[ https://issues.apache.org/jira/browse/HDFS-6985?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Colin Patrick McCabe updated HDFS-6985: --------------------------------------- Attachment: HDFS-6985.001.patch I changed it so that all implementations of {{ReplaceDatanodeOnFailure#Condition}} use the same names for variables (i.e. "short replication, final DatanodeInfo[] existings, int nExistings" rather than "short replication, final DatanodeInfo[] existings, int n") I didn't use the 'final' qualifier on primitives, since most Hadoop code doesn't do that. I add the 'final' qualitifer to all uses of the 'existings' array. I added some JavaDoc to {{ReplaceDatanodeOnFailure#Condition}}. Clarified that dfs.client.block.write.replace-datanode-on-failure.best-effort applies to pipeline recovery. > Add final keywords, documentation, etc. to ReplaceDatanodeOnFailure > ------------------------------------------------------------------- > > Key: HDFS-6985 > URL: https://issues.apache.org/jira/browse/HDFS-6985 > Project: Hadoop HDFS > Issue Type: Improvement > Reporter: Colin Patrick McCabe > Assignee: Colin Patrick McCabe > Priority: Minor > Attachments: HDFS-6985.001.patch > > > * use final qualifier consistently in {{ReplaceDatanodeOnFailure#Condition > classes}} > * add a debug log message in the DFSClient explaining which pipeline failure > policy is being used. > * add JavaDoc for ReplaceDatanodeOnFailure > * documentation > dfs.client.block.write.replace-datanode-on-failure.best-effort should make it > clear that that the configuration key refers to pipeline recovery. -- This message was sent by Atlassian JIRA (v6.3.4#6332)