Hide Forgot
Description of problem: update_cluster will fail for older apps where a gear's first port interface entry is not a web framework Version-Release number of selected component (if applicable): How reproducible: 100% for the affected gears Steps to Reproduce: 1. Find or create a scaled application where at least 1 of the gear's port_interfaces array has the haproxy port_interface first, followed by the web_framework port_interface 2. Attempt to scale up/down or run oo-admin-ctl-app -c update-cluster Actual results: update_cluster mcollective call to the node will fail with `rescue in with_container_from_args' proxy_port is required Expected results: update_cluster should succeed Additional info: This line (https://github.com/openshift/origin-server/blob/master/plugins/msg-broker/mcollective/lib/openshift/mcollective_application_container_proxy.rb#L1732) should change to get the first port_interface that has a web_framework type, instead of just using the first port_interface in the array
Fixed with https://github.com/openshift/origin-server/pull/4350
Verified on devenv_4148 and devenv-stage_619 Steps: 1. Create a scalable application, set min scale to 2 2. Reverse the order of port_interfaces of the app so that the haproxy is 1st port_interface of the application 3. Scale up this application Result: No exception was caught, this problem should be fixed