[HDFS-8674] Improve performance of postponed block scans

• Type: Improvement
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 2.6.0
• Fix Version/s: 2.8.0, 3.0.0-alpha2
• Component/s: namenode
• Target Version/s: 2.8.0, 3.0.0-alpha2
• Hadoop Flags: Reviewed
• Assignee: Daryn Sharp
• Reporter: Daryn Sharp
• Created: 26/Jun/15 16:22
• Updated: 27/May/17 02:05
• Resolved: 01/Dec/16 18:22

Description

When a standby goes active, it marks all nodes as "stale" which will cause block invalidations for over-replicated blocks to be queued until full block reports are received from the nodes with the block. The replication monitor scans the queue with O(N) runtime. It picks a random offset and iterates through the set to randomize blocks scanned.

The result is devastating when a cluster loses multiple nodes during a rolling upgrade. Re-replication occurs, the nodes come back, the excess block invalidations are postponed. Rescanning just 2k blocks out of millions of postponed blocks may take multiple seconds. During the scan, the write lock is held which stalls all other processing.

Issue Links

• relates to: HDFS-11868 Backport HDFS-8674 to branch 2.7