(accumulo) branch elasticity updated: Remove todo from SplitUtils (#4484)

[dlmarion]

This is an automated email from the ASF dual-hosted git repository.

dlmarion pushed a commit to branch elasticity
in repository https://gitbox.apache.org/repos/asf/accumulo.git


The following commit(s) were added to refs/heads/elasticity by this push:
     new 2bee648380 Remove todo from SplitUtils (#4484)
2bee648380 is described below

commit 2bee6483805ef2329aefb9385110289391c6266a
Author: Dave Marion <dlmar...@apache.org>
AuthorDate: Fri May 10 08:03:19 2024 -0400

    Remove todo from SplitUtils (#4484)
    
    I don't know that there is a reliable way to determine what the splits 
should be given the information that we have in the tablet metadata. The todo 
suggested running tests and doing some compactions, etc. But I think that it's 
really going to be situation dependent. Users can apply iterators to perform 
aggregation, deletes, etc. at compaction time that could influence the split 
points greatly. I think the better option here is to document compactions 
should be run first to get more a [...]
---
 .../src/main/java/org/apache/accumulo/server/split/SplitUtils.java | 7 -------
 1 file changed, 7 deletions(-)

diff --git 
a/server/base/src/main/java/org/apache/accumulo/server/split/SplitUtils.java 
b/server/base/src/main/java/org/apache/accumulo/server/split/SplitUtils.java
index 8f64c461bc..5cf5c9edc6 100644
--- a/server/base/src/main/java/org/apache/accumulo/server/split/SplitUtils.java
+++ b/server/base/src/main/java/org/apache/accumulo/server/split/SplitUtils.java
@@ -188,13 +188,6 @@ public class SplitUtils {
   }
 
   public static int calculateDesiredSplits(long esitimatedSize, long 
splitThreshold) {
-    // ELASTICITY_TODO tablets used to always split into 2 tablets. Now the 
split operation will
-    // split into many. How does this impact a tablet with many files and the 
estimated sizes after
-    // split vs the old method. Need to run test where we add lots of data to 
a single tablet,
-    // change the split thresh, wait for splits, then look at the estimated 
sizes, then compact and
-    // look at the sizes after. For example if a tablet has 10M of data and 
the split thesh is set
-    // to 100K, what will the est sizes look like across the tablets after 
splitting and then after
-    // compacting?
     return (int) Math.floor((double) esitimatedSize / (double) splitThreshold);
   }