HADOOP-2029 TestLogRolling fails too often in patch and nightlies

This is second commit against this issue.


git-svn-id: https://svn.apache.org/repos/asf/lucene/hadoop/trunk/src/contrib/hbase@583963 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Michael Stack 2007-10-11 21:49:40 +00:00
parent 00f32814d1
commit 8b387823be
3 changed files with 32 additions and 34 deletions

View File

@ -319,8 +319,10 @@ public class HLog implements HConstants {
// Now remove old log files (if any)
LOG.debug("Found " + sequenceNumbers.size() + " logs to remove " +
"using oldest outstanding seqnum of " + oldestOutstandingSeqNum);
for (Long seq : sequenceNumbers) {
deleteLogFile(this.outputfiles.remove(seq), seq);
if (sequenceNumbers.size() > 0) {
for (Long seq : sequenceNumbers) {
deleteLogFile(this.outputfiles.remove(seq), seq);
}
}
}
}

View File

@ -1104,10 +1104,9 @@ public class HRegionServer implements HConstants, HRegionInterface, Runnable {
/** {@inheritDoc} */
public void batchUpdate(Text regionName, long timestamp, BatchUpdate b)
throws IOException {
throws IOException {
checkOpen();
requestCount.incrementAndGet();
this.requestCount.incrementAndGet();
// If timestamp == LATEST_TIMESTAMP and we have deletes, then they need
// special treatment. For these we need to first find the latest cell so
// when we write the delete, we write it with the latest cells' timestamp
@ -1116,7 +1115,7 @@ public class HRegionServer implements HConstants, HRegionInterface, Runnable {
List<Text> deletes = null;
try {
long lockid = startUpdate(regionName, b.getRow());
for(BatchOperation op: b) {
for (BatchOperation op: b) {
switch(op.getOp()) {
case PUT:
put(regionName, lockid, op.getColumn(), op.getValue());
@ -1239,20 +1238,19 @@ public class HRegionServer implements HConstants, HRegionInterface, Runnable {
//
protected long startUpdate(Text regionName, Text row) throws IOException {
HRegion region = getRegion(regionName);
HRegion region = getRegion(regionName, false);
return region.startUpdate(row);
}
protected void put(final Text regionName, final long lockid,
final Text column, final byte [] val) throws IOException {
final Text column, final byte [] val)
throws IOException {
HRegion region = getRegion(regionName, true);
region.put(lockid, column, val);
}
protected void delete(Text regionName, long lockid, Text column)
throws IOException {
throws IOException {
HRegion region = getRegion(regionName);
region.delete(lockid, column);
}
@ -1297,8 +1295,7 @@ public class HRegionServer implements HConstants, HRegionInterface, Runnable {
* @throws NotServingRegionException
*/
protected HRegion getRegion(final Text regionName)
throws NotServingRegionException {
throws NotServingRegionException {
return getRegion(regionName, false);
}
@ -1311,8 +1308,8 @@ public class HRegionServer implements HConstants, HRegionInterface, Runnable {
* @throws NotServingRegionException
*/
protected HRegion getRegion(final Text regionName,
final boolean checkRetiringRegions) throws NotServingRegionException {
final boolean checkRetiringRegions)
throws NotServingRegionException {
HRegion region = null;
this.lock.readLock().lock();
try {
@ -1342,7 +1339,7 @@ public class HRegionServer implements HConstants, HRegionInterface, Runnable {
* @throws IOException
*/
private void checkOpen() throws IOException {
if (stopRequested.get() || abortRequested) {
if (this.stopRequested.get() || this.abortRequested) {
throw new IOException("Server not running");
}
if (!fsOk) {

View File

@ -53,8 +53,8 @@ public class MultiRegionTable extends HBaseTestCase {
*/
public static void makeMultiRegionTable(Configuration conf,
MiniHBaseCluster cluster, FileSystem localFs, String tableName,
String columnName) throws IOException {
String columnName)
throws IOException {
final int retries = 10;
final long waitTime =
conf.getLong("hbase.master.meta.thread.rescanfrequency", 10L * 1000L);
@ -62,7 +62,6 @@ public class MultiRegionTable extends HBaseTestCase {
// This size should make it so we always split using the addContent
// below. After adding all data, the first region is 1.3M. Should
// set max filesize to be <= 1M.
assertTrue(conf.getLong("hbase.hregion.max.filesize",
HConstants.DEFAULT_MAX_FILE_SIZE) <= 1024 * 1024);
@ -72,29 +71,25 @@ public class MultiRegionTable extends HBaseTestCase {
Path d = fs.makeQualified(new Path(conf.get(HConstants.HBASE_DIR)));
// Get connection on the meta table and get count of rows.
HTable meta = new HTable(conf, HConstants.META_TABLE_NAME);
int count = count(meta, HConstants.COLUMN_FAMILY_STR);
int count = count(meta, tableName);
HTable t = new HTable(conf, new Text(tableName));
addContent(new HTableIncommon(t), columnName);
// All is running in the one JVM so I should be able to get the single
// region instance and bring on a split.
HRegionInfo hri =
t.getRegionLocation(HConstants.EMPTY_START_ROW).getRegionInfo();
HRegion r = cluster.regionThreads.get(0).getRegionServer().
onlineRegions.get(hri.getRegionName());
onlineRegions.get(hri.getRegionName());
// Flush will provoke a split next time the split-checker thread runs.
r.flushcache(false);
// Now, wait until split makes it into the meta table.
int oldCount = count;
for (int i = 0; i < retries; i++) {
count = count(meta, HConstants.COLUMN_FAMILY_STR);
count = count(meta, tableName);
if (count > oldCount) {
break;
}
@ -111,7 +106,6 @@ public class MultiRegionTable extends HBaseTestCase {
// Get info on the parent from the meta table. Pass in 'hri'. Its the
// region we have been dealing with up to this. Its the parent of the
// region split.
Map<Text, byte []> data = getSplitParentInfo(meta, hri);
HRegionInfo parent =
Writables.getHRegionInfoOrNull(data.get(HConstants.COL_REGIONINFO));
@ -185,24 +179,28 @@ public class MultiRegionTable extends HBaseTestCase {
}
/*
* Count of rows in table for given column.
* Count of regions in passed meta table.
* @param t
* @param column
* @return
* @throws IOException
*/
private static int count(final HTable t, final String column)
private static int count(final HTable t, final String tableName)
throws IOException {
int size = 0;
Text [] cols = new Text[] {new Text(column)};
Text [] cols = new Text[] {HConstants.COLUMN_FAMILY};
HScannerInterface s = t.obtainScanner(cols, HConstants.EMPTY_START_ROW,
System.currentTimeMillis(), null);
try {
HStoreKey curKey = new HStoreKey();
TreeMap<Text, byte []> curVals = new TreeMap<Text, byte []>();
while(s.next(curKey, curVals)) {
size++;
HRegionInfo hri = Writables.
getHRegionInfoOrNull(curVals.get(HConstants.COL_REGIONINFO));
if (hri.getTableDesc().getName().toString().equals(tableName)) {
size++;
}
}
return size;
} finally {
@ -214,8 +212,8 @@ public class MultiRegionTable extends HBaseTestCase {
* @return Return row info for passed in region or null if not found in scan.
*/
private static Map<Text, byte []> getSplitParentInfo(final HTable t,
final HRegionInfo parent) throws IOException {
final HRegionInfo parent)
throws IOException {
HScannerInterface s = t.obtainScanner(HConstants.COLUMN_FAMILY_ARRAY,
HConstants.EMPTY_START_ROW, System.currentTimeMillis(), null);
try {
@ -223,10 +221,11 @@ public class MultiRegionTable extends HBaseTestCase {
TreeMap<Text, byte []> curVals = new TreeMap<Text, byte []>();
while(s.next(curKey, curVals)) {
HRegionInfo hri = Writables.
getHRegionInfoOrNull(curVals.get(HConstants.COL_REGIONINFO));
getHRegionInfoOrNull(curVals.get(HConstants.COL_REGIONINFO));
if (hri == null) {
continue;
}
// Make sure I get the parent.
if (hri.getRegionName().toString().
equals(parent.getRegionName().toString())) {
return curVals;