[HDFS-15543] RBF: Write Should allow, when a subcluster is unavailable for RANDOM mount points with fault Tolerance enabled.

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.1.1
• Fix Version/s: 3.4.0
• Component/s: rbf
• Assignee: Hemanth Boyina
• Reporter: Harshakiran Reddy
• Created: 27/Aug/20 10:03
• Updated: 07/Oct/20 04:32
• Resolved: 07/Oct/20 04:32

Description

A RANDOM mount point should allow to creating new files if one subcluster is down also with Fault Tolerance was enabled. but here it's failed.

MultiDestination_client]# hdfs dfsrouteradmin -ls /test_ec
Mount Table Entries:
Source Destinations Owner Group Mode Quota/Usage
/test_ec hacluster->/tes_ec,hacluster1->/tes_ec test ficommon rwxr-xr-x [NsQuota: -/-, SsQuota: -/-]

File Write throne the Exception:-

2020-08-26 19:13:21,839 WARN hdfs.DataStreamer: Abandoning blk_1073743375_2551
2020-08-26 19:13:21,877 WARN hdfs.DataStreamer: Excluding datanode DatanodeInfoWithStorage[DISK]
2020-08-26 19:13:21,878 WARN hdfs.DataStreamer: DataStreamer Exception
java.io.IOException: Unable to create new block.
at org.apache.hadoop.hdfs.DataStreamer.nextBlockOutputStream(DataStreamer.java:1758)
at org.apache.hadoop.hdfs.DataStreamer.run(DataStreamer.java:718)
2020-08-26 19:13:21,879 WARN hdfs.DataStreamer: Could not get block locations. Source file "/test_ec/f1.COPYING" - Aborting...block==null
put: Could not get block locations. Source file "/test_ec/f1.COPYING" - Aborting...block==null