Thursday, 2019-04-25

*** rm_work has joined #senlin02:51
rm_workhey ya'll :P02:51
rm_workI'm getting this error when trying to attach a policy to a cluster:02:52
rm_workRESP BODY: {"code": 409, "error": {"code": 409, "message": "CLUSTER_ATTACH_POLICY for cluster '22752854-e600-4688-a5e4-e1f176a7b12b' cannot be completed because it is already locked.", "type": "ResourceIsLocked"}, "explanation": "There was a conflict when trying to complete your request.", "title": "Conflict"}02:52
rm_workThe cluster is ACTIVE... And I've waited a minute+ since the last action... what's going on?02:52
rm_workHow can I debug this?02:52
rm_workthe command was: openstack cluster policy attach --debug --enabled true --policy loadbalance-policy-v6-1512550 hello_world_cluster-151255002:53
rm_work(I had just successfully run another policy attach a minute before: openstack cluster policy attach --enabled true --policy loadbalance-policy-v4-1512550 hello_world_cluster-1512550 )02:53
rm_workcan they only have one policy?02:53
*** openstackgerrit has joined #senlin03:39
*** ChanServ sets mode: +v openstackgerrit03:39
openstackgerritMerged openstack/senlin master: Upgrade openstacksdk version to solve connections leak  https://review.opendev.org/64922203:39
sapd1rm_work,  We can attach more than one policy to a cluster.04:39
sapd1Maybe your cluster is not completed a job before you restart senlin-engine, So senlin-engine does not release a lock for the cluster.04:42
rm_workHmm04:42
rm_workThis is a new cluster created during test04:43
rm_workLiterally add one policy, wait, add the second04:43
rm_workSecond fails like that04:43
sapd1I don't get this issue before, wait for other folks,04:44
rm_workGood to know that it's possible to have multiple, at least04:58
dtruongrm_work When you attach a policy, you get back an action ID. You have to check the action status for that action ID and make sure it’s succeeded.05:20
dtruongIf it’s still in ready or running state, you won’t be able to attach another policy05:20
rm_workAh hmm ok06:03
rm_workI'll have to figure out how to do that06:03
sapd1dtruong, Have you tried to healing instances which are attached to a load balancer pool?06:29
sapd1in short I get the issue when recreate instance in cluster.06:30
sapd1step to reproduce 1. create a cluster size >= 1; 2. attach health policy - recreate instance; 3. attach a loadbalance policy; 4. delete an instance in cluster;06:30
sapd1So new instance will not be created in pool. because senlin cannot get profile of that instance.06:31
*** zigo has quit IRC08:27
rm_workah, though apparently for THIS specific type, I can't do two: `Only one instance of policy type (senlin.policy.loadbalance-1.1) can be attached to a cluster, but another instance (b28fe955-c3de-4311-83c4-c6ef1063a1c7) is found attached to the cluster (1778e12c-7bda-4d59-8907-43d4f2b2761a) already. `10:20
rm_workbut, waiting for "SUCCEEDED" status on the first action did make me get farther :D10:20
*** zigo has joined #senlin11:54
*** altlogbot_1 has quit IRC12:32
*** altlogbot_2 has joined #senlin12:40
*** altlogbot_2 has quit IRC12:47
*** altlogbot_3 has joined #senlin12:50
*** altlogbot_3 has quit IRC13:24
*** altlogbot_0 has joined #senlin13:34
*** altlogbot_0 has quit IRC13:42
*** altlogbot_2 has joined #senlin13:58
*** hyang has joined #senlin18:37
dtruongsapd1 no, I have tried health policy with load balancing policy20:18
dtruongWhat is the exact problem that you see?  After the instance is recovered, it is not added back to the LB pool?20:18
dtruongrm_work There is a blueprint for attaching multiple LB policies to the same cluster: https://blueprints.launchpad.net/senlin/+spec/senlin-cluster-multiple-lb-policies20:21
dtruongThis is the related patch set but it is still WIP: https://review.opendev.org/#/c/644474/20:21
openstackgerritMerged openstack/python-senlinclient master: Dropping the py35 testing  https://review.opendev.org/65305421:22
rm_workok thanks dtruong23:04

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