HBASE-4789 On split, parent region is sticking around in oldest sequenceid to region map though not online; we don't cleanup WALs

git-svn-id: https://svn.apache.org/repos/asf/hbase/trunk@1204034 13f79535-47bb-0310-9956-ffa450edef68

src/main/java/org/apache/hadoop/hbase/regionserver/HRegion.java

@@ -1286,7 +1286,7 @@ public class HRegion implements HeapSize { // , Writable{
       w = mvcc.beginMemstoreInsert();
       mvcc.advanceMemstore(w);
 
-      sequenceId = (wal == null)? myseqid :
+      sequenceId = (wal == null)? myseqid:
         wal.startCacheFlush(this.regionInfo.getEncodedNameAsBytes());
       completeSequenceId = this.getCompleteCacheFlushSequenceId(sequenceId);
 
@@ -1302,7 +1302,7 @@ public class HRegion implements HeapSize { // , Writable{
       this.updatesLock.writeLock().unlock();
     }
     status.setStatus("Waiting for mvcc");
-    LOG.debug("Finished snapshotting, commencing waiting for mvcc");
+    LOG.debug("Finished snapshotting " + this + ", commencing wait for mvcc");
 
     // wait for all in-progress transactions to commit to HLog before
     // we can start the flush. This prevents

src/main/java/org/apache/hadoop/hbase/regionserver/LogRoller.java

@@ -115,11 +115,11 @@ class LogRoller extends HasThread implements WALActionsListener {
   }
 
   /**
-   * @param region Encoded name of region to flush.
+   * @param encodedRegionName Encoded name of region to flush.
    */
-  private void scheduleFlush(final byte [] region) {
+  private void scheduleFlush(final byte [] encodedRegionName) {
     boolean scheduled = false;
-    HRegion r = this.services.getFromOnlineRegions(Bytes.toString(region));
+    HRegion r = this.services.getFromOnlineRegions(Bytes.toString(encodedRegionName));
     FlushRequester requester = null;
     if (r != null) {
       requester = this.services.getFlushRequester();
@@ -129,8 +129,9 @@ class LogRoller extends HasThread implements WALActionsListener {
       }
     }
     if (!scheduled) {
-    LOG.warn("Failed to schedule flush of " +
-      Bytes.toString(region) + "r=" + r + ", requester=" + requester);
+      LOG.warn("Failed to schedule flush of " +
+        Bytes.toString(encodedRegionName) + ", region=" + r + ", requester=" +
+        requester);
     }
   }
 

src/main/java/org/apache/hadoop/hbase/regionserver/wal/HLog.java

@@ -187,8 +187,8 @@ public class HLog implements Syncable {
     Collections.synchronizedSortedMap(new TreeMap<Long, Path>());
 
   /*
-   * Map of regions to most recent sequence/edit id in their memstore.
-   * Key is encoded region name.
+   * Map of encoded region names to their most recent sequence/edit id in their
+   * memstore.
    */
   private final ConcurrentSkipListMap<byte [], Long> lastSeqWritten =
     new ConcurrentSkipListMap<byte [], Long>(Bytes.BYTES_COMPARATOR);
@@ -745,9 +745,8 @@ public class HLog implements Syncable {
     // If too many log files, figure which regions we need to flush.
     // Array is an array of encoded region names.
     byte [][] regions = null;
-    int logCount = this.outputfiles.size();
-    if (logCount > this.maxLogs && this.outputfiles != null &&
-        this.outputfiles.size() > 0) {
+    int logCount = this.outputfiles == null? 0: this.outputfiles.size();
+    if (logCount > this.maxLogs && logCount > 0) {
       // This is an array of encoded region names.
       regions = findMemstoresWithEditsEqualOrOlderThan(this.outputfiles.firstKey(),
         this.lastSeqWritten);
@@ -769,7 +768,7 @@ public class HLog implements Syncable {
    * Return regions (memstores) that have edits that are equal or less than
    * the passed <code>oldestWALseqid</code>.
    * @param oldestWALseqid
-   * @param regionsToSeqids
+   * @param regionsToSeqids Encoded region names to sequence ids
    * @return All regions whose seqid is < than <code>oldestWALseqid</code> (Not
    * necessarily in order).  Null if no regions found.
    */
@@ -780,6 +779,7 @@ public class HLog implements Syncable {
     for (Map.Entry<byte [], Long> e: regionsToSeqids.entrySet()) {
       if (e.getValue().longValue() <= oldestWALseqid) {
         if (regions == null) regions = new ArrayList<byte []>();
+        // Key is encoded region name.
         regions.add(e.getKey());
       }
     }
@@ -802,6 +802,7 @@ public class HLog implements Syncable {
     byte [] oldestRegion = null;
     for (Map.Entry<byte [], Long> e: this.lastSeqWritten.entrySet()) {
       if (e.getValue().longValue() == oldestOutstandingSeqNum.longValue()) {
+        // Key is encoded region name.
         oldestRegion = e.getKey();
         break;
       }
@@ -1092,9 +1093,9 @@ public class HLog implements Syncable {
         // is greater than or equal to the value in lastSeqWritten.
         // Use encoded name.  Its shorter, guaranteed unique and a subset of
         // actual  name.
-        byte [] hriKey = info.getEncodedNameAsBytes();
-        this.lastSeqWritten.putIfAbsent(hriKey, seqNum);
-        HLogKey logKey = makeKey(hriKey, tableName, seqNum, now, clusterId);
+        byte [] encodedRegionName = info.getEncodedNameAsBytes();
+        this.lastSeqWritten.putIfAbsent(encodedRegionName, seqNum);
+        HLogKey logKey = makeKey(encodedRegionName, tableName, seqNum, now, clusterId);
         doWrite(info, logKey, edits, htd);
         this.numEntries.incrementAndGet();
         txid = this.unflushedEntries.incrementAndGet();
@@ -1537,6 +1538,11 @@ public class HLog implements Syncable {
       // Cleaning up of lastSeqWritten is in the finally clause because we
       // don't want to confuse getOldestOutstandingSeqNum()
       this.lastSeqWritten.remove(getSnapshotName(encodedRegionName));
+      Long l = this.lastSeqWritten.remove(encodedRegionName);
+      if (l != null) {
+        LOG.warn("Why is there a raw encodedRegionName in lastSeqWritten? name=" +
+          Bytes.toString(encodedRegionName) + ", seqid=" + l);
+       }
       this.cacheFlushLock.unlock();
     }
   }