Class CommitProcessor
- java.lang.Object
-
- java.lang.Thread
-
- org.apache.zookeeper.server.ZooKeeperThread
-
- org.apache.zookeeper.server.ZooKeeperCriticalThread
-
- org.apache.zookeeper.server.quorum.CommitProcessor
-
- All Implemented Interfaces:
java.lang.Runnable,RequestProcessor
public class CommitProcessor extends ZooKeeperCriticalThread implements RequestProcessor
This RequestProcessor matches the incoming committed requests with the locally submitted requests. The trick is that locally submitted requests that change the state of the system will come back as incoming committed requests, so we need to match them up. Instead of just waiting for the committed requests, we process the uncommitted requests that belong to other sessions. The CommitProcessor is multi-threaded. Communication between threads is handled via queues, atomics, and wait/notifyAll synchronized on the processor. The CommitProcessor acts as a gateway for allowing requests to continue with the remainder of the processing pipeline. It will allow many read requests but only a single write request to be in flight simultaneously, thus ensuring that write requests are processed in transaction id order. - 1 commit processor main thread, which watches the request queues and assigns requests to worker threads based on their sessionId so that read and write requests for a particular session are always assigned to the same thread (and hence are guaranteed to run in order). - 0-N worker threads, which run the rest of the request processor pipeline on the requests. If configured with 0 worker threads, the primary commit processor thread runs the pipeline directly. Typical (default) thread counts are: on a 32 core machine, 1 commit processor thread and 32 worker threads. Multi-threading constraints: - Each session's requests must be processed in order. - Write requests must be processed in zxid order - Must ensure no race condition between writes in one session that would trigger a watch being set by a read request in another session The current implementation solves the third constraint by simply allowing no read requests to be processed in parallel with write requests.
-
-
Nested Class Summary
-
Nested classes/interfaces inherited from class java.lang.Thread
java.lang.Thread.State, java.lang.Thread.UncaughtExceptionHandler
-
Nested classes/interfaces inherited from interface org.apache.zookeeper.server.RequestProcessor
RequestProcessor.RequestProcessorException
-
-
Field Summary
Fields Modifier and Type Field Description protected java.util.concurrent.LinkedBlockingQueue<Request>committedRequestsRequests that have been committed.protected java.util.concurrent.atomic.AtomicIntegernumRequestsProcessingThe number of requests currently being processedprotected java.util.Map<java.lang.Long,java.util.Deque<Request>>pendingRequestsRequests that we are holding until commit comes in.protected java.util.concurrent.LinkedBlockingQueue<Request>queuedRequestsIncoming requests.protected java.util.concurrent.LinkedBlockingQueue<Request>queuedWriteRequestsIncoming requests that are waiting on a commit, contained in order of arrivalprotected booleanstoppedprotected booleanstoppedMainLoopFor testing purposes, we use a separated stopping condition for the outer loop.protected WorkerServiceworkerPoolstatic java.lang.StringZOOKEEPER_COMMIT_PROC_MAX_COMMIT_BATCH_SIZEDefault max commit batch size: 1static java.lang.StringZOOKEEPER_COMMIT_PROC_MAX_READ_BATCH_SIZEDefault max read batch size: -1 to disable the featurestatic java.lang.StringZOOKEEPER_COMMIT_PROC_NUM_WORKER_THREADSDefault: numCoresstatic java.lang.StringZOOKEEPER_COMMIT_PROC_SHUTDOWN_TIMEOUTDefault worker pool shutdown timeout in ms: 5000 (5s)
-
Constructor Summary
Constructors Constructor Description CommitProcessor(RequestProcessor nextProcessor, java.lang.String id, boolean matchSyncs, ZooKeeperServerListener listener)
-
Method Summary
All Methods Static Methods Instance Methods Concrete Methods Modifier and Type Method Description voidcommit(Request request)protected voidendOfIteration()static intgetMaxCommitBatchSize()static intgetMaxReadBatchSize()protected booleanneedCommit(Request request)voidprocessRequest(Request request)voidrun()static voidsetMaxCommitBatchSize(int size)static voidsetMaxReadBatchSize(int size)voidshutdown()voidstart()protected voidwaitForEmptyPool()-
Methods inherited from class org.apache.zookeeper.server.ZooKeeperCriticalThread
handleException
-
Methods inherited from class java.lang.Thread
activeCount, checkAccess, clone, countStackFrames, currentThread, dumpStack, enumerate, getAllStackTraces, getContextClassLoader, getDefaultUncaughtExceptionHandler, getId, getName, getPriority, getStackTrace, getState, getThreadGroup, getUncaughtExceptionHandler, holdsLock, interrupt, interrupted, isAlive, isDaemon, isInterrupted, join, join, join, onSpinWait, resume, setContextClassLoader, setDaemon, setDefaultUncaughtExceptionHandler, setName, setPriority, setUncaughtExceptionHandler, sleep, sleep, stop, suspend, toString, yield
-
-
-
-
Field Detail
-
ZOOKEEPER_COMMIT_PROC_NUM_WORKER_THREADS
public static final java.lang.String ZOOKEEPER_COMMIT_PROC_NUM_WORKER_THREADS
Default: numCores- See Also:
- Constant Field Values
-
ZOOKEEPER_COMMIT_PROC_SHUTDOWN_TIMEOUT
public static final java.lang.String ZOOKEEPER_COMMIT_PROC_SHUTDOWN_TIMEOUT
Default worker pool shutdown timeout in ms: 5000 (5s)- See Also:
- Constant Field Values
-
ZOOKEEPER_COMMIT_PROC_MAX_READ_BATCH_SIZE
public static final java.lang.String ZOOKEEPER_COMMIT_PROC_MAX_READ_BATCH_SIZE
Default max read batch size: -1 to disable the feature- See Also:
- Constant Field Values
-
ZOOKEEPER_COMMIT_PROC_MAX_COMMIT_BATCH_SIZE
public static final java.lang.String ZOOKEEPER_COMMIT_PROC_MAX_COMMIT_BATCH_SIZE
Default max commit batch size: 1- See Also:
- Constant Field Values
-
queuedRequests
protected java.util.concurrent.LinkedBlockingQueue<Request> queuedRequests
Incoming requests.
-
queuedWriteRequests
protected final java.util.concurrent.LinkedBlockingQueue<Request> queuedWriteRequests
Incoming requests that are waiting on a commit, contained in order of arrival
-
committedRequests
protected final java.util.concurrent.LinkedBlockingQueue<Request> committedRequests
Requests that have been committed.
-
pendingRequests
protected final java.util.Map<java.lang.Long,java.util.Deque<Request>> pendingRequests
Requests that we are holding until commit comes in. Keys represent session ids, each value is a linked list of the session's requests.
-
numRequestsProcessing
protected final java.util.concurrent.atomic.AtomicInteger numRequestsProcessing
The number of requests currently being processed
-
stoppedMainLoop
protected volatile boolean stoppedMainLoop
For testing purposes, we use a separated stopping condition for the outer loop.
-
stopped
protected volatile boolean stopped
-
workerPool
protected WorkerService workerPool
-
-
Constructor Detail
-
CommitProcessor
public CommitProcessor(RequestProcessor nextProcessor, java.lang.String id, boolean matchSyncs, ZooKeeperServerListener listener)
-
-
Method Detail
-
needCommit
protected boolean needCommit(Request request)
-
run
public void run()
- Specified by:
runin interfacejava.lang.Runnable- Overrides:
runin classjava.lang.Thread
-
endOfIteration
protected void endOfIteration()
-
waitForEmptyPool
protected void waitForEmptyPool() throws java.lang.InterruptedException- Throws:
java.lang.InterruptedException
-
start
public void start()
- Overrides:
startin classjava.lang.Thread
-
getMaxReadBatchSize
public static int getMaxReadBatchSize()
-
getMaxCommitBatchSize
public static int getMaxCommitBatchSize()
-
setMaxReadBatchSize
public static void setMaxReadBatchSize(int size)
-
setMaxCommitBatchSize
public static void setMaxCommitBatchSize(int size)
-
commit
public void commit(Request request)
-
processRequest
public void processRequest(Request request)
- Specified by:
processRequestin interfaceRequestProcessor
-
shutdown
public void shutdown()
- Specified by:
shutdownin interfaceRequestProcessor
-
-