[HADOOP-3564] Sometime after successful hod allocation datanode fails to come up with java.net.BindException for dfs.datanode.ipc.address

• Type: Bug
• Status: Closed
• Priority: Blocker
• Resolution: Fixed
• Affects Version/s: 0.18.0
• Fix Version/s: 0.18.0
• Component/s: contrib/hod
• Hadoop Flags: Reviewed
• Release Note: Modifed HOD to generate the dfs.datanode.ipc.address parameter in the hadoop-site.xml of datanodes that it launches.
• Assignee: Vinod Kumar Vavilapalli
• Reporter: Karam Singh
• Created: 16/Jun/08 06:57
• Updated: 22/Aug/08 19:50
• Resolved: 18/Jun/08 05:47

Description

From Jira: HADOOP-3283 which introduced new conf parameter dfs.datanode.ipc.address which defaults to 50020.
When static dfs of hadoop version 0.18.0 running and its conf is not having dfs.datanode.ipc.address specified, then datanode start with 50020 port for ipc, w
When we use hod allocate without using static dfs.datanode on some machine fails to come. On further investigation it has been found sometimes when torque provides list nodes, that list also contain some static dfs node.
When hodring tries to start datanode on a machine where a static dfs datanode of hadoop 0.18.0 is running, hod's dynamic dfs datanode fails to come with exception -: java.net.BindException: Problem binding to /0.0.0.0:50020 : Address already in use
beacuse hod provides ports for dfs.datanode.address and dfs.datanode.http.address.