Develop Reference

Kinesis 2.2.11 java unable to create consumer

I need to migrate to Kinesis library to version 2.2.11 so I followed the tutorial: https://docs.aws.amazon.com/streams/latest/dev/kcl-migration.html
I need to run multiple instances of my consumer app, so every one of them needs to have an unique application name in order to have a separate lease table in DynamoDb.
When initializing the consumer Kinesis runs DynamoDBLeaseRefresher.createLeaseTableIfNotExists which checks if a new table needs to be created for this application name and creates one if it cannot be found.
So 2 operations are performed:
DescribeTable - it returns the table info or throws a ResourceNotFoundExecption,
if needed - CreateTable.
The problem for me is with the DescribeTable method. When I am looking for an existing table it returns it with no problem. But when I am looking for a non-existent table it throws the ResourceNotFoundExecption -> so far so good. Unfortunately it then gets wrapped and is now:
java.util.concurrent.CompletionException: software.amazon.awssdk.core.exception.SdkClientException: Unable to execute HTTP request: software.amazon.awssdk.awscore.exception.AwsServiceException$Builder.extendedRequestId(Ljava/lang/String;)Lsoftware/amazon/awssdk/awscore/exception/AwsServiceException$Builder;
and the app expecting ResourceNotFoundException gets something different instead and crashes.
The wrapped exception message is a bit misleading: "Unable to execute HTTP request" since the request was performed and returned the proper message: "Resource not found".
Funny thing is that it sometimes works, the exception does not get wrapped, the CreateTable operation is performed and the consumer starts properly.
I have made a workaround for it for now where I just create the table before the initialization of the LeaseCoordinator, so it always gets the existing table.
here is my code:
public KinesisStreamReaderService(String streamName, String applicationName, String regionName) {
KinesisAsyncClient kinesisClient = KinesisAsyncClient.builder()
.credentialsProvider(EnvironmentVariableCredentialsProvider.create())
.region(Region.of(connectionProperties.getRegion()))
.httpClientBuilder(createHttpClientBuilder())
.build();
DynamoDbAsyncClient dynamoClient = DynamoDbAsyncClient.builder().region(Region.of(regionName)).build();
CloudWatchAsyncClient cloudWatchClient = CloudWatchAsyncClient.builder().region(Region.of(regionName)).build();
// if(!dynamoDbTableExists(dynamoClient, applicationName)) {
// createDynamoDbTable(dynamoClient, applicationName);
// }
ConfigsBuilder configsBuilder = new ConfigsBuilder(streamName, applicationName, kinesisClient,
dynamoClient, cloudWatchClient, workerId(), KinesisReaderProcessor::new);
configsBuilder.retrievalConfig().initialPositionInStreamExtended(
InitialPositionInStreamExtended.newInitialPosition(
InitialPositionInStream.LATEST));
scheduler = new Scheduler(
configsBuilder.checkpointConfig(),
configsBuilder.coordinatorConfig(),
configsBuilder.leaseManagementConfig(),
configsBuilder.lifecycleConfig(),
configsBuilder.metricsConfig(),
configsBuilder.processorConfig(),
configsBuilder.retrievalConfig().retrievalSpecificConfig(new PollingConfig(streamName, kinesisClient))
);
}
private void createDynamoDbTable(DynamoDbAsyncClient dynamoClient, String applicationName) {
log.info("Creating new lease table: {}", applicationName);
CompletableFuture<CreateTableResponse> createTableFuture = dynamoClient
.createTable(CreateTableRequest.builder()
.provisionedThroughput(ProvisionedThroughput.builder().readCapacityUnits(10L).writeCapacityUnits(10L).build())
.tableName(applicationName)
.keySchema(KeySchemaElement.builder().attributeName("leaseKey").keyType(KeyType.HASH).build())
.attributeDefinitions(AttributeDefinition.builder().attributeName("leaseKey").attributeType(
ScalarAttributeType.S).build())
.build());
try {
CreateTableResponse createTableResponse = createTableFuture.get();
log.debug("Created new lease table: {}", createTableResponse.tableDescription().tableName());
} catch (InterruptedException | ExecutionException e) {
throw new DataStreamException(e.getMessage(), e);
}
}
private boolean dynamoDbTableExists(DynamoDbAsyncClient dynamoClient, String tableName) {
CompletableFuture<DescribeTableResponse> describeTableResponseCompletableFutureNew = dynamoClient
.describeTable(DescribeTableRequest.builder()
.tableName(tableName).build());
try {
DescribeTableResponse describeTableResponseNew = describeTableResponseCompletableFutureNew
.get();
return nonNull(describeTableResponseNew);
} catch (InterruptedException | ExecutionException e) {
log.info(e.getMessage(), e);
}
return false;
}
private static String workerId() {
String workerId;
try {
workerId = format("%s_%s", getLocalHost().getCanonicalHostName(), randomUUID().toString());
} catch (UnknownHostException e) {
workerId = randomUUID().toString();
}
return workerId;
}
#Override
public void read(Consumer<String> consumer) {
this.consumer = consumer;
scheduler.run();
}
private class KinesisReaderProcessor implements ShardRecordProcessor {
private String shardId;
#Override
public void initialize(InitializationInput initializationInput) {
this.shardId = initializationInput.shardId();
log.info("Initializing record processor for shard: {}", shardId);
}
#Override
public void processRecords(ProcessRecordsInput processRecordsInput) {
log.debug("Checking shard {} for new records", shardId);
List<KinesisClientRecord> records = processRecordsInput.records();
if (!records.isEmpty()) {
log.debug("Processing {} records from kinesis stream shard {}", records.size(), shardId);
records.forEach(record -> {
String json = UTF_8.decode(record.data()).toString();
log.info(json);
consumer.accept(json);
});
}
}
#Override
public void leaseLost(LeaseLostInput leaseLostInput) {
log.info("Record processor has lost lease, terminating");
}
#Override
public void shardEnded(ShardEndedInput shardEndedInput) {
try {
shardEndedInput.checkpointer().checkpoint();
} catch (ShutdownException | InvalidStateException e) {
log.error(e.getMessage(), e);
}
}
#Override
public void shutdownRequested(ShutdownRequestedInput shutdownRequestedInput) {
try {
shutdownRequestedInput.checkpointer().checkpoint();
} catch (ShutdownException | InvalidStateException e) {
log.error(e.getMessage(), e);
}
}
}
}
Am I missing some configuration for the scheduler or something? Why is it sometimes working?
Thanks
Edit:
The problem is this block of code in DynamoDBLeaseRefresher.tableStatus() is invoked to check if the table exists:
DescribeTableResponse result;
try {
try {
result =
(DescribeTableResponse)FutureUtils.resolveOrCancelFuture(this.dynamoDBClient.describeTable(request), this.dynamoDbRequestTimeout);
} catch (ExecutionException var5) {
throw exceptionManager.apply(var5.getCause());
} catch (InterruptedException var6) {
throw new DependencyException(var6);
}
} catch (ResourceNotFoundException var7) {
log.debug("Got ResourceNotFoundException for table {} in leaseTableExists, returning false.", this.table);
return null;
}
and in my case it should get ResourceNotFoundException if the table is not found, but as I said the expection gets wrapped to CompletionException before it reaches the appropriate catch block and is caught in the code here:
catch (ExecutionException var5) {
throw exceptionManager.apply(var5.getCause());
This is happening 20 times in the loop while trying to Initialize the LeaseCoordinator and then just stops trying to initialize the connection. (As mentioned above it works occasionally, but that makes it even stranger to me)
With my workaround it only needs 1 try to get initialized

You don't need to create a lease table manually - DynamoDBLeaseCoordinator will create one if not exists on initialization and wait until it exists:
#Override
public void initialize() throws ProvisionedThroughputException, DependencyException, IllegalStateException {
final boolean newTableCreated =
leaseRefresher.createLeaseTableIfNotExists(initialLeaseTableReadCapacity, initialLeaseTableWriteCapacity);
if (newTableCreated) {
log.info("Created new lease table for coordinator with initial read capacity of {} and write capacity of {}.",
initialLeaseTableReadCapacity, initialLeaseTableWriteCapacity);
}
// Need to wait for table in active state.
final long secondsBetweenPolls = 10L;
final long timeoutSeconds = 600L;
final boolean isTableActive = leaseRefresher.waitUntilLeaseTableExists(secondsBetweenPolls, timeoutSeconds);
if (!isTableActive) {
throw new DependencyException(new IllegalStateException("Creating table timeout"));
}
}
The issue in your case, I think, is that it's eventually created and you probably should periodically check until table appears - like DynamoDBLeaseCoordinator#initialize() does.

Concurrency on Vertx

i have joined to one of those Vertx lovers , how ever the single threaded main frame may not be working for me , because in my server there might be 50 file download requests at a moment , as a work around i have created this class
public abstract T onRun() throws Exception;
public abstract void onSuccess(T result);
public abstract void onException();
private static final int poolSize = Runtime.getRuntime().availableProcessors();
private static final long maxExecuteTime = 120000;
private static WorkerExecutor mExecutor;
private static final String BG_THREAD_TAG = "BG_THREAD";
protected RoutingContext ctx;
private boolean isThreadInBackground(){
return Thread.currentThread().getName() != null && Thread.currentThread().getName().equals(BG_THREAD_TAG);
}
//on success will not be called if exception be thrown
public BackgroundExecutor(RoutingContext ctx){
this.ctx = ctx;
if(mExecutor == null){
mExecutor = MyVertxServer.vertx.createSharedWorkerExecutor("my-worker-pool",poolSize,maxExecuteTime);
}
if(!isThreadInBackground()){
/** we are unlocking the lock before res.succeeded , because it might take long and keeps any thread waiting */
mExecutor.executeBlocking(future -> {
try{
Thread.currentThread().setName(BG_THREAD_TAG);
T result = onRun();
future.complete(result);
}catch (Exception e) {
GUI.display(e);
e.printStackTrace();
onException();
future.fail(e);
}
/** false here means they should not be parallel , and will run without order multiple times on same context*/
},false, res -> {
if(res.succeeded()){
onSuccess((T)res.result());
}
});
}else{
GUI.display("AVOIDED DUPLICATE BACKGROUND THREADING");
System.out.println("AVOIDED DUPLICATE BACKGROUND THREADING");
try{
T result = onRun();
onSuccess((T)result);
}catch (Exception e) {
GUI.display(e);
e.printStackTrace();
onException();
}
}
}
allowing the handlers to extend it and use it like this
public abstract class DefaultFileHandler implements MyHttpHandler{
public abstract File getFile(String suffix);
#Override
public void Handle(RoutingContext ctx, VertxUtils utils, String suffix) {
new BackgroundExecutor<Void>(ctx) {
#Override
public Void onRun() throws Exception {
File file = getFile(URLDecoder.decode(suffix, "UTF-8"));
if(file == null || !file.exists()){
utils.sendResponseAndEnd(ctx.response(),404);
return null;
}else{
utils.sendFile(ctx, file);
}
return null;
}
#Override
public void onSuccess(Void result) {}
#Override
public void onException() {
utils.sendResponseAndEnd(ctx.response(),404);
}
};
}
and here is how i initialize my vertx server
vertx.deployVerticle(MainDeployment.class.getCanonicalName(),res -> {
if (res.succeeded()) {
GUI.display("Deployed");
} else {
res.cause().printStackTrace();
}
});
server.requestHandler(router::accept).listen(port);
and here is my MainDeployment class
public class MainDeployment extends AbstractVerticle{
#Override
public void start() throws Exception {
// Different ways of deploying verticles
// Deploy a verticle and don't wait for it to start
for(Entry<String, MyHttpHandler> entry : MyVertxServer.map.entrySet()){
MyVertxServer.router.route(entry.getKey()).handler(new Handler<RoutingContext>() {
#Override
public void handle(RoutingContext ctx) {
String[] handlerID = ctx.request().uri().split(ctx.currentRoute().getPath());
String suffix = handlerID.length > 1 ? handlerID[1] : null;
entry.getValue().Handle(ctx, new VertxUtils(), suffix);
}
});
}
}
}
this is working just fine when and where i need it , but i still wonder if is there any better way to handle concurencies like this on vertx , if so an example would be really appreciated . thanks alot

I don't fully understand your problem and reasons for your solution. Why don't you implement one verticle to handle your http uploads and deploy it multiple times? I think that handling 50 concurrent uploads should be a piece of cake for vert.x.
When deploying a verticle using a verticle name, you can specify the number of verticle instances that you want to deploy:
DeploymentOptions options = new DeploymentOptions().setInstances(16);
vertx.deployVerticle("com.mycompany.MyOrderProcessorVerticle", options);
This is useful for scaling easily across multiple cores. For example you might have a web-server verticle to deploy and multiple cores on your machine, so you want to deploy multiple instances to take utilise all the cores.
http://vertx.io/docs/vertx-core/java/#_specifying_number_of_verticle_instances

vertx is a well-designed model so that a concurrency issue does not occur.
generally, vertx does not recommend the multi-thread model.
(because, handling is not easy.)
If you select multi-thread model, you have to think about shared data..
Simply, if you just only want to split EventLoop Area,
first of all, you make sure Check your a number of CPU Cores.
and then Set up the count of Instances .
DeploymentOptions options = new DeploymentOptions().setInstances(4);
vertx.deployVerticle("com.mycompany.MyOrderProcessorVerticle", options);
But, If you have 4cores of CPU, you don't set up over 4 instances.
If you set up to number four or more, the performance won't improve.
vertx concurrency reference
http://vertx.io/docs/vertx-core/java/

Java threads - waiting on all child threads in order to proceed

So a little background;
I am working on a project in which a servlet is going to release crawlers upon a lot of text files within a file system. I was thinking of dividing the load under multiple threads, for example:
a crawler enters a directory, finds 3 files and 6 directories. it will start processing the files and start a thread with a new crawler for the other directories. So from my creator class I would create a single crawler upon a base directory. The crawler would assess the workload and if deemed needed it would spawn another crawler under another thread.
My crawler class looks like this
package com.fujitsu.spider;
import java.io.BufferedReader;
import java.io.File;
import java.io.FileReader;
import java.io.IOException;
import java.io.Serializable;
import java.util.ArrayList;
public class DocumentSpider implements Runnable, Serializable {
private static final long serialVersionUID = 8401649393078703808L;
private Spidermode currentMode = null;
private String URL = null;
private String[] terms = null;
private float score = 0;
private ArrayList<SpiderDataPair> resultList = null;
public enum Spidermode {
FILE, DIRECTORY
}
public DocumentSpider(String resourceURL, Spidermode mode, ArrayList<SpiderDataPair> resultList) {
currentMode = mode;
setURL(resourceURL);
this.setResultList(resultList);
}
#Override
public void run() {
try {
if (currentMode == Spidermode.FILE) {
doCrawlFile();
} else {
doCrawlDirectory();
}
} catch (Exception e) {
e.printStackTrace();
}
System.out.println("SPIDER # " + URL + " HAS FINISHED.");
}
public Spidermode getCurrentMode() {
return currentMode;
}
public void setCurrentMode(Spidermode currentMode) {
this.currentMode = currentMode;
}
public String getURL() {
return URL;
}
public void setURL(String uRL) {
URL = uRL;
}
public void doCrawlFile() throws Exception {
File target = new File(URL);
if (target.isDirectory()) {
throw new Exception(
"This URL points to a directory while the spider is in FILE mode. Please change this spider to FILE mode.");
}
procesFile(target);
}
public void doCrawlDirectory() throws Exception {
File baseDir = new File(URL);
if (!baseDir.isDirectory()) {
throw new Exception(
"This URL points to a FILE while the spider is in DIRECTORY mode. Please change this spider to DIRECTORY mode.");
}
File[] directoryContent = baseDir.listFiles();
for (File f : directoryContent) {
if (f.isDirectory()) {
DocumentSpider spider = new DocumentSpider(f.getPath(), Spidermode.DIRECTORY, this.resultList);
spider.terms = this.terms;
(new Thread(spider)).start();
} else {
DocumentSpider spider = new DocumentSpider(f.getPath(), Spidermode.FILE, this.resultList);
spider.terms = this.terms;
(new Thread(spider)).start();
}
}
}
public void procesDirectory(String target) throws IOException {
File base = new File(target);
File[] directoryContent = base.listFiles();
for (File f : directoryContent) {
if (f.isDirectory()) {
procesDirectory(f.getPath());
} else {
procesFile(f);
}
}
}
public void procesFile(File target) throws IOException {
BufferedReader br = new BufferedReader(new FileReader(target));
String line;
while ((line = br.readLine()) != null) {
String[] words = line.split(" ");
for (String currentWord : words) {
for (String a : terms) {
if (a.toLowerCase().equalsIgnoreCase(currentWord)) {
score += 1f;
}
;
if (currentWord.toLowerCase().contains(a)) {
score += 1f;
}
;
}
}
}
br.close();
resultList.add(new SpiderDataPair(this, URL));
}
public String[] getTerms() {
return terms;
}
public void setTerms(String[] terms) {
this.terms = terms;
}
public float getScore() {
return score;
}
public void setScore(float score) {
this.score = score;
}
public ArrayList<SpiderDataPair> getResultList() {
return resultList;
}
public void setResultList(ArrayList<SpiderDataPair> resultList) {
this.resultList = resultList;
}
}
The problem I am facing is that in my root crawler I have this list of results from every crawler that I want to process further. The operation to process the data from this list is called from the servlet (or main method for this example). However the operations is always called before all of the crawlers have completed their processing. thus launching the operation to process the results too soon, which leads to incomplete data.
I tried solving this using the join methods but unfortunately I cant seems to figure this one out.
package com.fujitsu.spider;
import java.util.ArrayList;
import com.fujitsu.spider.DocumentSpider.Spidermode;
public class Main {
public static void main(String[] args) throws InterruptedException {
ArrayList<SpiderDataPair> results = new ArrayList<SpiderDataPair>();
String [] terms = {"SERVER","CHANGE","MO"};
DocumentSpider spider1 = new DocumentSpider("C:\\Users\\Mark\\workspace\\Spider\\Files", Spidermode.DIRECTORY, results);
spider1.setTerms(terms);
DocumentSpider spider2 = new DocumentSpider("C:\\Users\\Mark\\workspace\\Spider\\File2", Spidermode.DIRECTORY, results);
spider2.setTerms(terms);
Thread t1 = new Thread(spider1);
Thread t2 = new Thread(spider2);
t1.start();
t1.join();
t2.start();
t2.join();
for(SpiderDataPair d : spider1.getResultList()){
System.out.println("PATH -> " + d.getFile() + " SCORE -> " + d.getSpider().getScore());
}
for(SpiderDataPair d : spider2.getResultList()){
System.out.println("PATH -> " + d.getFile() + " SCORE -> " + d.getSpider().getScore());
}
}
}
TL:DR
I really wish to understand this subject so any help would be immensely appreciated!.

You need a couple of changes in your code:
In the spider:
List<Thread> threads = new LinkedList<Thread>();
for (File f : directoryContent) {
if (f.isDirectory()) {
DocumentSpider spider = new DocumentSpider(f.getPath(), Spidermode.DIRECTORY, this.resultList);
spider.terms = this.terms;
Thread thread = new Thread(spider);
threads.add(thread)
thread.start();
} else {
DocumentSpider spider = new DocumentSpider(f.getPath(), Spidermode.FILE, this.resultList);
spider.terms = this.terms;
Thread thread = new Thread(spider);
threads.add(thread)
thread.start();
}
}
for (Thread thread: threads) thread.join()
The idea is to create a new thread for each spider and start it. Once they are all running, you wait until each on is done before the Spider itself finishes. This way each spider thread keeps running until all of its work is done (thus the top thread runs until all children and their children are finished).
You also need to change your runner so that it runs the two spiders in parallel instead of one after another like this:
Thread t1 = new Thread(spider1);
Thread t2 = new Thread(spider2);
t1.start();
t2.start();
t1.join();
t2.join();

You should use a higher-level library than bare Thread for this task. I would suggest looking into ExecutorService in particular and all of java.util.concurrent generally. There are abstractions there that can manage all of the threading issues while providing well-formed tasks a properly protected environment in which to run.
For your specific problem, I would recommend some sort of blocking queue of tasks and a standard producer-consumer architecture. Each task knows how to determine if its path is a file or directory. If it is a file, process the file; if it is a directory, crawl the directory's immediate contents and enqueue new tasks for each sub-path. You could also use some properly-synchronized shared state to cap the number of files processed, depth, etc. Also, the service provides the ability to await termination of its tasks, making the "join" simpler.
With this architecture, you decouple the notion of threads and thread management (handled by the ExecutorService) with your business logic of tasks (typically a Runnable or Callable). The service itself has the ability to tune how to instantiate, such as a fixed maximum number of threads or a scalable number depending on how many concurrent tasks exist (See factory methods on java.util.concurrent.Executors). Threads, which are more expensive than the Runnables they execute, are re-used to conserve resources.
If your objective is primarily something functional that works in production quality, then the library is the way to go. However, if your objective is to understand the lower-level details of thread management, then you may want to investigate the use of latches and perhaps thread groups to manage them at a lower level, exposing the details of the implementation so you can work with the details.

How can I check if a hazelcast cluster is alive from a java client?

We use hazelcast in client-server mode. The hazelcast cluster contains 2 hazelcast nodes and we have about 25 clients connected to the cluster.
What I am lookin for now is a simple check that tries to figure out if the cluster is still alive. It should be a rather cheap operation because this check will occure on every client quite frequently (once every second I could imagine).
What is the best way to do so?

The simplest way would be the register a LifecycleListener to the client HazelcastInstance:
HazelcastInstance client = HazelcastClient.newHazelcastClient();
client.getLifecycleService().addLifecycleListener(new LifecycleListener() {
#Override
public void stateChanged(LifecycleEvent event) {
}
})
The client uses a periodic heartbeat to detect if the cluster is still running.

You can use the LifecycleService.isRunning() method as well:
HazelcastInstance hzInstance = HazelcastClient.newHazelcastClient();
hzInstance.getLifecycleService().isRunning()

As isRunning() may be true even if cluster is down, I'd go for the following approach (a mixture of #konstantin-zyubin's answer and this). This doesn't need an event-listener, which is an advantage in my setup:
if (!hazelcastInstance.getLifecycleService().isRunning()) {
return Health.down().build();
}
int parameterCount;
LocalTopicStats topicStats;
try {
parameterCount = hazelcastInstance.getMap("parameters").size();
topicStats = hazelcastInstance.getTopic("myTopic").getLocalTopicStats();
} catch (Exception e) {
// instance may run but cluster is down:
Health.Builder builder = Health.down();
builder.withDetail("Error", e.getMessage());
return builder.build();
}
Health.Builder builder = Health.up();
builder.withDetail("parameterCount", parameterCount);
builder.withDetail("receivedMsgs", topicStats.getReceiveOperationCount());
builder.withDetail("publishedMsgs", topicStats.getPublishOperationCount());
return builder.build();

I have found a more reliable way to check hazelcast availability, because
client.getLifecycleService().isRunning()
when you use async reconnection mode is always return true, as was mentioned.
#Slf4j
public class DistributedCacheServiceImpl implements DistributedCacheService {
private HazelcastInstance client;
#Autowired
protected ConfigLoader<ServersConfig> serversConfigLoader;
#PostConstruct
private void initHazelcastClient() {
ClientConfig config = new ClientConfig();
if (isCacheEnabled()) {
ServersConfig.Hazelсast hazelcastConfig = getWidgetCacheSettings().getHazelcast();
config.getGroupConfig().setName(hazelcastConfig.getName());
config.getGroupConfig().setPassword(hazelcastConfig.getPassword());
for (String address : hazelcastConfig.getAddresses()) {
config.getNetworkConfig().addAddress(address);
}
config.getConnectionStrategyConfig()
.setAsyncStart(true)
.setReconnectMode(ClientConnectionStrategyConfig.ReconnectMode.ASYNC);
config.getNetworkConfig()
.setConnectionAttemptLimit(0) // infinite (Integer.MAX_VALUE) attempts to reconnect
.setConnectionTimeout(5000);
client = HazelcastClient.newHazelcastClient(config);
}
}
#Override
public boolean isCacheEnabled() {
ServersConfig.WidgetCache widgetCache = getWidgetCacheSettings();
return widgetCache != null && widgetCache.getEnabled();
}
#Override
public boolean isCacheAlive() {
boolean aliveResult = false;
if (isCacheEnabled() && client != null) {
try {
IMap<Object, Object> defaultMap = client.getMap("default");
if (defaultMap != null) {
defaultMap.size(); // will throw Hazelcast exception if cluster is down
aliveResult = true;
}
} catch (Exception e) {
log.error("Connection to hazelcast cluster is lost. Reason : {}", e.getMessage());
}
}
return aliveResult;
}
}

How to implement fair usage policy of threads in API backend?

I am building an REST API in Java which I would be exposing to the outside world. People who would be invoking the API would have to be registered and would be sending their userid in the request.
There would be a maximum of, say, 10 concurrent threads available for executing the API request. I am maintaining a queue which holds all the request ids to be serviced (the primary key of the DB entry).
I need to implement some fair usage policy as follows -
If there are more than 10 jobs in the queue (i.e more than max number of threads), a user is allowed to execute only one request at a time (the other requests submitted by him/her, if any, would remain in the queue and would be taken up only once his previous request has completed execution). In case there are free threads, i.e. even after allotting threads to requests submitted by different users, then the remaining threads in the thread pool can be distributed among the remaining requests (even if the user who has submitted the request is already holding one thread at that moment).
The current implementation is as follows -
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import java.util.concurrent.PriorityBlockingQueue;
import java.util.concurrent.Semaphore;
public class APIJobExecutor implements Runnable{
private static PriorityBlockingQueue<Integer> jobQueue = new PriorityBlockingQueue<Integer>();
private static ExecutorService jobExecutor = Executors.newCachedThreadPool();
private static final int MAX_THREADS = 10;
private static Semaphore sem = new Semaphore(MAX_THREADS, true);
private APIJobExecutor(){
}
public static void addJob(int jobId)
{
if(!jobQueue.contains(jobId)){
jobQueue.add(new Integer(jobId));
}
}
public void run()
{
while (true) {
try {
sem.acquire();
}catch (InterruptedException e1) {
e1.printStackTrace();
//unable to acquire lock. retry.
continue;
}
try {
Integer jobItem = jobQueue.take();
jobExecutor.submit(new APIJobService(jobItem));
} catch (InterruptedException e) {
e.printStackTrace();
} finally {
sem.release();
}
}
}
}
Edit:
Is there any out of the box Java data structure that gives me this functionality. If not, how do I go about implementing the same?

This is a fairly common "quality of service" pattern and can be solved using the bucket idea within a job-queue. I do not know of a standard Java implementation and/or datastructure for this pattern (maybe the PriorityQueue?), but there should be at least a couple of implementations available (let us know if you find a good one).
I did once create my own implementation and I've tried to de-couple it from the project so that you may modify and use it (add unit-tests!). A couple of notes:
a default-queue is used in case QoS is not needed (e.g. if less than 10 jobs are executing).
the basic idea is to store tasks in lists per QoS-key (e.g. the username), and maintain a separate "who is next" list.
it is intended to be used within a job queue (e.g. part of the APIJobExecutor, not a replacement). Part of the job queue's responsibility is to always call remove(taskId) after a task is executed.
there should be no memory leaks: if there are no tasks/jobs in the queue, all internal maps and lists should be empty.
The code:
import java.util.concurrent.*;
import java.util.concurrent.atomic.AtomicInteger;
import java.util.concurrent.locks.ReentrantLock;
import java.util.*;
import org.slf4j.*;
/** A FIFO task queue. */
public class QosTaskQueue<TASKTYPE, TASKIDTYPE> {
private static final Logger log = LoggerFactory.getLogger(QosTaskQueue.class);
public static final String EMPTY_STRING = "";
/** Work tasks queued which have no (relevant) QoS key. */
private final ConcurrentLinkedQueue<TASKIDTYPE> defaultQ = new ConcurrentLinkedQueue<TASKIDTYPE>();
private final AtomicInteger taskQSize = new AtomicInteger();
private final Map<TASKIDTYPE, TASKTYPE> queuedTasks = new ConcurrentHashMap<TASKIDTYPE, TASKTYPE>();
/** Amount of tasks in queue before "quality of service" distribution kicks in. */
private int qosThreshold = 10;
/** Indicates if "quality of service" distribution is in effect. */
private volatile boolean usingQos;
/**
* Lock for all modifications to Qos-queues.
* <br>Must be "fair" to ensure adding does not block polling threads forever and vice versa.
*/
private final ReentrantLock qosKeyLock = new ReentrantLock(true);
/*
* Since all QoS modifications can be done by multiple threads simultaneously,
* there is never a good time to add or remove a Qos-key with associated queue.
* There is always a chance that a key is added while being removed and vice versa.
* The simplest solution is to make everything synchronized, which is what qosKeyLock is used for.
*/
private final Map<String, Queue<TASKIDTYPE>> qosQueues = new HashMap<String, Queue<TASKIDTYPE>>();
private final Queue<String> qosTurn = new LinkedList<String>();
public boolean add(TASKTYPE wt, TASKIDTYPE taskId, String qosKey) {
if (queuedTasks.containsKey(taskId)) {
throw new IllegalStateException("Task with ID [" + taskId + "] already enqueued.");
}
queuedTasks.put(taskId, wt);
return addToQ(taskId, qosKey);
}
public TASKTYPE poll() {
TASKIDTYPE taskId = pollQos();
return (taskId == null ? null : queuedTasks.get(taskId));
}
/**
* This method must be called after a task is taken from the queue
* using {#link #poll()} and executed.
*/
public TASKTYPE remove(TASKIDTYPE taskId) {
TASKTYPE wt = queuedTasks.remove(taskId);
if (wt != null) {
taskQSize.decrementAndGet();
}
return wt;
}
private boolean addToQ(TASKIDTYPE taskId, String qosKey) {
if (qosKey == null || qosKey.equals(EMPTY_STRING) || size() < getQosThreshold()) {
defaultQ.add(taskId);
} else {
addSynced(taskId, qosKey);
}
taskQSize.incrementAndGet();
return true;
}
private void addSynced(TASKIDTYPE taskId, String qosKey) {
qosKeyLock.lock();
try {
Queue<TASKIDTYPE> qosQ = qosQueues.get(qosKey);
if (qosQ == null) {
if (!isUsingQos()) {
// Setup QoS mechanics
qosTurn.clear();
qosTurn.add(EMPTY_STRING);
usingQos = true;
}
qosQ = new LinkedList<TASKIDTYPE>();
qosQ.add(taskId);
qosQueues.put(qosKey, qosQ);
qosTurn.add(qosKey);
log.trace("Created QoS queue for {}", qosKey);
} else {
qosQ.add(taskId);
if (log.isTraceEnabled()) {
log.trace("Added task to QoS queue {}, size: " + qosQ.size(), qosKey);
}
}
} finally {
qosKeyLock.unlock();
}
}
private TASKIDTYPE pollQos() {
TASKIDTYPE taskId = null;
qosKeyLock.lock();
try {
taskId = pollQosRecursive();
} finally {
qosKeyLock.unlock();
}
return taskId;
}
/**
* Poll the work task queues according to qosTurn.
* Recursive in case empty QoS queues are removed or defaultQ is empty.
* #return
*/
private TASKIDTYPE pollQosRecursive() {
if (!isUsingQos()) {
// QoS might have been disabled before lock was released or by this recursive method.
return defaultQ.poll();
}
String qosKey = qosTurn.poll();
Queue<TASKIDTYPE> qosQ = (qosKey.equals(EMPTY_STRING) ? defaultQ : qosQueues.get(qosKey));
TASKIDTYPE taskId = qosQ.poll();
if (qosQ == defaultQ) {
// DefaultQ should always be checked, even if it was empty
qosTurn.add(EMPTY_STRING);
if (taskId == null) {
taskId = pollQosRecursive();
} else {
log.trace("Removed task from defaultQ.");
}
} else {
if (taskId == null) {
qosQueues.remove(qosKey);
if (qosQueues.isEmpty()) {
usingQos = false;
}
taskId = pollQosRecursive();
} else {
qosTurn.add(qosKey);
if (log.isTraceEnabled()) {
log.trace("Removed task from QoS queue {}, size: " + qosQ.size(), qosKey);
}
}
}
return taskId;
}
#Override
public String toString() {
StringBuilder sb = new StringBuilder(this.getClass().getName());
sb.append(", size: ").append(size());
sb.append(", number of QoS queues: ").append(qosQueues.size());
return sb.toString();
}
public boolean containsTaskId(TASKIDTYPE wid) {
return queuedTasks.containsKey(wid);
}
public int size() {
return taskQSize.get();
}
public void setQosThreshold(int size) {
this.qosThreshold = size;
}
public int getQosThreshold() {
return qosThreshold;
}
public boolean isUsingQos() {
return usingQos;
}
}