Description of problem: During upgrade to v3.9.0-0.21.0: TASK [openshift_web_console : Apply the web console template file] ************* Friday 19 January 2018 00:26:06 +0000 (0:00:00.506) 0:11:16.462 ******** fatal: [free-int-master-3c664]: FAILED! => { "changed": true, "cmd": "oc process -f \"/tmp/console-ansible-zeZmuo/console-template.yaml\" --param API_SERVER_CONFIG=\"extensionScripts:\nextensionDevelopment: false\nclusterInfo:\n logoutPublicURL: ''\n metricsPublicURL: ''\n loggingPublicURL: ''\n consolePublicURL: https://api.free-int.openshift.com/console/\n masterPublicURL: https://api.free-int.openshift.com\nmasterPublicURL: https://api.free-int.openshift.com\nextensionProperties:\nservingInfo:\n namedCertificates:\n certFile: /var/serving-cert/tls.crt\n bindAddress: 0.0.0.0:8443\n bindNetwork: tcp4\n maxRequestsInFlight: 0\n keyFile: /var/serving-cert/tls.key\n clientCA: ''\n requestTimeoutSeconds: 0\nkind: WebConsoleConfiguration\napiVersion: webconsole.config.openshift.io/v1\nextensions:\nfeatures:\n inactivityTimeoutMinutes: 0\npublicURL: https://api.free-int.openshift.com/console/\nlogoutURL: ''\nmetricsPublicURL: ''\nextensionStylesheets:\nloggingPublicURL: ''\n\" --param IMAGE=\"registry.access.redhat.com/openshift3/ose-web-console:v3.9\" --param NODE_SELECTOR='{\"region\": \"infra\"}' --param REPLICA_COUNT=\"3\" | oc apply -f -", "delta": "0:00:00.364853", "end": "2018-01-19 00:26:06.845070", "failed": true, "rc": 1, "start": "2018-01-19 00:26:06.480217" } STDERR: error: unable to process template namespaces "aos-cd-smoketest-265672" not found error: no objects passed to apply MSG: non-zero return code
This namespace did in fact not exist at the time. The last process using the root account deleted the project in use. I've switched to a valid project and restarted the upgrade. However, I don't think this particular error should occur.
We need to either switch to using oc_* modules for the client bits or create a copy of /etc/origin/master/admin.kubeconfig in {{ mktemp.stdout }} and use that when we invoke the client which would ensure that we're operating as an admin in the default namespace.
This looks like a problem for template service broker, too (and maybe other places).
https://github.com/openshift/openshift-ansible/pull/6799
Commits pushed to master at https://github.com/openshift/openshift-ansible https://github.com/openshift/openshift-ansible/commit/7d7499efb48b00a320148d71e915193635707ea6 Bug 1536253: Pass `--config` flag on oc commands when installing console Fixes https://bugzilla.redhat.com/show_bug.cgi?id=1536253 https://github.com/openshift/openshift-ansible/commit/e692d7d757742f82dc6ba554110c78a0d0ccbcf6 Merge pull request #6799 from spadgett/console-admin-kubeconfig Automatic merge from submit-queue. Bug 1536253: Pass `--config` flag on oc commands when installing console Fixes https://bugzilla.redhat.com/show_bug.cgi?id=1536253 /assign @sdodson /kind bug @jupierce cc
Will check in OCP upgrade testing these days
We didn't meet this error during OCP upgrade, could you help check for free-int upgrade?
Verified the bug per comment 10, feel free to re-open if we met the issue against online