[HDFS-11015] Enforce timeout in balancer

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 2.8.0, 2.7.4, 3.0.0-alpha2
• Component/s: balancer & mover
• Target Version/s: 2.8.0
• Hadoop Flags: Reviewed
• Assignee: Kihwal Lee Assign to me
• Reporter: Kihwal Lee
• Created: 14/Oct/16 19:05
• Updated: 09/May/18 18:24
• Resolved: 25/Oct/16 19:21

Description

1) Hung node detection: HDFS-6247 has removed the socket read timeout while adding the periodic response for slow block moves. However, the removal of the long timeout wasn't necessary. The timeout is still useful for avoiding hung nodes and does not abort slow moves.

2) Enforcing the iteration limit:The 20 minute iteration limit is supposed to be enforced, but it is not. An iteration can easily stretch to 30 to 40 minutes with a long tail. Because of the long tails, the balancer throughput does not reach its full potential.

3) Slow move detection: For improved throughput, imposing block move timeout is sometimes necessary. We have seen an iteration taking over 2 hours because of one slow block move. This is mainly for catching exceptionally slow moves. Even if the balancer stops waiting, the move will continue and finish.

In order to not undo what HDFS-6247 tried to achieve, it should be possible to configure off 3).

Issue Links

• causes: HDFS-13174 hdfs mover -p /path times out after 20 min
• is related to: HDFS-10977 Balancer should query NameNode with a timeout
• is related to: HDFS-10966 Enhance Dispatcher logic on deciding when to give up a source DataNode
• is related to: HDFS-11051 Test Balancer behavior when some block moves are slow