[GitHub] [accumulo] keith-turner commented on issue #3078: Mark blocks for compacted files as not needed to be cached

[GitBox <gi...@apache.org>]

keith-turner commented on issue #3078:
URL: https://github.com/apache/accumulo/issues/3078#issuecomment-1315283430

   > I think there are two things that could be done here (assuming successful compaction):
   >
   > 1. Call setDropBehind(true) for the compaction input files so that the DataNodes can drop the files from their page caches.
   > 2. Remove (or cause to be removed faster) any cached blocks in the BlockCache for the compaction input files.
   
   We should not have to worry about compactions causing blocks to be cached. When a file is opened for compaction [here](https://github.com/apache/accumulo/blob/f389518b6b9c4897e017edade95f927846214e6e/server/base/src/main/java/org/apache/accumulo/server/compaction/FileCompactor.java#L314-L316) no accumulo block caches are passed in.
   
   Would be nice to setDropBehind(true) for compaction input files.  Could also do that for the output file.  The output file is created [here](https://github.com/apache/accumulo/blob/f389518b6b9c4897e017edade95f927846214e6e/server/base/src/main/java/org/apache/accumulo/server/compaction/FileCompactor.java#L215).  
   
   Seems like we would need to modify the internal reader and writer builders to support the set drop behind option.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: notifications-unsubscribe@accumulo.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org