_group_x2mde78a”. "04:33
tmanI expected that senlin would automatically create the server_group with a random name04:35
tmanis there a mistake in the guide?04:36
ruijiehi tman, looks like a bug05:19
ruijieby the way, which version of senlin are you using?05:21
*** liyi has quit IRC05:33
*** liyi has joined #senlin05:41
*** liyi has quit IRC05:56
*** liyi has joined #senlin06:01
tman@ruijie: I'm using v4.0.006:02
ruijiehttp://git.openstack.org/cgit/openstack/senlin/tree/senlin/profiles/os/nova/server.py?h=4.0.0&id=59c6ea6617f1199d26d6e00ef431c3e5e42a7089#n84806:06
ruijieem.. this 'hints' should not be a 'NoneType'06:07
*** maestropandy has joined #senlin07:28
*** maestropandy has left #senlin07:36
tmanI got another error when creating a receiver07:41
tmanhttp://paste.ubuntu.com/25959124/07:43
tmanmy command to create a receiver:07:44
tman# openstack cluster receiver create --cluster demo --action CLUSTER_SCALE_IN w_scale_in 'unicode' object has no attribute 'get'07:44
tman'unicode' object has no attribute 'get' --> the output of the command07:44
*** liyi has quit IRC07:56
*** chenyb4 has quit IRC07:57
ruijiehttp://paste.openstack.org/show/626239/08:02
*** liyi has joined #senlin08:02
ruijiemine is ok, @tman08:04
ruijieare you using devstack? looks like some of your dependency packages are not latest version08:05
*** chenyb4 has joined #senlin08:07
tman@ruijie, I'm not using devstack08:09
tmandeploying in docker container08:10
tmanI just rebuilt the image and recreate the containers (senlin api/engine)08:10
tmanthe problem solved08:10
tmanthe prv. image using master branch08:11
tmanthe rebuilt image, i switch to 4.0.0/pike08:11
*** XueFeng has quit IRC08:11
ruijiewe are using versioned object to construct the request object: oslo.versionedobjects>=1.28.0 # Apache-2.008:12
tmanin the Dockerfile, I use that file http://git.openstack.org/cgit/openstack/requirements/plain/upper-constraints.txt?h=stable/pike08:14
tmanI checked the version, it is: oslo.versionedobjects===1.26.008:15
tmanbut it would still ok08:15
tman@ruijie: I'm confused with another case08:19
tmanI am testing health policy08:19
tmanhere is the spec:08:19
tmanhttp://paste.ubuntu.com/25959268/08:20
tmanI shutdown one Node in 2-node Cluster manually (Shutoff button in Horizon)08:21
tmanand expect that Senlin would (hard) reboot it08:21
tmanbut actually, Senlin just delete it and recreate a new node08:22
ruijieyes tman, it is a bug, senlin will not use the user defined action to recover the server08:22
ruijieefforts are needed to improve it08:23
ruijiehi Qiming, there08:24
Qiminghi08:24
QimingI believe there are some missing links in the call path08:24
Qimingshould not be difficult to fix though08:25
ruijieyes Qiming, these action parameters are not set to HealthRegistry object08:27
Qimingwe know what happened, we just need hands to work on it08:28
ruijieI think chenyb4 is working on it, discussed that yesterday08:28
ruijiebut the problem of the version object,http://paste.ubuntu.com/25959124/08:29
chenyb4About this question, I was fix it finish in this week. cc ruijie Qiming08:29
*** tman has quit IRC08:31
Qimingwhat is that about?08:32
Qimingsomething changed in versioned_object?08:32
chenyb4About health manager node recover parameter08:33
*** tman has joined #senlin08:34
ruijieI tried it in my local environment, every thing is ok08:35
ruijie# openstack cluster receiver create --cluster demo --action CLUSTER_SCALE_IN w_scale_in 'unicode' object has no attribute 'get'08:36
ruijie'unicode' object has no attribute 'get' --> the output of the command08:36
Qimingwhen did that happen?08:37
Qimingare you using master ?08:37
Qimingcould be caused by this one: http://git.openstack.org/cgit/openstack/senlin/commit/?id=30af0102e4fe98bd03f2ed3a5a1c26399b833d7008:39
tmanQiming, yes, I would use master branch when got the error "'unicode' object has no attribute 'get'"08:41
Qiminghave you upgraded your oslo.versionedobjects to >=1.28 ?08:42
tmanNope08:43
Qimingit could be that your lib is too old08:43
Qimingthe patch merged in oslo.versionedobjects is still not there08:43
Qimingyou may want to try: sudo pip install oslo.versionedobjects==1.28.008:44
tmannow I using 4.0.0/pike, with oslo.versionedobjects===1.26.008:44
tmanthe problem dissappear08:44
*** AlexeyAbashkin has joined #senlin08:47
*** XueFeng has joined #senlin09:18
chenyb4hi, ruijie09:59
*** liyi has quit IRC10:02
*** XueFeng has quit IRC10:44
*** XueFeng has joined #senlin10:44
tmanhi10:44
tmanI tried but failed to run this guide10:45
tmanhttps://docs.openstack.org/senlin/latest/scenarios/autoscaling_heat.html10:45
tmanit seems likely that there was a problem when Senlin had to handle many actions concurrently10:46
tmanit's ok if I run steps one by one as here: https://docs.openstack.org/senlin/latest/scenarios/autoscaling_ceilometer.html10:47
tmansenlin-engine.log: http://paste.ubuntu.com/25959867/10:49
tmanI found this as the same case, but maybe it is not resolved10:50
tmanhttps://bugs.launchpad.net/senlin/+bug/168162010:50
openstackLaunchpad bug 1681620 in senlin "ACTION failed when can't get the cluster lock" [Critical,Fix committed] - Assigned to XueFeng Liu (jonnary-liu)10:50
*** maestropandy has joined #senlin10:59
*** tman has quit IRC10:59
*** liyi has joined #senlin11:14
*** liyi has quit IRC11:25
*** liyi has joined #senlin11:33
*** maestropandy has quit IRC11:34
*** maestropandy has joined #senlin11:45
*** maestropandy1 has joined #senlin11:53
*** maestropandy1 has left #senlin11:53
*** maestropandy has quit IRC11:54
*** liyi has quit IRC12:17
*** liyi has joined #senlin12:17
*** liyi has quit IRC12:18
*** liyi has joined #senlin12:18
*** ruijie_ has joined #senlin12:42
*** zhurong has joined #senlin12:46
*** chenyb4_ has joined #senlin12:54
*** liyi has quit IRC13:04
*** chenyb4_ has quit IRC13:07
*** chenyb4_ has joined #senlin13:07
*** maestropandy has joined #senlin13:11
*** maestropandy has left #senlin13:11
openstackgerritQiming Teng proposed openstack/senlin master: Add lock retry logic  https://review.openstack.org/51964813:16
openstackgerritQiming Teng proposed openstack/senlin master: Improve action logging for easier debugging  https://review.openstack.org/51965013:16
*** zhurong has quit IRC13:26
XueFenghi13:35
*** liyi has joined #senlin13:49
*** liyi has quit IRC13:53
ruijie_we can persist the recovery action to db when attaching health-policy: http://git.openstack.org/cgit/openstack/senlin/tree/senlin/policies/health_policy.py#n19414:02
chenyb4_ruijie_: hand node-recover add before to health policy node-recover, the hand node-recover don't load action name, maybe use in node_action add pd = self.data.get('health', None)14:05
chenyb4_ruijie_: cluster_action was add. So, the cluster-recover can able to get health policy action name14:06
chenyb4_ruijie_: https://github.com/openstack/senlin/blob/master/senlin/engine/actions/cluster_action.py#L67914:06
ruijie_we will trigger node_recover action when detecting bad status of node14:09
ruijie_that is more reasonable to load HealthRegistry object to get the recovery action directly in HealthManager14:09
XueFengHi, share two things.One is self-healing forum in Sydeny submmit. https://etherpad.openstack.org/p/self-healing-rocky-forum14:09
ruijie_thanks XueFeng14:10
chenyb4_ruijie_: ok.14:11
XueFengThe second is Senlin Session in Sydeny summit14:11
chenyb4_XueFeng: thanks14:11
XueFenghttps://www.openstack.org/videos/sydney-2017/standing-up-kubernetes-on-openstack-is-cool-then-what14:13
ruijie_Qiming had a session about the HA: https://www.slideshare.net/cliffton75/high-availability-in-openstack-cloud14:14
XueFengAnd  another thing is that we should do more about https://www.openstack.org/videos/sydney-2017/standing-up-kubernetes-on-openstack-is-cool-then-what14:14
XueFengWe'd better to talk with the forum14:15
XueFengThis time in the forum, the ptl of vitrage give a picture central  by Vitrage.14:17
ruijie_vitrage --> senlin --> recover/mistral14:18
XueFengMay be this is not good choice. Need discuss again14:30
XueFengAbout k8s session, we should continue to do it. This session is just a start14:37
XueFengruijie_, chenyb4_14:38
ruijie_yes XueFeng, agree14:59
*** chenyb4_ has quit IRC15:02
*** chenyb4_ has joined #senlin15:03
*** chenyb4_ has quit IRC15:07
*** ruijie_ has quit IRC15:10
*** jmlowe has joined #senlin15:44
*** liyi has joined #senlin15:56
*** liyi has quit IRC16:01
openstackgerritEthan Lynn proposed openstack/senlin master: [WIP] Kubernetes on senlin  https://review.openstack.org/51532116:03
*** jmlowe has quit IRC16:43
*** jmlowe has joined #senlin17:02
*** jmlowe has quit IRC17:04
*** liyi has joined #senlin17:06
*** liyi has quit IRC17:10
*** AlexeyAbashkin has quit IRC17:41
*** liyi has joined #senlin18:06
*** liyi has quit IRC18:10
openstackgerritOpenStack Proposal Bot proposed openstack/python-senlinclient master: Updated from global requirements  https://review.openstack.org/51917119:08
*** AlexeyAbashkin has joined #senlin19:23
*** AlexeyAbashkin has quit IRC19:28
openstackgerritOpenStack Proposal Bot proposed openstack/python-senlinclient master: Updated from global requirements  https://review.openstack.org/51917119:28
*** liyi has joined #senlin19:45
*** liyi has quit IRC19:49
*** AlexeyAbashkin has joined #senlin20:23
*** AlexeyAbashkin has quit IRC20:28
*** liyi has joined #senlin20:35
*** liyi has quit IRC20:39
*** AlexeyAbashkin has joined #senlin21:25
*** AlexeyAbashkin has quit IRC21:29
*** jmlowe has joined #senlin21:35
*** liyi has joined #senlin21:44
*** liyi has quit IRC21:48
*** zigo has quit IRC21:59
*** zigo has joined #senlin22:01
*** jmlowe has quit IRC22:19
openstackgerritOpenStack Proposal Bot proposed openstack/python-senlinclient master: Updated from global requirements  https://review.openstack.org/51917122:37
openstackgerritOpenStack Proposal Bot proposed openstack/senlin master: Updated from global requirements  https://review.openstack.org/51980922:38
openstackgerritOpenStack Proposal Bot proposed openstack/python-senlinclient master: Updated from global requirements  https://review.openstack.org/51917122:50
openstackgerritOpenStack Proposal Bot proposed openstack/senlin master: Updated from global requirements  https://review.openstack.org/51980922:52
« Monday, 2017-11-13 Index Wednesday, 2017-11-15 »

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!