If zookeeper is not running, then decreasing replicas will make cluster of zookeeper Unrecoverable #398
Comments
|
@stop-coding Could you please let us know how to reproduce this issue. After this once, if zookeeper starts running is the replica set not updated correctly? |
@anishakj Thank you for attention. For example：
I think removing a pod needs to ensure atomicity, that include connect success and reconfig success... |
stop-coding
changed the title
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Description
when a cluster of zookeeper is not running for some error, then decreasing replicas will delete pod automatically.
pod exec zookeeperTeardown.sh to connect zookeeper will fail ,then it will not remove node on this cluster of zookeeper.
But the pod have been delete without update zookeeper configure on zoo.cfg
Importance
must-have
Location
ZNODE_PATH="/zookeeper-operator/$CLUSTER_NAME"
CLUSTERSIZE=
java -Dlog4j.configuration=file:"$LOG4J_CONF" -jar /root/zu.jar sync $ZKURL $ZNODE_PATHecho "CLUSTER_SIZE=$CLUSTERSIZE, MyId=$MYID"
if [[ -n "$CLUSTERSIZE" && "$CLUSTERSIZE" -lt "$MYID" ]]; then
java -Dlog4j.configuration=file:"$LOG4J_CONF" -jar /root/zu.jar remove $ZKURL $MYID
echo $?
fi
Suggestions for an improvement
fix on zookeepercluster_controller.go:reconcileStatefulSet
If ClusterSize decrease, do reconfig remove here
The text was updated successfully, but these errors were encountered: