Chen He created AMBARI-20785:
--------------------------------
Summary: Ambari report datanode decommissioned but datanode is
still in decommissing
Key: AMBARI-20785
URL: https://issues.apache.org/jira/browse/AMBARI-20785
Project: Ambari
Issue Type: Bug
Components: infra
Affects Versions: 2.4.0
Reporter: Chen He
If we decommission HDFS datanode through ambari REST API call. It will create a
new request http://ambari_server:8080/api/v1/clusters/cluster_name/requests/
However, the request quickly response "COMPLETED" only after it added the given
datanode into dfs.exclude. It does not block till datanode fully
decommissioned. It should block till the datanode completely decommissioned.
At the same time, org.apache.ambari.groovy.client.decommissionDataNode() is
using the same way to decommission datanode. It could cause data loss if
cluster shutdown this node instantly after decommission the datanode.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
