Details
-
Task
-
Status: Open
-
Major
-
Resolution: Unresolved
-
None
-
None
-
None
Description
SCM container placement policies should check for duplicate hostnames while allocating pipelines. This can happen when a volume containing datanode.id.dir fails and reinitializes the datanode on restart. This could lead to SCM containing 2 nodes with the same ip & hostname but different datanode.id. This can happen because datanode.id is the primary identifier of a datanode in SCM. On pipeline allocation there is a possiblity of these 2 nodes getting picked up in a timeframe where the old node has not been marked dead and the new node has been registered.