Description of problem: We have seen two instances now where, after a cluster was migrated from etcd v2 to v3, a master will revert to using v2 data after a restart of the atomic-openshift-master-api and atomic-openshift-master-controller services. In both cases, the clusters had been upgraded hours or days prior, and only after a restart of these services did they revert to using the old data. Version-Release number of selected component (if applicable): oc v3.6.173.0.96 kubernetes v1.6.1+5115d708d7 How reproducible: We have seen this twice so far on two different clusters. Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
etcd is at v3.1.3
Upstream PR that fixes it: https://github.com/openshift/openshift-ansible/pull/7551
This is already fixed in 3.7 and 3.6. Fix for master has been picked: https://github.com/openshift/openshift-ansible/pull/7556
Fix for 3.9: https://github.com/openshift/openshift-ansible/pull/7559
3.6 and 3.7 merged 8 days ago: 3.7: https://github.com/openshift/openshift-ansible/pull/7313 3.6: https://github.com/openshift/openshift-ansible/pull/7226
Related: https://bugzilla.redhat.com/show_bug.cgi?id=1544399
Tried both rpm and containerized etcd migration. Works well on openshift-ansible-3.6.173.0.110-1.git.0.ca81843.el7.noarch. After migration, new created data was stored in etcdv3 only. Combined comment11&comment13&comment15, change bug status.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2018:1106