Description of problem: OCP 3.7 uses Cassandra 3.0.14 which has a critical bug that we have hit - https://issues.apache.org/jira/browse/CASSANDRA-13696. The effect of this bug is that Cassandra keeps running but shuts down most services such that it stops accepting client requests and stops gossiping with other Cassandra nodes. The rest of the C* cluster will mark the effected C* node down. Any requests targeted for the node will result in UnavilableExceptions. The nodetool status command still reports UN, i.e., up and normal; consequently, the liveness probe does not fail, and the C* cluster remains in a bad state. Running nodetool status from another C* pod in the cluster will however report the effected C* node as down. There is a temporary work around. First, delete anything in the /cassandra_data/data/system/hints-<some_hash>/ directory. Then restart the Cassandra pod. Restarting the pod is a bit of a brute force approach. There is another option that is slightly more involved but will get Cassandra back running much more quickly. Here are the steps: 1) oc rsh <cassandra_pod> 2) nodetool enablebackup 3) nodetool enablegossip 4) nodetool enablebinary Within a few seconds the rest of the C* cluster should start reporting the effected C* node as up. Hinted handoff will still be disabled which is fine given the impact of CASSANDRA-13696. Note though that a pod restart will re-enable hinted handoff. It can be disabled with `nodetool disablehandoff`. Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
Verified with metrics-cassandra/images/v3.7.31-1, CASSANDRA_VERSION is 3.0.15 now, and passed smoke testing. Set it to VERIFIED # openshift version openshift v3.7.31 kubernetes v1.7.6+a08f5eeb62 etcd 3.2.8 # oc rsh ${hawkular-cassandra_pod} sh-4.2$ env | grep -i version JBOSS_IMAGE_VERSION=1.3 JAVA_VERSION=1.8.0 CASSANDRA_VERSION=3.0.15.redhat-1
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-2018:0636