Remove broken "defragment-on-read" optimization

The read path for names queries has had a "defragment-on-read"
optimization for a while whereby if too many sstables are hit by the
read, the result is written back into memtable, in the hope that later
reads will only read that newly written data in a single sstable (or at
least fewer).

The principle of that optimisation does not work however as data is
written back with the same timestamp as it originally has and that means
future reads cannot know to skip older sstables (at least with the
metadata we currently store).

As such, this optimisation never saved anything and in fact added load.

The patch removes that broken code.

Patch by Sylvain Lebresne, reviewed by Aleksey Yeschenko for
CASSANDRA-15432

CHANGES.txt

@@ -1,4 +1,5 @@
 3.0.22:
+ * Remove broken 'defrag-on-read' optimization (CASSANDRA-15432)
  * Check for endpoint collision with hibernating nodes (CASSANDRA-14599)
  * Operational improvements and hardening for replica filtering protection (CASSANDRA-15907)
  * stop_paranoid disk failure policy is ignored on CorruptSSTableException after node is up (CASSANDRA-15191)

src/java/org/apache/cassandra/db/SinglePartitionReadCommand.java

@@ -883,7 +883,6 @@ public class SinglePartitionReadCommand extends ReadCommand
 
         /* add the SSTables on disk */
         Collections.sort(view.sstables, SSTableReader.maxTimestampComparator);
-        boolean onlyUnrepaired = true;
         // read sorted sstables
         SSTableReadMetricsCollector metricsCollector = new SSTableReadMetricsCollector();
         for (SSTableReader sstable : view.sstables)
@@ -952,9 +951,6 @@ public class SinglePartitionReadCommand extends ReadCommand
                 if (iter.isEmpty())
                     continue;
 
-                if (sstable.isRepaired())
-                    onlyUnrepaired = false;
-
                 result = add(
                     RTBoundValidator.validate(isForThrift() ? ThriftResultsMerger.maybeWrap(iter, nowInSec()) : iter, RTBoundValidator.Stage.SSTABLE, false),
                     result,
@@ -972,30 +968,6 @@ public class SinglePartitionReadCommand extends ReadCommand
         DecoratedKey key = result.partitionKey();
         cfs.metric.samplers.get(TableMetrics.Sampler.READS).addSample(key.getKey(), key.hashCode(), 1);
 
-        // "hoist up" the requested data into a more recent sstable
-        if (metricsCollector.getMergedSSTables() > cfs.getMinimumCompactionThreshold()
-            && onlyUnrepaired
-            && !cfs.isAutoCompactionDisabled()
-            && cfs.getCompactionStrategyManager().shouldDefragment())
-        {
-            // !!WARNING!!   if we stop copying our data to a heap-managed object,
-            //               we will need to track the lifetime of this mutation as well
-            Tracing.trace("Defragmenting requested data");
-
-            try (UnfilteredRowIterator iter = result.unfilteredIterator(columnFilter(), Slices.ALL, false))
-            {
-                final Mutation mutation = new Mutation(PartitionUpdate.fromIterator(iter));
-                StageManager.getStage(Stage.MUTATION).execute(new Runnable()
-                {
-                    public void run()
-                    {
-                        // skipping commitlog and index updates is fine since we're just de-fragmenting existing data
-                        Keyspace.open(mutation.getKeyspaceName()).apply(mutation, false, false);
-                    }
-                });
-            }
-        }
-
         return result.unfilteredIterator(columnFilter(), Slices.ALL, clusteringIndexFilter().isReversed());
     }
 

src/java/org/apache/cassandra/db/compaction/AbstractCompactionStrategy.java

@@ -299,11 +299,6 @@ public abstract class AbstractCompactionStrategy
         return new ScannerList(scanners);
     }
 
-    public boolean shouldDefragment()
-    {
-        return false;
-    }
-
     public String getName()
     {
         return getClass().getSimpleName();

src/java/org/apache/cassandra/db/compaction/CompactionStrategyManager.java

@@ -235,12 +235,6 @@ public class CompactionStrategyManager implements INotificationConsumer
         return res;
     }
 
-    public boolean shouldDefragment()
-    {
-        assert repaired.getClass().equals(unrepaired.getClass());
-        return repaired.shouldDefragment();
-    }
-
     public Directories getDirectories()
     {
         assert repaired.getClass().equals(unrepaired.getClass());

src/java/org/apache/cassandra/db/compaction/SizeTieredCompactionStrategy.java

@@ -309,12 +309,6 @@ public class SizeTieredCompactionStrategy extends AbstractCompactionStrategy
         return uncheckedOptions;
     }
 
-    @Override
-    public boolean shouldDefragment()
-    {
-        return true;
-    }
-
     @Override
     public synchronized void addSSTable(SSTableReader added)
     {