Description of problem: openshift-ansble created a load balancer in PENDING_WAIT state. Amphora stuck in booting state. This could be related to https://bugzilla.redhat.com/show_bug.cgi?id=1620169, which had same symptoms. Version-Release number of selected component (if applicable): octavia-amphora-image-13.0-20190109.1.el7ost.noarch python2-octaviaclient-1.4.0-1.1.el7ost.noarch puppet-octavia-12.4.0-7.el7ost.noarch octavia-amphora-image-x86_64-13.0-20190109.1.el7ost.noarch How reproducible: I reproduced this 3 times in a row. openshift-ansible-3.11.82-3.git.0.9718d0a.el7.noarch Steps to Reproduce: 1. Deploy openstack 13 2. Deploy OCP 3.11 via openshift-ansible with ovds-multitenant sdn. Actual results: (overcloud) [stack@undercloud ~]$ openstack loadbalancer list -f value 1d3f3c90-56cb-429a-a22b-cf19f22cdb8b openshift-ansible-openshift.example.com-api-lb aa7c8d73e2674806ab7d00ee89a1ad54 172.16.1.10 ACTIVE octavia 2639d4d3-f531-428c-a4e3-84d4b7e6bfb7 openshift-cluster-router_lb-mxadjq7vjovd aa7c8d73e2674806ab7d00ee89a1ad54 172.16.1.15 PENDING_UPDATE octavia (overcloud) [stack@undercloud ~]$ openstack server list --all -f value | grep amphora 911b1f79-0049-42d9-97b4-a9f7468849a2 amphora-ceb47f7e-b43f-4026-9954-1a87abd2bf24 ACTIVE octavia-amphora-13.0-20190109.1.x86_64 1ae77193-3e50-44e5-b457-b149536925a1 amphora-4aca1cd5-75a2-4e8f-bcb9-fa36989a0a44 ACTIVE lb-mgmt-net=172.24.0.6; bastion_net=172.16.1.16 octavia-amphora-13.0-20190109.1.x86_64 (overcloud) [stack@undercloud ~]$ openstack loadbalancer amphora list -f value 4aca1cd5-75a2-4e8f-bcb9-fa36989a0a44 1d3f3c90-56cb-429a-a22b-cf19f22cdb8b ALLOCATED STANDALONE 172.24.0.6 172.16.1.10 9e93ee42-ecf7-45cb-813f-bd757433638a None BOOTING None None None Expected results: amphora boots and gets assigned network address. openshift-installer should probably fail if the LB router never comes online. Additional info: I sued the new THT timeout parameters during deployment instead of manually setting the timeout values: (overcloud) [stack@undercloud ~]$ cat templates/environments/24-octavia-timeout.yaml parameter_defaults: OctaviaTimeoutClientData: 1200000 OctaviaTimeoutMemberData: 1200000
Jacob, any chance of attaching sosreport from the controller nodes?
I redeployed the environment. Next time I see this behavior I will capture a sosreport.
I'm closing this one for now. Please don't hesitate to re-open it and attach sosreports.