avoid writing empty rows when scrubbing tombstoned rows

patch by jbellis; reviewed by slebresne for CASSANDRA-2296 git-svn-id: https://svn.apache.org/repos/asf/cassandra/branches/cassandra-0.7@1079816 13f79535-47bb-0310-9956-ffa450edef68
apache · Mar 9, 2011 · 0087c38 · 0087c38
1 parent 942a432
commit 0087c38
Show file tree

Hide file tree

Showing 2 changed files with 23 additions and 6 deletions.
diff --git a/CHANGES.txt b/CHANGES.txt
@@ -13,6 +13,7 @@
    (CASSANDRA-2259)
  * move sample schema from cassandra.yaml to schema-sample.txt,
    a cli scripts (CASSANDRA-2007)
+ * avoid writing empty rows when scrubbing tombstoned rows (CASSANDRA-2296)
 
 
 0.7.3

diff --git a/src/java/org/apache/cassandra/db/CompactionManager.java b/src/java/org/apache/cassandra/db/CompactionManager.java
@@ -534,7 +534,7 @@ private void doScrub(ColumnFamilyStore cfs) throws IOException
 
             SSTableWriter writer = maybeCreateWriter(cfs, compactionFileLocation, expectedBloomFilterSize, null);
             executor.beginCompaction(cfs.columnFamily, new ScrubInfo(dataFile, sstable));
-            int goodRows = 0, badRows = 0;
+            int goodRows = 0, badRows = 0, emptyRows = 0;
 
             while (!dataFile.isEOF())
             {
@@ -588,8 +588,16 @@ private void doScrub(ColumnFamilyStore cfs) throws IOException
                     if (dataSize > dataFile.length())
                         throw new IOError(new IOException("Impossible row size " + dataSize));
                     SSTableIdentityIterator row = new SSTableIdentityIterator(sstable, dataFile, key, dataStart, dataSize, true);
-                    writer.append(getCompactedRow(row, cfs, sstable.descriptor, true));
-                    goodRows++;
+                    AbstractCompactedRow compactedRow = getCompactedRow(row, cfs, sstable.descriptor, true);
+                    if (compactedRow.isEmpty())
+                    {
+                        emptyRows++;
+                    }
+                    else
+                    {
+                        writer.append(compactedRow);
+                        goodRows++;
+                    }
                     if (!key.key.equals(currentIndexKey) || dataStart != dataStartFromIndex)
                         logger.warn("Row scrubbed successfully but index file contains a different key or row size; consider rebuilding the index as described in http://www.mail-archive.com/user@cassandra.apache.org/msg03325.html");
                 }
@@ -608,8 +616,16 @@ private void doScrub(ColumnFamilyStore cfs) throws IOException
                         try
                         {
                             SSTableIdentityIterator row = new SSTableIdentityIterator(sstable, dataFile, key, dataStartFromIndex, dataSizeFromIndex, true);
-                            writer.append(getCompactedRow(row, cfs, sstable.descriptor, true));
-                            goodRows++;
+                            AbstractCompactedRow compactedRow = getCompactedRow(row, cfs, sstable.descriptor, true);
+                            if (compactedRow.isEmpty())
+                            {
+                                emptyRows++;
+                            }
+                            else
+                            {
+                                writer.append(compactedRow);
+                                goodRows++;
+                            }
                         }
                         catch (Throwable th2)
                         {
@@ -634,7 +650,7 @@ private void doScrub(ColumnFamilyStore cfs) throws IOException
             {
                 SSTableReader newSstable = writer.closeAndOpenReader(sstable.maxDataAge);
                 cfs.replaceCompactedSSTables(Arrays.asList(sstable), Arrays.asList(newSstable));
-                logger.info("Scrub of " + sstable + " complete: " + goodRows + " rows in new sstable");
+                logger.info("Scrub of " + sstable + " complete: " + goodRows + " rows in new sstable and " + emptyRows + " empty (tombstoned) rows dropped");
                 if (badRows > 0)
                     logger.warn("Unable to recover " + badRows + " rows that were skipped.  You can attempt manual recovery from the pre-scrub snapshot.  You can also run nodetool repair to transfer the data from a healthy replica, if any");
             }