I have a class like this
import java.util.concurrent.*;
public class TestClass<V> {
private final ConcurrentMap<String, Future<V>> requests;
private final ExecutorService executorService;
public TestClass(final ExecutorService executorService) {
this.executorService = executorService;
this.requests = new ConcurrentHashMap<>();
}
public V submitRequest(String cacheKey, Callable<V> request) throws Exception {
final Future<V> task = getOrCreateTask(cacheKey, request);
final V results;
try {
results = task.get();
} catch (InterruptedException | ExecutionException e) {
throw new IllegalStateException(String.format("Exception while executing request for key '%s'", cacheKey),
e);
} finally {
//Nullpointer here
requests.remove(cacheKey);
}
return results;
}
private synchronized Future<V> getOrCreateTask(String key, Callable<V> request) {
if (requests.containsKey(key)) {
return requests.get(key);
} else {
final Future<V> newTask = executorService.submit(request);
requests.put(key, newTask);
return newTask;
}
}
}
but sometimes under heavy load server throws nullpointer on requests.remove(cacheKey). I have read final when not escaped by this in the constructor is write guaranteed. i.e. other threads can see what is going on with my requests map.
Not sure how do i fix efficiently? Does not like that idea of adding synchronised on the whole parent level method
I'm not actually sure the NPE is where you're identifying it is unless cacheKey is null, which you could check for. The concurrentmap is set correctly so the requests field should never be null. Nevertheless, this code is not correctly synchronized. You are attempting to perform two operations in getOrCreateTask() that while under the synchronized keyword are not correctly synchronized with the map because the map is interacted with in submitRequest when you remove the values.
What is likely happening is that between the check ConcurrentMap#containsKey and ConcurrentMap#get that another thread has removed the value from the cache (ConcurrentMap#remove).
Thread A: Check Contains "foobar" => true
Thread B: Remove "foobar"
Thread A: Call get("foobar") => null
Thread A: Calls Future#get on a null pointer, which then throws a NPE.
Since you control the concurrentmap you can know you'll never have null values. In that case you should instead just call the #get method and check if the returned value is null. This will prevent another thread from removing the value between a contains/get pair since you'll be only accessing the map through one atomic operation.
Related
I have method in class MyClassB which is triggered asynchronously from a method of MyClassA:
public void getProductCall()
{
new Thread(new Runnable() {
#Override
public void run() {
try {
productRequest = service.createS4ProductRequest(getRepriceItems());
//Below is a rest call to another system
String response = pricing.getS4ProductResponse(quote.getAssetQuoteNrAndVrsn(), productRequest);
//I'm using the below 2 lines to check from ClassA's method to see if this process has ended
setProductResponse(response);
productPriceProcessEnded=true;
} catch (Exception e) {
productPriceErrorOccured=true;
e.printStackTrace();
}
}
}).start();
}
This is the piece of code in MyClassA i used to check if the above method is complete.
for(int i=0;i<1000000000;i++)
{
if(!networkAsynCalls.isListPriceErrorOccured())
{
if(networkAsynCalls.isListPriceprocessEnded())
{
return networkAsynCalls.getListReponse();
}
else
{
Thread.sleep(250);
continue;
}
}
else
return null;
}
instead of using this random for loop can i use some inbuilt method or service pool or something ?
Because,
1) This thread on method is in another class
2) In class MyClassB i have few more methods like this, so i need to check the status of all the methods in MyClassA
Thanks for any help.
If I undestand what you're trying to do is dispatch some code to be ran asynchronously, then be able to wait until it is completed (successfully or failed). If that's the case, you should take a look at Futures.
Here is an example based on the Javadoc:
FutureTask<String> future =
new FutureTask<String>(new Callable<String>() {
public String call() {
// do stuff
return "result";
}});
This code creates an object "future" that can be invoked to execute searcher.search(target). At this point, the code is not executed at all. You simply have an object representing a computation that may be executed asynchronously. To do so, you'd call:
ExecutorService executor = Executors.newFixedThreadPool(5);
executor.execute(future);
This snippet created an Executor (which is a fixed pool of 5 threads), then handed over the future to it for execution. The executor will run the computation from Future asynchronously.
Future offers some methods (see the Javadoc) to wait until completion, cancel, check completion status, etc. For example,
String result = future.get();
will block, waiting for the result indefinitely. A get(10, TimeUnit.SECONDS) will wait for 10 seconds and if the future has not completed, throw.
I have a scenario where two threads invoke a method and this method generated a sequence using postgres nextval(test_sequence).
test_sequence is initailly assigned to 1.
public String createNotification() {
logger.info("createNotification ENTRY");
Future<String> futRes = this.threadPool.submit(new Callable<String>() {
#Override
public String call() {
String notificationID = getNotificationId();//DB CALL TO GENERATE THE NEXT SEQUENCE.
boolean isInsertSuccess = notificationDaoService.insertNotificationIntoDB(notificationID);
if (isInsertSuccess == true) {
return notificationID;
} else {
return null;
}
}
});
try {
return futRes.get(5, TimeUnit.SECONDS);
} catch (Exception e) {
logger.error("Issue while getting value from future with exception :", e);
return null;
}
}
So in the above snippet, getNotificationId() will generate the sequence and insertNotificationIntoDB() wil insert the generated notification id to the table.
I some times observing the primary key voilation exception when multiple threads try to invoke createNotification().
So i am thinking to synchronise the db calls as mentioned below,
synchronised(object)
{
String notificationID = getNotificationId();
boolean isInsertSuccess = notificationDaoService.insertNotificationIntoDB(notificationID);
}
is this solution ok?
and also i want to ask if i can generalise that if multiple threads are accessing a function and if that function has DB calls that does basic CRUD, then all the DB calls needs to be synchronised. Is this right inference?
Has anyone retrieved the auto-generated keys for a database insert while using Groovy SQL's withBatch method? I have the following code
def Sql target = ...//database connection
target.withBatch { ps ->
insertableStuff.each { ps.addBatch ( it ) }
ps.executeBatch()
def results = ps.getGeneratedKeys() //what do I do with this?
}
We're using DB2, and I've successfully tested the getGeneratedKeys method with a single statement/result set, but once I wrap the process in a batch, I'm not sure what objects I'm dealing with anymore.
According to IBM, it is possible to get the results back, but their example is using standard JDBC objects, not the groovy ones. Any ideas?
I took the Groovy SQL stuff out the picture to see if I could get something working, I wanted to make sure that DB2 for z/OS actually supported the function, and was able to get the generated values. I was using IBM's example, however I had to add some extra code to handle for the casting that the IBM example is using.
SQL target = ...//get database connection
def preparedStatement = target.connection.prepareStatement(statement, ['ISN'] as String[])
ResultSet[] resultSets = ((DB2PreparedStatement) (ps.getDelegate().getDelegate())).getDBGeneratedKeys()
resultSets.each { ResultSet results ->
while(results.next()) {
println results.getInt(1)
}
}
So... that's a little clunky, but it's functional. Unfortunately, by controlling the statement myself, I lost all of the parameter mapping that Groovy normally does for me.
I was looking through the groovy Sql source code and can see where they are explicitly telling the database connection not to handle parameters, so I'm thinking I'll add a new method to Sql.metaClass that can pass in a list of the auto-generated column names or something to make this more palatable.
I also want to see if there's a way to get the getGeneratedKeys method working so that I don't have to do all of that casting. At the very least, a utility method to safely handle the casting for me.
try {
withinBatch = true;
PreparedStatement statement = (PreparedStatement) getAbstractStatement(new CreatePreparedStatementCommand(0), connection, sql);
configure(statement);
psWrapper = new BatchingPreparedStatementWrapper(statement, indexPropList, batchSize, LOG, this);
closure.call(psWrapper);
return psWrapper.executeBatch();
} catch (SQLException e) {
The createNewPreparedStatement(0) prevents the creation of a statement which could return the auto-generated keys.
Just to make sure I wasn't crazy, I re-tried the 'getGeneratedKeys' method again with a statement that I know works and I got no results (see below). I had to recursively spin through the results to find the IBM class. So... not my favorite code, it's pretty brittle, but it's functional. Now I just need to see if I can still use the withBatch method somehow, I'll obviously need to override some things.
println 'print using getGeneratedKeys'
def results = preparedStatement.getGeneratedKeys()
while (results.next()) {
println SqlGroovyMethods.toRowResult(results)
}
println 'print using delegate processing'
println getGeneratedKeys(preparedStatement)
private List getGeneratedKeys(PreparedStatement statement) {
switch (statement) {
case DelegatingStatement:
return getGeneratedKeys(DelegatingStatement.cast(statement).getDelegate())
case DB2PreparedStatement:
ResultSet[] resultSets = DB2PreparedStatement.cast(statement).getDBGeneratedKeys()
List keys = []
resultSets.each { ResultSet results ->
while (results.next()) {
keys << SqlGroovyMethods.toRowResult(results)
}
}
return keys
default:
return [SqlGroovyMethods.toRowResult(statement.getGeneratedKeys())]
}
}
---- Console Output ----
print using getGeneratedKeys
print using delegate processing
[[KEY:7391], [KEY:7392]]
Okay, got it working. I had to hack my way into the Groovy SQL class, and there are some things that I just couldn't do because the methods in the Groovy class were private, so this implementation doesn't support cachedStatements, the isWithinBatch method won't operate correctly in the closure, and there's no access to the number of rows that were updated.
It'd be nice to see some variation of this in the base Groovy code, perhaps with a extension point where you put in your own handler (since you wouldn't want the IBM specific stuff in the base Groovy code), but at least I have a workable solution now.
public class SqlWithGeneratedKeys extends Sql {
public SqlWithGeneratedKeys(Sql parent) {
super(parent);
}
public List<GroovyRowResult> withBatch(String pSql, String [] keys, Closure closure) throws SQLException {
return this.withBatch(0, pSql, keys, closure);
}
public List<GroovyRowResult> withBatch(int batchSize, String pSql, String [] keys, Closure closure) throws SQLException {
final Connection connection = this.createConnection();
List<Tuple> indexPropList = null;
final SqlWithParams preCheck = this.buildSqlWithIndexedProps(pSql);
BatchingPreparedStatementWrapper psWrapper = null;
String sql = pSql;
if (preCheck != null) {
indexPropList = new ArrayList<Tuple>();
for (final Object next : preCheck.getParams()) {
indexPropList.add((Tuple) next);
}
sql = preCheck.getSql();
}
PreparedStatement statement = null;
try {
statement = connection.prepareStatement(sql, keys);
this.configure(statement);
psWrapper = new BatchingPreparedStatementWrapper(statement, indexPropList, batchSize, LOG, this);
closure.call(psWrapper);
psWrapper.executeBatch();
return this.getGeneratedKeys(statement);
} catch (final SQLException e) {
LOG.warning("Error during batch execution of '" + sql + "' with message: " + e.getMessage());
throw e;
} finally {
BaseDBServices.closeDBElements(connection, statement, null);
}
}
protected List<GroovyRowResult> getGeneratedKeys(Statement statement) throws SQLException {
if (statement instanceof DelegatingStatement) {
return this.getGeneratedKeys(DelegatingStatement.class.cast(statement).getDelegate());
} else if (statement instanceof DB2PreparedStatement) {
final ResultSet[] resultSets = DB2PreparedStatement.class.cast(statement).getDBGeneratedKeys();
final List<GroovyRowResult> keys = new ArrayList<GroovyRowResult>();
for (final ResultSet results : resultSets) {
while (results.next()) {
keys.add(SqlGroovyMethods.toRowResult(results));
}
}
return keys;
}
return Arrays.asList(SqlGroovyMethods.toRowResult(statement.getGeneratedKeys()));
}
}
Calling it is nice and clean.
println new SqlWithGeneratedKeys(target).withBatch(statement, ['ISN'] as String[]) { ps ->
rows.each {
ps.addBatch(it)
}
}
I have an MVC application in which I have a controller that receives data from the user and then uploads a file to Azure blob storage. The application is using Unity IoC to handle dependency injection.
During the workflow I have isolated the following code as demonstrating the problem
public class MvcController : Controller
{
private IDependencyResolver _dependencyResolver;
public MvcController() : this(DependencyResolver.Current)
{
}
public MvcController(IDependencyResolver dependencyResolver)
{
this._dependencyResolver = dependencyResolver;
}
public GetService<T>()
{
T resolved = _dependencyResolver.GetService<T>()
if (resolved == null)
throw new Exception(string.Format("Dependency resolver does not contain service of type {0}", typeof(T).Name));
return resolved;
}
}
public class MyController : MvcController
{
[NoAsyncTimeout]
public async Task<ActionResult> SaveFileAsync(/* A bunch of arguments */)
{
/* A bunch of code */
//This line gets a concrete instance from HttpContext.Current successfully...
IMyObject o = GetService<IMyObject>();
await SaveFileToAzure(/* A bunch of parameters */);
.
.
/* Sometime later */
Method2(/* A bunch of parameters */);
}
private Method2(/* A bunch of parameters */)
{
//This line fails because HttpContext.Current is null
IMyObject o = GetService<IMyObject>();
/* A bunch of other code */
}
private async Task SaveFileToAzure(/* A bunch of parameters */)
{
//Grab a blob container to store the file data...
CloudBlobContainer blobContainer = GetBlobContainer();
ICloudBlob blob = blobContainer.GetBlockBlobReference(somePath);
Stream dataStream = GetData();
System.Threading.CancellationToken cancelToken = GetCancellationToken();
//All calls to DependencyResolver.GetService<T>() after this line of code fail...
response = await blob.UploadStreamAsync(dataStream, cancelToken);
/* A bunch of other code */
}
}
Unity has a registration for my object:
container.RegisterType<IMyObject, MyObject>(new HttpLifetimeManager());
My lifetime manager is defined as follows:
public sealed class HttpRequestLifetimeManager : LifetimeManager
{
public Guid Key { get; private set; }
public HttpRequestLifetimeManager()
{
this.Key = Guid.NewGuid();
}
public override object GetValue()
{
return HttpContext.Current.Items[(object)this.Key];
}
public override void SetValue(object newValue)
{
HttpContext.Current.Items[(object)this.Key] = newValue;
}
public override void RemoveValue()
{
HttpContext.Current.Items.Remove((object)this.Key);
}
}
Nothing complicated.
Stepping into the HttpRequestLifetimeManager on the failing GetService() calls shows that after the UploadStreamAsync() call HttpContext.Current is null...
Has anyone else come across this problem? If so, is this a bug? Is this expected behaviour? Am I doing something out of the ordinary? What should I do to resolve it?
I can hack around it by storing a reference to HttpContext.Current prior to the offending call and restoring it after, but that doesn't seem like the right approach.
Any ideas?
To echo #Joachim - http context may not be available to your async thread. Compare the current thread id where you can see httpcontext is available, to the thread id where you can see that it isn't - i'm assuming you will see they are 2 different threads. If my assumption is correct this may be a sign that your main thread (the one with httpcontext) does not have a "synchronizationcontext". (you can see http://blogs.msdn.com/b/pfxteam/archive/2012/01/20/10259049.aspx for more details of how that works) If so, it may mean that the code immediately after your await statement is actually not running on the same thread as the code prior to the await statement! So from your perspective, one moment you have http context and the next you don't because execution has actually been switched to another thread! You should probably look at implementing / setting a synchronizationcontext on your main thread if that's the case and then control will be returned to your original thread with http context and that should fix your problem, or alternatively you could retrieve your object from http context on the original thread and find a way to pass it as a parameter to the async method/s so that they don't need to access http context to get their state.
We are launching a website that will have a very heavy volume for a short period of time. It is basically giving tickets. The code is written in Java, Spring & Hibernate. I want to mimic the high volume by spawning multiple threads and trying to get the ticket using JUnit test case. The problem is that in my DAO class the code just simply dies after I begin transaction. I mean there is no error trace in the log file or anything like that. Let me give some idea about the way my code is.
DAO code:
#Repository("customerTicketDAO")
public class CustomerTicketDAO extends BaseDAOImpl {// BaseDAOImpl extends HibernateDaoSupport
public void saveCustomerTicketUsingJDBC(String customerId) {
try{
getSession().getTransaction().begin(); //NOTHING HAPPENS AFTER THIS LINE OF CODE
// A select query
Query query1 = getSession().createSQLQuery("my query omitted on purpose");
.
.
// An update query
Query query2 = getSession().createSQLQuery("my query omitted on purpose");
getSession().getTransaction().commite();
} catch (Exception e) {
}
}
Runnable code:
public class InsertCustomerTicketRunnable implements Runnable {
#Autowired
private CustomerTicketDAO customerTicketDAO;
public InsertCustomerTicketRunnable(String customerId) {
this.customerId = customerId;
}
#Override
public void run() {
if (customerTicketDAO != null) {
customerTicketDAO.saveCustomerTicketUsingJDBC(customerId);
}
}
}
JUnit method:
#RunWith(SpringJUnit4ClassRunner.class)
#ContextConfiguration(locations={"file:src/test/resources/applicationContext-test.xml"})
public class DatabaseTest {
#Before
public void init() {
sessionFactory = (SessionFactory)applicationContext.getBean("sessionFactory");
Session session = SessionFactoryUtils.getSession(sessionFactory, true);
TransactionSynchronizationManager.bindResource(sessionFactory, new SessionHolder(session));
customerTicketDAO = (CustomerTicketDAO)applicationContext.getBean("customerTicketDAO");
}
#After
public void end() throws Exception {
SessionHolder sessionHolder = (SessionHolder) TransactionSynchronizationManager.unbindResource(sessionFactory);
SessionFactoryUtils.closeSession(session);
}
#Test
public void saveCustomerTicketInMultipleThreads () throws Exception {
ExecutorService executor = Executors.newFixedThreadPool(NTHREDS);
for (int i=0; i<1000; i++) {
executor.submit(new InsertCustomerTicketRunnable(i));
}
// This will make the executor accept no new threads
// and finish all existing threads in the queue
executor.shutdown();
// Wait until all threads are finish
executor.awaitTermination(1, TimeUnit.SECONDS);
}
I see no data being inserted into the database. Can someone please point me as to where I am going wrong?
Thanks
Raj
SessionFactory is thread safe but Session is not. So my guess is that you need to call SessionFactoryUtils.getSession() from within each thread, so that each thread gets its own instance. You are currently calling it from the main thread, so all children threads try to share the same instance.
Naughty, naughty!
public void saveCustomerTicketUsingJDBC(String customerId) {
try {
getSession().getTransaction().begin(); //NOTHING HAPPENS AFTER THIS LINE OF CODE
.
.
} catch (Exception e) {
}
}
You should never (well, hardly ever) have an empty catch block, if there is a problem you will find that your code 'just simply dies' with no log messages. Oh look, that's what's happening ;)
At the very minimum you should log the exception, that will go a long way towards you helping you find what the problem is (and from there, the solution).