Backport #18719
tests: lfn-upgrade-hammer: cluster stuck in HEALTH_WARN after last upgraded node reboots
History
#1 Updated by Nathan Cutler about 7 years ago
- Description updated (diff)
description¶
Scenario: jewel 10.2.6 integration testing
Symptom: "rados/singleton-nomsgr/{all/lfn-upgrade-hammer.yaml rados.yaml}" fails because, after last upgraded node is rebooted, cluster gets stuck in "HEALTH_WARN all OSDs are running jewel or later but the 'require_jewel_osds' osdmap flag is not set"
Root cause: the test does not issue the "ceph osd set require_jewel_osds" command after rebooting the last upgraded node.
Example failure: http://pulpito.ceph.com/loic-2017-01-26_22:01:29-rados-wip-jewel-backports-distro-basic-smithi/753628/
#2 Updated by Nathan Cutler about 7 years ago
Fixed in ceph-qa-suite master by http://github.com/ceph/ceph-qa-suite/commit/50758a4810794d265c5d36a71d1e16799251a00d but then the tests themselves were dropped by http://github.com/ceph/ceph-qa-suite/commit/660a2edaa2bea8597e7eccca31fb6c6aaf2d4501
#3 Updated by Nathan Cutler about 7 years ago
Have high hopes for a successful test run at http://tracker.ceph.com/issues/17851#note-45 (at the very end of the comment)
#4 Updated by Nathan Cutler almost 7 years ago
- Status changed from Fix Under Review to Resolved
- Target version set to v10.2.6