HDFS-7180. NFSv3 gateway frequently gets stuck due to GC. Contributed by Brandon Li

This commit is contained in:
Brandon Li 2014-10-22 21:27:01 -07:00
parent b94b8b30f2
commit d71d40a63d
6 changed files with 73 additions and 50 deletions

View File

@ -228,7 +228,7 @@ public void onRemoval(
RemovalNotification<DFSInputStreamCaheKey, FSDataInputStream> notification) {
try {
notification.getValue().close();
} catch (IOException e) {
} catch (IOException ignored) {
}
}
};

View File

@ -175,7 +175,10 @@ public String toString() {
private volatile boolean enabledDump;
private FileOutputStream dumpOut;
/** Tracks the data buffered in memory related to non sequential writes */
private AtomicLong nonSequentialWriteInMemory;
private RandomAccessFile raf;
private final String dumpFilePath;
private Daemon dumpThread;
@ -205,7 +208,7 @@ boolean hasPendingWork() {
return (pendingWrites.size() != 0 || pendingCommits.size() != 0);
}
// Increase or decrease the memory occupation of non-sequential writes
/** Increase or decrease the memory occupation of non-sequential writes */
private long updateNonSequentialWriteInMemory(long count) {
long newValue = nonSequentialWriteInMemory.addAndGet(count);
if (LOG.isDebugEnabled()) {
@ -214,8 +217,8 @@ private long updateNonSequentialWriteInMemory(long count) {
}
Preconditions.checkState(newValue >= 0,
"nonSequentialWriteInMemory is negative after update with count "
+ count);
"nonSequentialWriteInMemory is negative " + newValue
+ " after update with count " + count);
return newValue;
}
@ -248,7 +251,7 @@ private long updateNonSequentialWriteInMemory(long count) {
nonSequentialWriteInMemory = new AtomicLong(0);
this.dumpFilePath = dumpFilePath;
enabledDump = dumpFilePath == null ? false: true;
enabledDump = dumpFilePath != null;
nextOffset = new AtomicLong();
nextOffset.set(latestAttr.getSize());
assert(nextOffset.get() == this.fos.getPos());
@ -269,7 +272,7 @@ private long getFlushedOffset() {
}
// Check if need to dump the new writes
private void checkDump() {
private void waitForDump() {
if (!enabledDump) {
if (LOG.isDebugEnabled()) {
LOG.debug("Do nothing, dump is disabled.");
@ -294,6 +297,14 @@ private void checkDump() {
this.notifyAll();
}
}
while (nonSequentialWriteInMemory.get() >= DUMP_WRITE_WATER_MARK) {
try {
this.wait();
} catch (InterruptedException ignored) {
}
}
}
}
@ -380,6 +391,7 @@ public void run() {
}
synchronized (OpenFileCtx.this) {
if (nonSequentialWriteInMemory.get() < DUMP_WRITE_WATER_MARK) {
OpenFileCtx.this.notifyAll();
try {
OpenFileCtx.this.wait();
if (LOG.isDebugEnabled()) {
@ -396,8 +408,13 @@ public void run() {
+ " enabledDump: " + enabledDump);
}
} catch (Throwable t) {
// unblock threads with new request
synchronized (OpenFileCtx.this) {
OpenFileCtx.this.notifyAll();
}
LOG.info("Dumper get Throwable: " + t + ". dumpFilePath: "
+ OpenFileCtx.this.dumpFilePath, t);
activeState = false;
}
}
}
@ -561,10 +578,15 @@ private synchronized WriteCtx addWritesToCache(WRITE3Request request,
// check if there is a WriteCtx with the same range in pendingWrites
WriteCtx oldWriteCtx = checkRepeatedWriteRequest(request, channel, xid);
if (oldWriteCtx == null) {
addWrite(writeCtx);
pendingWrites.put(new OffsetRange(offset, offset + count), writeCtx);
if (LOG.isDebugEnabled()) {
LOG.debug("New write buffered with xid " + xid + " nextOffset "
+ cachedOffset + " req offset=" + offset + " mapsize="
+ pendingWrites.size());
}
} else {
LOG.warn("Got a repeated request, same range, with xid:"
+ writeCtx.getXid());
LOG.warn("Got a repeated request, same range, with xid:" + xid
+ " nextOffset " + +cachedOffset + " req offset=" + offset);
}
return writeCtx;
}
@ -646,7 +668,7 @@ private void receivedNewWriteInternal(DFSClient dfsClient,
boolean startWriting = checkAndStartWrite(asyncDataService, writeCtx);
if (!startWriting) {
// offset > nextOffset. check if we need to dump data
checkDump();
waitForDump();
// In test, noticed some Linux client sends a batch (e.g., 1MB)
// of reordered writes and won't send more writes until it gets
@ -681,7 +703,7 @@ private void receivedNewWriteInternal(DFSClient dfsClient,
private WRITE3Response processPerfectOverWrite(DFSClient dfsClient,
long offset, int count, WriteStableHow stableHow, byte[] data,
String path, WccData wccData, IdUserGroup iug) {
WRITE3Response response = null;
WRITE3Response response;
// Read the content back
byte[] readbuffer = new byte[count];
@ -882,13 +904,6 @@ synchronized COMMIT_STATUS checkCommitInternal(long commitOffset,
return COMMIT_STATUS.COMMIT_WAIT;
}
private void addWrite(WriteCtx writeCtx) {
long offset = writeCtx.getOffset();
int count = writeCtx.getCount();
// For the offset range (min, max), min is inclusive, and max is exclusive
pendingWrites.put(new OffsetRange(offset, offset + count), writeCtx);
}
/**
* Check stream status to decide if it should be closed
* @return true, remove stream; false, keep stream
@ -1181,7 +1196,7 @@ synchronized void cleanup() {
dumpThread.interrupt();
try {
dumpThread.join(3000);
} catch (InterruptedException e) {
} catch (InterruptedException ignored) {
}
}

View File

@ -220,12 +220,12 @@ void cleanAll() {
void shutdown() {
// stop the dump thread
if (streamMonitor != null && streamMonitor.isAlive()) {
if (streamMonitor.isAlive()) {
streamMonitor.shouldRun(false);
streamMonitor.interrupt();
try {
streamMonitor.join(3000);
} catch (InterruptedException e) {
} catch (InterruptedException ignored) {
}
}

View File

@ -282,7 +282,7 @@ GETATTR3Response getattr(XDR xdr, SecurityHandler securityHandler,
return response;
}
GETATTR3Request request = null;
GETATTR3Request request;
try {
request = GETATTR3Request.deserialize(xdr);
} catch (IOException e) {
@ -374,7 +374,7 @@ SETATTR3Response setattr(XDR xdr, SecurityHandler securityHandler,
return response;
}
SETATTR3Request request = null;
SETATTR3Request request;
try {
request = SETATTR3Request.deserialize(xdr);
} catch (IOException e) {
@ -459,7 +459,7 @@ LOOKUP3Response lookup(XDR xdr, SecurityHandler securityHandler,
return response;
}
LOOKUP3Request request = null;
LOOKUP3Request request;
try {
request = LOOKUP3Request.deserialize(xdr);
} catch (IOException e) {
@ -527,7 +527,7 @@ ACCESS3Response access(XDR xdr, SecurityHandler securityHandler,
return response;
}
ACCESS3Request request = null;
ACCESS3Request request;
try {
request = ACCESS3Request.deserialize(xdr);
} catch (IOException e) {
@ -594,7 +594,7 @@ READLINK3Response readlink(XDR xdr, SecurityHandler securityHandler,
return response;
}
READLINK3Request request = null;
READLINK3Request request;
try {
request = READLINK3Request.deserialize(xdr);
@ -668,7 +668,7 @@ READ3Response read(XDR xdr, SecurityHandler securityHandler,
return response;
}
READ3Request request = null;
READ3Request request;
try {
request = READ3Request.deserialize(xdr);
@ -710,7 +710,7 @@ READ3Response read(XDR xdr, SecurityHandler securityHandler,
securityHandler.getUid(), securityHandler.getGid(),
securityHandler.getAuxGids(), attrs);
if ((access & Nfs3Constant.ACCESS3_READ) != 0) {
eof = offset < attrs.getSize() ? false : true;
eof = offset >= attrs.getSize();
return new READ3Response(Nfs3Status.NFS3_OK, attrs, 0, eof,
ByteBuffer.wrap(new byte[0]));
} else {
@ -749,7 +749,7 @@ READ3Response read(XDR xdr, SecurityHandler securityHandler,
} catch (IOException e) {
// TODO: A cleaner way is to throw a new type of exception
// which requires incompatible changes.
if (e.getMessage() == "Stream closed") {
if (e.getMessage().equals("Stream closed")) {
clientCache.invalidateDfsInputStream(userName,
Nfs3Utils.getFileIdPath(handle));
continue;
@ -769,7 +769,7 @@ READ3Response read(XDR xdr, SecurityHandler securityHandler,
if (readCount < 0) {
readCount = 0;
}
eof = (offset + readCount) < attrs.getSize() ? false : true;
eof = (offset + readCount) >= attrs.getSize();
return new READ3Response(Nfs3Status.NFS3_OK, attrs, readCount, eof,
ByteBuffer.wrap(readbuffer));
@ -801,7 +801,7 @@ WRITE3Response write(XDR xdr, Channel channel, int xid,
return response;
}
WRITE3Request request = null;
WRITE3Request request;
try {
request = WRITE3Request.deserialize(xdr);
@ -883,7 +883,7 @@ CREATE3Response create(XDR xdr, SecurityHandler securityHandler,
return response;
}
CREATE3Request request = null;
CREATE3Request request;
try {
request = CREATE3Request.deserialize(xdr);
@ -1017,7 +1017,7 @@ MKDIR3Response mkdir(XDR xdr, SecurityHandler securityHandler,
return response;
}
MKDIR3Request request = null;
MKDIR3Request request;
try {
request = MKDIR3Request.deserialize(xdr);
@ -1114,7 +1114,7 @@ REMOVE3Response remove(XDR xdr, SecurityHandler securityHandler,
return response;
}
REMOVE3Request request = null;
REMOVE3Request request;
try {
request = REMOVE3Request.deserialize(xdr);
} catch (IOException e) {
@ -1194,7 +1194,7 @@ RMDIR3Response rmdir(XDR xdr, SecurityHandler securityHandler,
return response;
}
RMDIR3Request request = null;
RMDIR3Request request;
try {
request = RMDIR3Request.deserialize(xdr);
} catch (IOException e) {
@ -1375,7 +1375,7 @@ SYMLINK3Response symlink(XDR xdr, SecurityHandler securityHandler,
return response;
}
SYMLINK3Request request = null;
SYMLINK3Request request;
try {
request = SYMLINK3Request.deserialize(xdr);
} catch (IOException e) {
@ -1431,7 +1431,7 @@ public READDIR3Response link(XDR xdr, RpcInfo info) {
*/
private DirectoryListing listPaths(DFSClient dfsClient, String dirFileIdPath,
byte[] startAfter) throws IOException {
DirectoryListing dlisting = null;
DirectoryListing dlisting;
try {
dlisting = dfsClient.listPaths(dirFileIdPath, startAfter);
} catch (RemoteException e) {
@ -1468,7 +1468,7 @@ public READDIR3Response readdir(XDR xdr, SecurityHandler securityHandler,
return response;
}
READDIR3Request request = null;
READDIR3Request request;
try {
request = READDIR3Request.deserialize(xdr);
} catch (IOException e) {
@ -1492,9 +1492,9 @@ public READDIR3Response readdir(XDR xdr, SecurityHandler securityHandler,
+ cookie + " count: " + count);
}
HdfsFileStatus dirStatus = null;
DirectoryListing dlisting = null;
Nfs3FileAttributes postOpAttr = null;
HdfsFileStatus dirStatus;
DirectoryListing dlisting;
Nfs3FileAttributes postOpAttr;
long dotdotFileId = 0;
try {
String dirFileIdPath = Nfs3Utils.getFileIdPath(handle);
@ -1657,8 +1657,8 @@ READDIRPLUS3Response readdirplus(XDR xdr, SecurityHandler securityHandler,
}
HdfsFileStatus dirStatus;
DirectoryListing dlisting = null;
Nfs3FileAttributes postOpDirAttr = null;
DirectoryListing dlisting;
Nfs3FileAttributes postOpDirAttr;
long dotdotFileId = 0;
HdfsFileStatus dotdotStatus = null;
try {
@ -1803,7 +1803,7 @@ FSSTAT3Response fsstat(XDR xdr, SecurityHandler securityHandler,
return response;
}
FSSTAT3Request request = null;
FSSTAT3Request request;
try {
request = FSSTAT3Request.deserialize(xdr);
} catch (IOException e) {
@ -1877,7 +1877,7 @@ FSINFO3Response fsinfo(XDR xdr, SecurityHandler securityHandler,
return response;
}
FSINFO3Request request = null;
FSINFO3Request request;
try {
request = FSINFO3Request.deserialize(xdr);
} catch (IOException e) {
@ -1941,7 +1941,7 @@ PATHCONF3Response pathconf(XDR xdr, SecurityHandler securityHandler,
return response;
}
PATHCONF3Request request = null;
PATHCONF3Request request;
try {
request = PATHCONF3Request.deserialize(xdr);
} catch (IOException e) {
@ -1992,7 +1992,7 @@ COMMIT3Response commit(XDR xdr, Channel channel, int xid,
return response;
}
COMMIT3Request request = null;
COMMIT3Request request;
try {
request = COMMIT3Request.deserialize(xdr);
} catch (IOException e) {

View File

@ -47,14 +47,20 @@ class WriteCtx {
public static enum DataState {
ALLOW_DUMP,
NO_DUMP,
DUMPED;
DUMPED
}
private final FileHandle handle;
private final long offset;
private final int count;
//Only needed for overlapped write, referring OpenFileCtx.addWritesToCache()
/**
* Some clients can send a write that includes previously written data along
* with new data. In such case the write request is changed to write from only
* the new data. {@code originalCount} tracks the number of bytes sent in the
* request before it was modified to write only the new data.
* @see OpenFileCtx#addWritesToCache for more details
*/
private final int originalCount;
public static final int INVALID_ORIGINAL_COUNT = -1;
@ -173,7 +179,7 @@ private void loadData() throws IOException {
public void writeData(HdfsDataOutputStream fos) throws IOException {
Preconditions.checkState(fos != null);
ByteBuffer dataBuffer = null;
ByteBuffer dataBuffer;
try {
dataBuffer = getData();
} catch (Exception e1) {

View File

@ -896,6 +896,8 @@ Release 2.6.0 - UNRELEASED
HDFS-7215.Add JvmPauseMonitor to NFS gateway (brandonli)
HDFS-7180. NFSv3 gateway frequently gets stuck due to GC (brandonli)
BREAKDOWN OF HDFS-6134 AND HADOOP-10150 SUBTASKS AND RELATED JIRAS
HDFS-6387. HDFS CLI admin tool for creating & deleting an