[HDDS-3703] Consider avoiding file lookup calls in writeChunk hotpath

• Type: Improvement
• Status: Resolved
• Priority: Blocker
• Resolution: Fixed
• Fix Version/s: 1.0.0
• Labels: Triaged performance
• Target Version/s: 1.0.0
• Assignee: Attila Doroszlai
• Reporter: Rajesh Balamohan
• Created: 02/Jun/20 12:07
• Updated: 08/Jun/20 18:58
• Resolved: 05/Jun/20 08:36

Description

In getChunkFile internally, it invokes "verifyChunkDirExists". This causes file existence checks for the directory and throws IO exception accordingly. If the file is anyways going to be written, it is better to handle it later and throw the same exception. This could avoid file checks for every "writechunk"

https://github.com/apache/hadoop-ozone/blob/master/hadoop-hdds/container-service/src/main/java/org/apache/hadoop/ozone/container/keyvalue/impl/FilePerBlockStrategy.java#L106

 

File channels are cached anyways in "OpenFiles". So if we can avoid "file.getAbsolutePath()", this could save memory and resolving paths. 

https://github.com/apache/hadoop-ozone/blob/master/hadoop-hdds/container-service/src/main/java/org/apache/hadoop/ozone/container/keyvalue/impl/FilePerBlockStrategy.java#L118

 

Also "validateChunkForOverwrite" can be optimised, as "isOverWritePermitted" would be false most of the times.

 

 

Issue Links

• supercedes: HDDS-2382 Consider reducing number of file::exists() calls during write operation
• links to: GitHub Pull Request #1013