Description of problem: Could get return when curl app's ip:port, but failed to curl with app's route Version-Release number of selected component (if applicable): openshift v3.6.151 kubernetes v1.6.1+5115d708d7 How reproducible: always Steps to Reproduce: 1.Prepare an app with route 2.Create app which check step1's route in builder container in same project $oc new-app ruby https://github.com/openshift-qe/sti-ruby-test.git $oc patch buildconfig sti-ruby-test -p '{"spec": {"strategy":{"sourceStrategy":{"env":[{"name":"APP_ROUTE","value":"rails-postgresql-example-xiu.0717-n1v.qe.rhcloud.com"}]}}}}' Actual results: Build failed to curl route in builder container. ---> Cleaning up unused ruby gems ... == Assemble === 54.165.58.205 STREAM rails-postgresql-example-xiu.0717-n1v.qe.rhcloud.com 54.165.58.205 DGRAM 54.165.58.205 RAW 54.82.188.5 STREAM 54.82.188.5 DGRAM 54.82.188.5 RAW error: build error: non-zero (13) exit code from registry.access.redhat.com/rhscl/ruby-23-rhel7@sha256:ef3ddd245a1d596e2c8ea320d705479a4c293d5a42c058ac82df4da91d8918d8 Expected results: Could access route in builder contianer Additional info: 1. Could access step1's route out of builder container. 2. Could access step1's app ip:port in builder container 3. Here is my change to include app's route in assemble script https://github.com/openshift-qe/sti-ruby-test/blob/master/.sti/bin/assemble#L66
This is a known issue, the workaround is to include the cluster DNS in the node's /etc/resolv.conf See further discussion here: https://trello.com/c/3Mb8EIh0 and here: https://github.com/openshift/origin/issues/14196 We'll leave this open to track the issue as well.
According to Mrunal, /etc/resolv.conf is supposed to be getting setup along with the container's networking group, so transferring to him to investigate.
still an issue guys?
Can't reproduce this issue in ocp 3.6（latest one is 3.6.173.0.104）and latest OCP 3.9 (v3.9.3) now
Could you help to set this bug to on_qa? Thanks
Move this bug to verified as the comment #4