Details
-
Type:
Bug
-
Status: Resolved
-
Priority:
Major
-
Resolution: Fixed
-
Affects Version/s: CDH4.2.1
-
Fix Version/s: None
-
Component/s: HDFS
-
Labels:None
Description
There are examples of the Standby Name Node sending administrative messages to the Data Nodes. There was a very high number of replication messages sent, these were rejected by the Data Nodes. The issue is viewed as relatively harmless, in that no corruption or service disruption is caused, however it does cause alarms in monitoring systems as there are a high number of messages such as:
2013-08-12 12:15:04,028 WARN org.apache.hadoop.hdfs.server.blockmanagement.BlockManager: PendingReplicationMonitor timed out blk_-123123123123123123_1862152
There is also a lot of data generated in the logs. The SBNN logs were over 5 times larger than the corresponding logs for the NN.