HDFS-5919. FileJournalManager doesn't purge empty and corrupt inprogress edits files (vinayakumarb)

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/trunk@1613355 13f79535-47bb-0310-9956-ffa450edef68

hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt

@@ -370,6 +370,9 @@ Release 2.6.0 - UNRELEASED
     HDFS-6715. Webhdfs wont fail over when it gets java.io.IOException: Namenode
     is in startup mode. (jing9)
 
+    HDFS-5919. FileJournalManager doesn't purge empty and corrupt inprogress edits
+    files (vinayakumarb)
+
 Release 2.5.0 - UNRELEASED
 
   INCOMPATIBLE CHANGES

hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/namenode/FileJournalManager.java

@@ -71,6 +71,8 @@ public class FileJournalManager implements JournalManager {
     NameNodeFile.EDITS.getName() + "_(\\d+)-(\\d+)");
   private static final Pattern EDITS_INPROGRESS_REGEX = Pattern.compile(
     NameNodeFile.EDITS_INPROGRESS.getName() + "_(\\d+)");
+  private static final Pattern EDITS_INPROGRESS_STALE_REGEX = Pattern.compile(
+      NameNodeFile.EDITS_INPROGRESS.getName() + "_(\\d+).*(\\S+)");
 
   private File currentInProgress = null;
 
@@ -162,8 +164,7 @@ public class FileJournalManager implements JournalManager {
       throws IOException {
     LOG.info("Purging logs older than " + minTxIdToKeep);
     File[] files = FileUtil.listFiles(sd.getCurrentDir());
-    List<EditLogFile> editLogs = 
-      FileJournalManager.matchEditLogs(files);
+    List<EditLogFile> editLogs = matchEditLogs(files, true);
     for (EditLogFile log : editLogs) {
       if (log.getFirstTxId() < minTxIdToKeep &&
           log.getLastTxId() < minTxIdToKeep) {
@@ -244,8 +245,13 @@ public class FileJournalManager implements JournalManager {
   public static List<EditLogFile> matchEditLogs(File logDir) throws IOException {
     return matchEditLogs(FileUtil.listFiles(logDir));
   }
-  
+
   static List<EditLogFile> matchEditLogs(File[] filesInStorage) {
+    return matchEditLogs(filesInStorage, false);
+  }
+
+  private static List<EditLogFile> matchEditLogs(File[] filesInStorage,
+      boolean forPurging) {
     List<EditLogFile> ret = Lists.newArrayList();
     for (File f : filesInStorage) {
       String name = f.getName();
@@ -256,6 +262,7 @@ public class FileJournalManager implements JournalManager {
           long startTxId = Long.parseLong(editsMatch.group(1));
           long endTxId = Long.parseLong(editsMatch.group(2));
           ret.add(new EditLogFile(f, startTxId, endTxId));
+          continue;
         } catch (NumberFormatException nfe) {
           LOG.error("Edits file " + f + " has improperly formatted " +
                     "transaction ID");
@@ -270,12 +277,30 @@ public class FileJournalManager implements JournalManager {
           long startTxId = Long.parseLong(inProgressEditsMatch.group(1));
           ret.add(
               new EditLogFile(f, startTxId, HdfsConstants.INVALID_TXID, true));
+          continue;
         } catch (NumberFormatException nfe) {
           LOG.error("In-progress edits file " + f + " has improperly " +
                     "formatted transaction ID");
           // skip
         }
       }
+      if (forPurging) {
+        // Check for in-progress stale edits
+        Matcher staleInprogressEditsMatch = EDITS_INPROGRESS_STALE_REGEX
+            .matcher(name);
+        if (staleInprogressEditsMatch.matches()) {
+          try {
+            long startTxId = Long.valueOf(staleInprogressEditsMatch.group(1));
+            ret.add(new EditLogFile(f, startTxId, HdfsConstants.INVALID_TXID,
+                true));
+            continue;
+          } catch (NumberFormatException nfe) {
+            LOG.error("In-progress stale edits file " + f + " has improperly "
+                + "formatted transaction ID");
+            // skip
+          }
+        }
+      }
     }
     return ret;
   }

hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/server/namenode/TestNNStorageRetentionManager.java

@@ -212,18 +212,25 @@ public class TestNNStorageRetentionManager {
     tc.addImage("/foo1/current/" + getImageFileName(300), false);
     tc.addImage("/foo1/current/" + getImageFileName(400), false);
 
+    // Segments containing txns upto txId 250 are extra and should be purged.
     tc.addLog("/foo2/current/" + getFinalizedEditsFileName(1, 100), true);
-    // Without lowering the max segments to retain, we'd retain all segments
-    // going back to txid 150 (300 - 150).
     tc.addLog("/foo2/current/" + getFinalizedEditsFileName(101, 175), true);
+    tc.addLog("/foo2/current/" + getInProgressEditsFileName(176) + ".empty",
+        true);
     tc.addLog("/foo2/current/" + getFinalizedEditsFileName(176, 200), true);
     tc.addLog("/foo2/current/" + getFinalizedEditsFileName(201, 225), true);
+    tc.addLog("/foo2/current/" + getInProgressEditsFileName(226) + ".corrupt",
+        true);
     tc.addLog("/foo2/current/" + getFinalizedEditsFileName(226, 240), true);
     // Only retain 2 extra segments. The 301-350 and 351-400 segments are
     // considered required, not extra.
     tc.addLog("/foo2/current/" + getFinalizedEditsFileName(241, 275), false);
     tc.addLog("/foo2/current/" + getFinalizedEditsFileName(276, 300), false);
+    tc.addLog("/foo2/current/" + getInProgressEditsFileName(301) + ".empty",
+        false);
     tc.addLog("/foo2/current/" + getFinalizedEditsFileName(301, 350), false);
+    tc.addLog("/foo2/current/" + getInProgressEditsFileName(351) + ".corrupt",
+        false);
     tc.addLog("/foo2/current/" + getFinalizedEditsFileName(351, 400), false);
     tc.addLog("/foo2/current/" + getInProgressEditsFileName(401), false);
     runTest(tc);