Description of problem: Jenkins migration is leaving oo-accept-node with an error: "Gear has a websocket framework cartridge but no websocket configuration" Version-Release number of selected component (if applicable): 1.2/2013-06-05.9 How reproducible: Always Steps to Reproduce: 1. Setup ose-1.1.3 env 2. Create a jenkins app 3. Following http://etherpad.corp.redhat.com/OSE-1-2-upgrade-notes to do upgrade testing 4. After upgrade, run oo-accept-node Actual results: # oo-accept-node FAIL: Gear has a websocket framework cartridge but no websocket configuration: 4a105df2e6e74e5e9de144b40a7927f6 Expected results: There should be no such error. Additional info:
Good catch. I noticed this and was digging through the oo-accept-node logic to try to figure why it was detecting that. It turns out, it was looking at the v1 cart and finding "--websocket" in the proxy hook. However, according to dmcphers that logic should be v1/v2-aware and so this shouldn't be happening. So it's either faulty oo-accept-node logic or (more likely) something about our migrated gears is misleading it. It appears to be ONLY Jenkins BTW.
Kenny and I ran into it as well as an issue where the DNS variable had capitals in it after migration. Pull requests for master: https://github.com/openshift/origin-server/pull/2780 Pull requests for stage: https://github.com/openshift/origin-server/pull/2781
Verify this against puddle: http://buildvm-devops.usersys.redhat.com/puddle/build/OpenShiftEnterprise/1.2/2013-06-13.1/ Setup an OSE env using puddle 1.1.z/2013-06-11.2. Create a series of applications including jenkins-1.4 app. Then update this env to puddle 1.2/2013-06-13.1, run oo-accept-node. [root@node ~]# oo-accept-node -v INFO: using default accept-node extensions INFO: loading node configuration file /etc/openshift/node.conf INFO: loading resource limit file /etc/openshift/resource_limits.conf INFO: checking node public hostname resolution INFO: checking selinux status INFO: checking selinux openshift-origin policy INFO: checking selinux booleans INFO: checking package list INFO: checking services INFO: checking kernel semaphores >= 512 INFO: checking cgroups configuration INFO: checking cgroups processes INFO: checking filesystem quotas INFO: checking quota db file selinux label INFO: checking 27 user accounts INFO: checking application dirs INFO: checking system httpd configs PASS No errors throw out, so verify this bug.
Closing all bugs introduced, fixed, and verified during 1.2 release work (thus never shipped).