[HDFS-7498] Simplify the logic in INodesInPath

• Type: Sub-task
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 2.7.0
• Component/s: namenode
• Target Version/s: 2.7.0
• Hadoop Flags: Reviewed
• Assignee: Jing Zhao
• Reporter: Jing Zhao
• Created: 09/Dec/14 07:15
• Updated: 20/Nov/17 20:48
• Resolved: 09/Dec/14 19:40

Description

Currently we have relatively complicated logic in INodesInPath:
1) It can contain null elements in its INode array, and in mkdirRecursively these null INodes are replaced with new directories.
2) Operations like rename may also replace the inode in its INode array
3) getINodes requires trimming inodes array if the INodesInPath is derived from a dot-snapshot path
4) A lot of methods directly use/manipulate its INode array

We aim to simplify the logic of INodesInPath in this jira. Specifically, we can
make INodesInPath an immutable data structure and move the inode trimming logic to path resolving.

Issue Links

• breaks: HDFS-12171 Reduce IIP object allocations for inode lookup
• is related to: HDFS-7104 Fix and clarify INodeInPath getter functions