Develop Reference

How to create a non durable queue in ActiveMQ Artemis?

I have a application in which I want to have 1 durable and 1 non-durable queue in Active MQ Artemis.
For connecting to this message bus I use amqpnetlite.
var source = new Source()
{
};
if (durable)
{
source.Address = amqpAddressConverter.GetSubscriberAddress(address, useLoadBalancing);
source.Durable = 1;
source.ExpiryPolicy = new Symbol("never");
source.DistributionMode = new Symbol("copy");
}
else
{
source.Address = amqpAddressConverter.GetSubscriberAddress(address);
source.Durable = 0;
source.ExpiryPolicy = "never";
}
var receiverLink = new ReceiverLink(session, linkName, source, null);
So this is my receiver link. As shown I set the Durable uint of the Source which will given into the ReceiverLink.
Because as I saw in the Active MQ Artemis documentation, that the Durable is a boolean but within the amqpnetlite library it is an uint my understanding is that everything over 0 should be true and 0 should be false.
At first the behaviour was very strange: Even when the Aretemis Web interface was shown a queue as durable it would be deleted as soon as no consumer would be connected.
I found this:
ActiveMQ Artemis queue deleted after shutdown of consuming client
which describes that even durable queues get deleted because of the default behaviour.
So I manipulated the broker.xml and set AUTO-DELETE-QUEUE to false.
Since then the behaviour completly switched:
Both (durable = 1 and durable = 0) queues are being still there after the connection disconnected.
So how to create a durable and a non-durable connection correctly?

The Artemis source carries an example in .NET that creates a durable topic subscription and also shows how to later recover it using AmqpNetLite.
One key thing many folks miss is that your client needs to use a unique container ID analogous to the JMS Client ID concept.
For Queue specific subscriptions the client should indicate in the link capabilities that it wants a Queue based address created as the default is a multicast Queue which won't behave the same.
Source source = new Source() {
Address = address,
Capabilities = new Symbol[] {"queue"},
};
vs topic specific source configuration:
Source source = new Source() {
Address = address,
Capabilities = new Symbol[] {"topic"},
};

Kafka very high latency C#

I am doing some performance tests on Apache Kafka to compare it with others like RabbitMQ and ActiveMQ. The idea is to use it on a messaging system for agents' communication.
I am testing multiple scenarios (one to one, broadcast and many to one) with different numbers of publishers and subscribers and so different loads. Even in the lowest load scenario of one to one with 10 pairs of agents sending 500 messages with 1ms delay between sends I am experiencing very high latencies (average of ~200ms). And if we go to 100 pairs the numbers rise to ~1500ms. The same thing happens on broadcast and many to one.
I am using Windows with Kafka 2.12-2.5.0 and zookeeper 3.6.1 with C# .Net client Confluent.Kafka 1.4.2. I have already tried some properties like LingerMs = 0 according to some posts I found. I have both Kafka and zookeeper with default settings.
I made a simple test code in which the problem happens:
using System;
using System.Collections.Generic;
using System.IO;
using System.Linq;
using System.Text;
using System.Threading;
using System.Threading.Tasks;
using Confluent.Kafka;
namespace KafkaSetupAgain
{
class Program
{
static void Main(string[] args)
{
int numberOfMessages = 500;
int numberOfPublishers = 10;
int numberOfSubscribers = 10;
int timeOfRun = 30000;
List<MCVESubscriber> Subscribers = new List<MCVESubscriber>();
for (int i = 0; i < numberOfSubscribers; i++)
{
MCVESubscriber ZeroMqSubscriber = new MCVESubscriber();
new Thread(() =>
{
ZeroMqSubscriber.read(i.ToString());
}).Start();
Subscribers.Add(ZeroMqSubscriber);
}
Thread.Sleep(10000);//to make sure all subscribers started
for (int i = 0; i < numberOfPublishers; i++)
{
MCVEPublisher ZeroMqPublisherBroadcast = new MCVEPublisher();
new Thread(() =>
{
ZeroMqPublisherBroadcast.publish(numberOfMessages, i.ToString());
}).Start();
}
Thread.Sleep(timeOfRun);
foreach (MCVESubscriber Subscriber in Subscribers)
{
Subscriber.PrintMessages("file.csv");
}
}
public class MCVEPublisher
{
public void publish(int numberOfMessages, string topic)
{
var config = new ProducerConfig
{
BootstrapServers = "localhost:9092",
LingerMs = 0,
Acks = 0,
};
var producer = new ProducerBuilder<Null, string>(config).Build();
int success = 0;
int failure = 0;
Thread.Sleep(3500);
for (int i = 0; i < numberOfMessages; i++)
{
Thread.Sleep(1);
long milliseconds = System.Diagnostics.Stopwatch.GetTimestamp() / TimeSpan.TicksPerMillisecond;
var t = producer.ProduceAsync(topic, new Message<Null, string> { Value = milliseconds.ToString() });
t.ContinueWith(task => {
if (task.IsFaulted)
{
failure++;
}
else
{
success++;
}
});
}
Console.WriteLine("Success: " + success + " Failure:" + failure);
}
}
public class MCVESubscriber
{
private List<string> prints = new List<string>();
public void read(string topic)
{
var config = new ConsumerConfig()
{
BootstrapServers = "localhost:9092",
EnableAutoCommit = false,
FetchErrorBackoffMs = 1,
};
var consumerConfig = new ConsumerConfig(config);
consumerConfig.GroupId = Guid.NewGuid().ToString();
consumerConfig.AutoOffsetReset = AutoOffsetReset.Earliest;
consumerConfig.EnableAutoCommit = false;
using (var consumer = new ConsumerBuilder<Ignore, string>(config).Build())
{
consumer.Subscribe(new[] { topic });
while (true)
{
var consumeResult = consumer.Consume();
long milliseconds = System.Diagnostics.Stopwatch.GetTimestamp() / TimeSpan.TicksPerMillisecond;
prints.Add(consumeResult.Message.Value + ";" + milliseconds.ToString());
}
consumer.Close();
}
}
public void PrintMessages(string path)
{
Console.WriteLine("printing " + prints.Count);
File.AppendAllLines(path, prints);
}
}
}
}
Does someone what can be the problem? What configs can I change to
improve latency?
Thanks,
Davide Costa

Kafka is not really built for low latency message distribution, but for high availability. It can be configured to have lower latency, but you start losing a lot of the advantages Kafka offers.
A few tips/comments below:
On the KafkaProducer side, in general, you want to wait until there's enough messages to send, to as to batch messages more efficiently. That's the linger.ms property you already mentioned. Typically that is set to something like 50ms, so by setting it to zero, you're effectively telling the producer to send data as fast as it gets it. This may make the producer more "chatty", but you have the assurance it will send the data to the cluster as soon as it gets it.
However, once a message is "produced" into Kafka, it waits until it gets an ACK from the lower layer that the broker has received the message successfully. There's multiple options here:
Consider a message as "received" once the message has been sent by the producer. That is, locally, once the network layer has finished sending it, the producer will consider it "sent and acknowledged"
Wait for an ACK from the leader broker which you're sending the message to, depending on which partition it gets assigned, so you at least know one broker has it. THIS IS THE DEFAULT.
Wait for an ACK from the leader broker which you're sending the message to, PLUS an ACK from each of that partitions' replicas on the other brokers. This means, if your cluster has a replication factor of 3, that the message is sent to broker 1 for example, it then replicates that to brokers 2 and 3, which have copies of the same partition, waits for those brokers to reply back saying they got the message, and only THEN reply back to the producer saying the message has been ACK'd. This is typically used in environments where you never want the possibility of losing a single message, so you always guarantee that there will be three copies of your message before the producer moves on.
Official acks explanation from the Kafka docs:
https://kafka.apache.org/25/documentation.html#acks
There are other settings to consider like kafka producer compression and broker compression settings that might add more latency/overhead, but if you're using the defaults (no producer compression and producer option in the broker compression), there should be no additional latency in those steps.
Having said all that, I would suggest you try to set the acks option in the producer to 0, and see how your latency changes. My guess is you will get much better latency, BUT also understand that there are no guarantees your messages are actually being received and stored correctly. A flaky network, a network partition, etc, could cause you to lose data. That might be ok for your use case, but just make sure you're aware of it.

How can I both Send and Receive from a Router socket in ZeroMQ or NetMQ?

I have a Dealer <--> Router setup in NetMQ v4 where I can asynchronously send and receive messages in any direction with no problems.
I now want to formalize that into an abstraction where the server (Router) listens for any incoming message but it also need to on demand broadcasts messages to any of the connected clients (Dealers).
I am trying to avoid using Pub <--> Sub sockets as I need the subscribers to also send messages to the server. The closest pattern to what I am trying to achieve is a WebSocket client-server communication.
The first part of listening to the client messages are done in something like:
using (var server = new RouterSocket("#tcp://*:80"))
{
var addresses = new HashSet<string>();
while (true)
{
var msg = server.ReceiveMultipartMessage();
var address = Encoding.UTF8.GetString(msg[0].Buffer);
var payload = Encoding.UTF8.GetString(msg[2].Buffer);
Console.WriteLine("[Server] - Client: {0} Says: {1}", address, payload);
var contains = addresses.Contains(address);
if (!contains) { addresses.Add(address); }
msg.Clear();
msg.Append(address);
msg.AppendEmptyFrame();
msg.Append("Reply for: " + address);
server.SendMultipartMessage(msg);
}
}
Now given that the sockets are not Thread-Safe, I am stuck on finding a way to broadcasts messages (coming from a different thread on demand) to all the clients.
I can probably use the TryReceiveMultipartMessage method in the loop instead with a set timeout after which I can check a queue for any broadcast messages and then loop through each client sending such message. Something like:
using (var server = new RouterSocket("#tcp://*:80"))
{
var addresses = new HashSet<string>();
var msg = new NetMQMessage();
while (true)
{
var clientHasMsg = server.TryReceiveMultipartMessage(TimeSpan.FromSeconds(1), ref msg);
if (!clientHasMsg)
{
// Check any incoming broacast then loop through all the clients
// sending each the brodcast msg
var broadMsg = new NetMQMessage();
foreach (var item in addresses)
{
broadMsg.Append(item);
broadMsg.AppendEmptyFrame();
broadMsg.Append("This is a broadcast");
server.SendMultipartMessage(broadMsg);
broadMsg.Clear();
}
// Go back into the loop waiting for client messages
continue;
}
var address = Encoding.UTF8.GetString(msg[0].Buffer);
var payload = Encoding.UTF8.GetString(msg[2].Buffer);
Console.WriteLine("[Server] - Client: {0} Says: {1}", address, payload);
var contains = addresses.Contains(address);
if (!contains) { addresses.Add(address); }
msg.Clear();
msg.Append(address);
msg.AppendEmptyFrame();
msg.Append("Reply for: " + address);
server.SendMultipartMessage(msg);
}
}
This somehow does not feel right mainly due to:
What value for the timeout is a good value? 1 sec, 100 ms etc;
Is this the most efficient/performing solution as this program will be used to have 100k+ clients connected with each sending thousands of messages per second.
Any pointers on what's the best approach to this is very much appreciated.

You can use netmqqueue, it multi producer single consumer queue. You can add it to NetMQPoller and enqueue from multiple threads without lock.

I think PUB/SUB is a proper approach for your requirements of 100k+ clients. Nevertheless, it doesn't mean that you can't communicate back to server: use DEALER/ROUTER. Why do you think this solution is not acceptable?

ZeroMQ/0MQ Push/Pull memory and routing issues

I have played a while with ZeroMQ and have couple questions/problems that I came up with. Would appreciate if any contributer to ZeroMQ could chime in or anyone who has used or currently uses the library.
* Let's say I have one router/forwarder and 2 different clients(c1,c2). I want to push messages from client1 to client2 through the routing device. The router pulls messages from whichever client (here client1) and publishes them to any subscribed client (here client2). I currently the only way to route such messages to the appropriate client is through pub/sub, however , a) I want to decide how to route at runtime by sending a routingTo tag along with the message body, b) I want to use push/pull to forward to clients, not pub/sub because I want to implement blocking functionality when setting the high water mark property, c) I want to have c1 and c2 connect on exactly 1 port for pushing and 1 port for subscribing. Can I somehow make changes on the router side in order to not having to use pub/sub or is pub/sub the only way to route to clients even I know on the routing side where a message is supposed to be forwarded to? I read that pub/sub drops messages when queue size exceeds the hwm which I dont want. I also do not want to implement the request/reply patters because it adds unnecessary overhead as I do not need replies.
* After running below code (Push/Pull -> Pub/Sub) and sent all messages and have received confirmation that all messages were received the client that pushed messages out still displays a huge memory footprint, apparently there are still huge amounts of messages in the Push socket's queue. Why is that and what can I do to fix that?
Here is my code:
ROUTER:
class Program
{
static void Main(string[] args)
{
using (var context = new Context(1))
{
using (Socket socketIn = context.Socket(SocketType.PULL), socketOut = context.Socket(SocketType.XPUB))
{
socketIn.HWM = 10000;
socketOut.Bind("tcp://*:5560"); //forwards on this port
socketIn.Bind("tcp://*:5559"); //listens on this port
Console.WriteLine("Router started and running...");
while (true)
{
//Receive Message
byte[] address = socketIn.Recv();
byte[] body = socketIn.Recv();
//Forward Message
socketOut.SendMore(address);
socketOut.Send(body);
}
}
}
}
}
CLIENT1:
class Program
{
static void Main(string[] args)
{
using (var context = new Context(1))
{
using (Socket socketIn = context.Socket(SocketType.SUB), socketOut= context.Socket(SocketType.PUSH))
{
byte[] iAM = Encoding.Unicode.GetBytes("Client1");
byte[] youAre = Encoding.Unicode.GetBytes("Client2");
byte[] msgBody = new byte[16];
socketOut.HWM = 10000;
socketOut.Connect("tcp://localhost:5559");
socketIn.Connect("tcp://localhost:5560");
socketIn.Subscribe(iAM);
Console.WriteLine("Press key to kick off Test Client1 Sending Routine");
Console.ReadLine();
for (int counter = 1; counter <= 10000000; counter++)
{
//Send Message
socketOut.SendMore(youAre);
socketOut.Send(msgBody);
}
Console.WriteLine("Client1: Finished Sending");
Console.ReadLine();
}
}
}
}
CLIENT2:
class Program
{
public static int msgCounter;
static void Main(string[] args)
{
msgCounter = 0;
using (var context = new Context(1))
{
using (Socket socketIn = context.Socket(SocketType.SUB), socketOut = context.Socket(SocketType.PUSH))
{
byte[] iAM = Encoding.Unicode.GetBytes("Client2");
socketOut.Connect("tcp://localhost:5559");
socketIn.Connect("tcp://localhost:5560");
socketIn.Subscribe(iAM);
Console.WriteLine("Client2: Started Listening");
//Receive First Message
byte[] address = socketIn.Recv();
byte[] body = socketIn.Recv();
msgCounter += 1;
Console.WriteLine("Received first message");
Stopwatch watch = new Stopwatch();
watch.Start();
while (msgCounter < 10000000)
{
//Receive Message
address = socketIn.Recv();
body = socketIn.Recv();
msgCounter += 1;
}
watch.Stop();
Console.WriteLine("Elapsed Time: " + watch.ElapsedMilliseconds + "ms");
Console.ReadLine();
}
}
}
}

I'm going to suggest that your architecture may be a bit off here.
1) If you need exactly one PUSH and exactly one PULL, remove the device from the middle. Devices are added to an architecture explicitly to mange multiple consumers so that you don't have to update producers each time you add a node. When/If you do get to where you need multiple consumers and/or producers, you're going to need a connection to each node on your device - that's just how they work. In this case, it sounds as though the device is overly complicating your solution.
2) The idea of having the "route to" tag really boggles my mind. Probably the biggest reason to choose messaging over other integration options is to decouple your producers and consumers so that neither side has to know anything about the other (other than where to send the messages in the case of broker-less designs). Adding routing information directly to your logic breaks this.
As to the overhead, I've never experienced this. But then, I've never used the .Net driver for ZeroMQ before and so an uneducated guess would be to look at the .Net driver itself.

Message Queue Exception : Queue does not exist or you do not have sufficient permissions to perform the operation

At this line of code i am getting the error as i mentioned
I declared MSMQ_NAME as string as follows
private const string MSMQ_NAME = ".\\private$\\ASPNETService";
private void DoSomeMSMQStuff()
{
using (MessageQueue queue = new MessageQueue(MSMQ_NAME))
{
queue.Send(DateTime.Now); //Exception raises
queue.Close();
}
}

Can you first verify the queue is existing with the name 'ASPNETService' at below location?
Computer Management -> Services and Applications -> Message Queuing -> Private Queues

I had a similar problem. I was confused because my code worked on my local development machine, but not in production. Even stranger, the queues were created the exact same way.
It turns out that IIS doesn't have access to them by default. I just opened up the permissions.
Computer Management -> Private Queues -> right-click queue name -> Properties -> Security Tab -> click "Everyone" user -> click Full Control/Allow checkbox -> click OK
This fixed it for me, and in my case it's not an issue, but you may want to think about the ramifications of just opening it up for all users.
Also, I had to do this across all queues on all servers. There doesn't seem to be a way to multi-select queues or folders in order to set permissions for multiple queues simultaneously.

I was having the same problem.
I had created a new private queue and gave Full Permission to Everyone.
But I was still catching a "Queue does not exist or you do not have sufficient permissions to perform the operation" when trying to Send() to the queue. And I was able to verify that MessageQueue.Exists(".\\private$\\myqueue") was returning true.
Restarting the Message Queuing Service resolved my the problem for me.

I had same problem and I did like below where I check whether queue exists or not. If yes send message else create queue and then send message
MessageQueue msgQueue = null;
string queuePath = ".\\Private$\\billpay";
Payment newPayment = new Payment()
{
Payee = txtPayee.Text,
Payor = txtPayor.Text,
Amount = Convert.ToInt32(txtAmount.Text),
DueDate = dpDueDate.SelectedDate.Value.ToShortDateString()
};
Message msg = new Message();
msg.Body = newPayment;
msg.Label = "Gopala - Learning Message Queue";
if (MessageQueue.Exists(queuePath) == false)
{
//Queue doesnot exist so create it
msgQueue = MessageQueue.Create(queuePath);
}
else
{
msgQueue = new MessageQueue(queuePath);
}
msgQueue.Send(msg);

I was facing the same problem, I had resolved it using the following class to create queue
private MessageQueue messageQueue;
public const string DEFAULT_QUEUE_NAME = "newQueue";
public const string QUEUENAME_PREFIX = ".\\Private$\\";
public static string QueueName
{
get
{
string result = string.Format("{0}{1}", QUEUENAME_PREFIX, DEFAULT_QUEUE_NAME);
return result;
}
}
public void SendMessage()
{
string queuePath = QueueName;
MessageQueue messageQueue = MessageQueue.Create(queuePath);
messageQueue.Send("msg");
}
Create message queue in same manner for receiving the message.

For others struggling with this and pulling their hair out like I have been, I finally found something that works when all of the upvoted suggestions failed.
Even if you think the host name of your target queue's hosting system is being resolved correctly, don't believe it. Try replacing the host name with an IP address and see if it works. It does for me. I can WRITE to a public queue using a host name on my remote server without problems, but trying to READ from it produces exactly the error listed for this question.
For example, if I declare the following:
private static string QueueName = #"FormatName:DIRECT=TCP:SOMEHOST\MyQueue";
private static System.Messaging.MessageQueue Queue = new System.Messaging.MessageQueue(QueueName);
Where "MyQueue" is a public queue on server SOMEHOST, the following code will successfully insert messages to the queue, but always fails on the Receive():
Queue.Formatter = new XmlMessageFormatter(new Type[] { typeof(String) });
// The Receive() call here is a blocking call. We'll wait if there is no message in the queue, and processing
// is halted until there IS a message in the queue.
//
try
{
Queue.Send("hello world", System.Messaging.MessageQueueTransactionType.Single);
var msg = Queue.Receive(MessageQueueTransactionType.Single);
}
catch (Exception ex)
{
// todo error handling
}
One simple change in how I specify the queue location is all that's needed to make the Receive() stop failing with the dreaded "queue does not exist or you do not have sufficient permissions" error:
private static string QueueName = #"FormatName:DIRECT=TCP:192.168.1.100\MyQueue";
(Obviously I've obfuscated IP addresses and other sensitive info). Using the IP address is not obviously a production-worthy scenario, but it did point me to some type of name resolution problem as being the possible cause of the error. I cannot explain why Send() works but Receive() does not when I am using a host name instead of IP, but I can reproduce these results consistently. Until I can figure out what's going on with the name resolution, I'm no longer wasting a day trying to read messages from a queue.