[
https://issues.apache.org/jira/browse/HDFS-1646?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12999041#comment-12999041
]
Suresh Srinivas commented on HDFS-1646:
---------------------------------------
Tests such as TestDFSUpgrade start datanodes with invalid data (invalid version
file) etc. and expect datanode start to
fail. In these cases, Datanode#startDatanode() method throws an exception
during startup and datanode is not started.
With the introduction of separate thread for communicating with namenodes, the
method Datanode#startDatanodes() can not throw the exception. This results in
MiniDFSCluster.waitActive() to wait forever and tests timeout.
MiniDFSCluster.waitActive() waits until:
# Number of registered data nodes at namenode == Number of datanodes
MiniDFSCluster was started with.
However, if BPOfferService fails to start, it never registers with the
namenode, consequently waitActive() exit is condition is never met.
> MiniDFSCluster#waitActive() waits for ever in federation
> --------------------------------------------------------
>
> Key: HDFS-1646
> URL: https://issues.apache.org/jira/browse/HDFS-1646
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: test
> Affects Versions: Federation Branch
> Reporter: Suresh Srinivas
> Assignee: Suresh Srinivas
> Fix For: Federation Branch
>
>
> Tests where datanodes are expected to fail, hang in federation.
--
This message is automatically generated by JIRA.
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
