I'm working on an application that is divided in a thin client and a server part, communicating over TCP. We frequently let the server make asynchronous calls (notifications) to the client to report state changes. This avoids that the server loses too much time waiting for an acknowledgement of the client. More importantly, it avoids deadlocks.
Such deadlocks can happen as follows. Suppose the server would send the state-changed-notification synchronously (please note that this is a somewhat constructed example). When the client handles the notification, the client needs to synchronously ask the server for information. However, the server cannot respond, because he is waiting for an answer to his question.
Now, this deadlock is avoided by sending the notification asynchronously, but this introduces another problem. When asynchronous calls are made more rapidly than they can be processed, the call queue keeps growing. If this situation is maintained long enough, the call queue will get totally full (flooded with messages). My question is: what can be done when that happens?
My problem can be summarized as follows. Do I really have to choose between sending notifications without blocking at the risk of flooding the message queue, or blocking when sending notifications at the risk of introducing a deadlock? Is there some trick to avoid flooding the message queue?
Note: To repeat, the server does not stall when sending notifications. They are sent asynchronously.
Note: In my example I used two communicating processes, but the same problem exists with two communicating threads.
If the server is sending informational messages to the client, which you yourself say are asynchronous, it should not have to wait for a reply from the client. If they are not informational, in other words they require an answer, I would say a server should never send such messages to a client, and their presence indicates a poor design.
If you have a constant congestion problem, there is little you can do other than gracefully fail and notify the client that no new messages can be posted; then it is up to the client to maintain a backlog of messages to be posted.
Introducing a priority queue and using message expiration/filtering could allow you to free up space in the queue, but that really just postpones the problem. If possible, you could also aggregate messages or ignore duplicate messages, but again the problem does not seem to be the queue itself. (Not to mention that the more complex queue logic could eat up valuable resources that would be better used actually processing messages.)
Depending on what the server side does, you could introduce result hashing for long computations, offload some types of messages to a dedicated device, check if the server waits unreasonably long for I/O operations, and a myriad of other techniques. Profile if possible, at least try to find out which message(s) causes congestion.
Oh, and the business solution: Compare cost of estimated development time to the cost of better hardware and conclude that you should just buy a more powerful server (or an additional one).
Depending on how important these messages are you might want to look into Message Expiration, or perhaps a Message Filter, though it sounds like your architecture may be incorrect.
I would rather fix the logic in the server side. The message queue should not stall waiting for the answer. Rather have a state machine which can also receive those info queries while it is waiting for the answer from the client.
Of course you can still flood your message queue, but with TCP you can handle it pretty easily.
The best way, I believe, would be to add another state to your client. This I borrowed from the SMPP protocol specs.
Add a congestion state to the client, whereby it always checks the queue length, assuming this is possible, and therefore once a certain threshold is attained, say 1000 unprocessed messages, the client sends the server a message indicating that it's congested and the server will be required to cease all messaging until it receives a notification indicating that the client is no longer congested.
Alternatively, on the server side, if there is a certain number of pending replies, the server could simply cease sending messages until the client replies a certain number of them.
These thresholds can be dynamically calculated or fixed, depending.....
Related
I'm having issues with Node.js and the "WS" implementation of websocket (https://www.npmjs.com/package/ws). After a surge (plenty of messages in a short window of time), I'm having data that suggests that I've "missed" a message.
I've contacted the owner of the emitter server and he assures me that all messages have been sent on his side.
I've logger every message received on my side (at the start of the function on('message', () => {}), and I can't seem to find the message, so my assumption is that it doesn't even reached this point
So I'm wondering:
Messages are reveived and treated in a FIFO order. During the treatment of the current message, new ones will be stacked in the node event loop to be computed immediatly after. Correct ? Is there a way for that event loop to be "too big" that may drop new incomming messages ? If so, does it drop it quietly ? or does the program crashes vigorously (in other words, how can I see if a message has been dropped this way ?)
Does the 'ws' module have any kind of kown limitations for a maximum number of message received ? Does it have an internal way of dropping messages ?
Is there a better alternative than the 'ws' module ?
Is there any other ways to explain a "missed" message ?
Thanks a lot for your insights,
I use ws in nodejs to handle large message flows from many clients simultaneously in production, and I have never had it lose messages. Each server handles several thousand messages each second from hundreds of different client connections. The way my system works, if ws dropped messages or changed their order, my users would complain loudly.
That makes me guess you are not hitting any limitation of ws.
Early in my programming work I had the not-so-bright idea of putting incoming messages in queue objects in my nodejs code and processing them "later." That led to a hideously confusing message flow through my server. It sometimes looked like I had lost ws messages. I was happy to delete all that code, and dispatch every message completely within its message event handler.
Websocket connections sometimes close abnormally. Because network. You can catch those situations with error and close event handlers. It can take a while for the sender of a message, or the receiver, to detect that a network fault of some kind disrupted its connection. That can lead to disagreement about message count between sender and receiver. It's worth investigating.
I adorn ws's connection objects with message counts ("adorn" -- add an application-specific property to an object) and put those message counts into the log when a connection closes.
I've got a basic knowledge from here about epoll. I know that epoll can monitor multiple FDs and handle them.
My question is: can a heavy event block the server so I must use multithreading?
For example, the epoll of a server is monitoring 2 sockets A and B. Now A starts to send lot of messages to the server so the server starts to read them. One second later, B starts to send messages too while A is still sending. In this case, Need I create a thread for these read actions? If I don't, does it mean that the server has no chance to get the messages from B until A finishes its sending?
If you can process incoming messages fast enough (no blocking calls, no heavy computations), you don't need a separate thread. Otherwise, you would benefit from going multi-threaded.
In any case, it helps to understand what happens when you have only one thread and you can't process messages fast enough. If you are working with TCP protocol, the machines sending you the data will simply reduce their transmission rate. When using UDP, some incoming packets will get dropped.
We are using socketIO on a large chat application.
At some points we want to dispatch "presence" (user availability) to all other users.
io.in('room1').emit('availability:update', {userid='xxx', isAvailable: false});
room1 may contains a lot of users (500 max). We observe a significant raise in our NodeJS load when many availability updates are triggered.
The idea was to use something similar to redis store with Socket IO. Have web browser clients to connect to different NodeJS servers.
When we want to emit to a room we dispatch the "emit to room1" payload to all other NodeJS processes using Redis PubSub ZeroMQ or even RabbitMQ for persistence. Each process will itself call his own io.in('room1').emit to target his subset of connected users.
One of the concern with this setup is that the inter-process communication may become quite busy and I was wondering if it may become a problem in the future.
Here is the architecture I have in mind.
Could you batch changes and only distribute them every 5 seconds or so? In other words, on each node server, simply take a 'snapshot' every X seconds of the current state of all users (e.g. 'connected', 'idle', etc.) and then send that to the other relevant servers in your cluster.
Each server then does the same, every 5 seconds or so it sends the same message - of only the changes in user state - as one batch object array to all connected clients.
Right now, I'm rather surprised you are attempting to send information about each user as a packet. Batching seems like it would solve your problem quite well, as it would also make better use of standard packet sizes that are normally transmitted via routers and switches.
You are looking for this library:
https://github.com/automattic/socket.io-redis
Which can be used with this emitter:
https://github.com/Automattic/socket.io-emitter
About available users function, I think there are two alternatives,you can create a "queue Users" where will contents "public data" from connected users or you can use exchanges binding information for show users connected. If you use an "user's queue", this will be the same for each "room" and you could update it when an user go out, "popping" its state message from queue (Although you will have to "reorganize" all queue message for it).
Nevertheless, I think that RabbitMQ is designed for asynchronous communication and it is not very useful approximation have a register for presence or not from users. I think it's better for applications where you don't know when the user will receive the message and its "real availability" ("fire and forget architectures"). ZeroMQ require more work from zero but you could implement something more specific for your situation with a better performance.
An publish/subscribe example from RabbitMQ site could be a good point to begin a new design like yours where a message it's sent to several users at same time. At summary, I will create two queues for user (receive and send queue messages) and I'll use specific exchanges for each "room chat" controlling that users are in each room using exchange binding's information. Always you have two queues for user and you create exchanges to binding it to one or more "chat rooms".
I hope this answer could be useful for you ,sorry for my bad English.
This is the common approach for sharing data across several Socket.io processes. You have done well, so far, with a single process and a single thread. I could lamely assume that you could pick any of the mentioned technologies for communicating shared data without hitting any performance issues.
If all you need is IPC, you could perhaps have a look at Faye. If, however, you need to have some data persisted, you could start a Redis cluster with as many Redis masters as you have CPUs, though this will add minor networking noise for Pub/Sub.
I'm surely missing something about how the whole MQTT protocol works, as I can't grasp the usage pattern of Last Will Testament messages: what's their purpose?
One example I often see is about informing that a device has gone offline. It doesn't make very much sense to me, since it's obvious that if a device isn't publishing any data it may be offline or there could be some network problems.
So, what are some practical usages of the LWT? What was it invented for?
LWT messages are not really concerned about detecting whether a client has gone offline or not (that task is handled by keepAlive messages).
LWT messages are about what happens after the client has gone offline.
The analogy is that of a real last will:
If a person dies, she can formulate a testament, in which she declares what actions should be taken after she has passed away. An executor will heed those wishes and execute them on her behalf.
The analogy in the MQTT world is that a client can formulate a testament, in which it declares what message should be sent on it's behalf by the broker, after it has gone offline.
A fictitious example:
I have a sensor, which sends crucial data, but very infrequently.
It has formulated a last will statement in the form of [topic: '/node/gone-offline', message: ':id'], with :id being a unique id for the sensor. I also have a emergency-subscriber for the topic 'node/gone-offline', which will send a SMS to my phone every time a message is published on that channel.
During normal operation, the sensor will keep the connection to the MQTT-broker open by sending periodic keepAlive messages interspersed with the actual sensor readings. If the sensor goes offline, the connection to the broker will time out, due to the lack of keepAlives.
This is where LWT comes in: If no LWT is specified, the broker doesn't care and just closes the connection. In our case however, the broker will execute the sensor's last will and publish the LWT-message '/node/gone-offline: :id'. The message will then be consumed to my emergency-subscriber and I will be notified of the sensor's ID via SMS so that I can check up on what's going on.
In short:
Instead of just closing the connection after a client has gone offline, LWT messages can be leveraged to define a message to be published by the broker on behalf of the client, since the client is offline and cannot publish anymore.
Just because a device is not publishing does not mean it is not online or there is a network problem.
Take for example a sensor that monitors a value that only changes very infrequently, good design says that the sensor should only publish the changes to help reduce bandwidth usage as periodically publishing the same value is wasteful. If the value is published as a retained value then any new subscriber will always get the current value without having to wait for the sensor value to change and it publish again.
In this case the LWT is used to published when the sensor fails (or there is a network problem) so we know of the problem as soon at the client keep alive times out.
A in-depth article about Last-Will-and-Testament messages is available in the MQTT Essentials Blog Post series: http://www.hivemq.com/mqtt-essentials-part-9-last-will-and-testament/.
To summarize the blog post:
The Last Will and Testament feature is used in MQTT to notify other clients about an ungracefully disconnected client.
MQTT is often used in scenarios were unreliable networks are very common. Therefore it is assumed that some clients will disconnect ungracefully from time to time, because they lost the connection, the battery is empty or any other imaginable case. It would be good to know if a connected client has disconnected gracefully (which means with a MQTT DISCONNECT message) or not, in order to take appropriate action.
I have to listen for rpc calls , stack them somewhere , process them, and answer. The thing is that they are not run as soon as they come. The response is an ACK for each rpc call recieved.
The problem is that i want to design it in a way that i can have many listening servers writing in the same stack of calls, piling them up as they come.
My objective is to listen to as many calls as possible. How should i achieve this?
My main technology is Perl and node.js but would use any open source software for this task.
It sounds like any kind of job queue will do what you need it to; I'm personally a big fan of using Redis for this kind of thing. Since Redis lists maintain insertion order, you can simply LPUSH your RPC call info on to the end of the list from any number of web servers listening to the RPC calls, and somewhere else (in another process/on another machine, I assume) RPOP (or BRPOP) them off and process them.
Since Node.js uses fully asynchronous IO, assuming you're not doing a lot of processing in your RPC listeners (that is, you're only listening for requests, sending an ACK, and pushing onto Redis), my guess is that Node would be exceedingly efficient at this.
An aside on using Redis for a queue: if you want to ensure that, in the event of a catastrophic failure, jobs are not lost, you'll need to implement a little more logic; from the RPOPLPUSH documentation:
Pattern: Reliable queue
Redis is often used as a messaging server to implement processing of background jobs or other kinds of messaging
tasks. A simple form of queue is often obtained pushing values into a
list in the producer side, and waiting for this values in the consumer
side using RPOP (using polling), or BRPOP if the client is better
served by a blocking operation.
However in this context the obtained
queue is not reliable as messages can be lost, for example in the case
there is a network problem or if the consumer crashes just after the
message is received but it is still to process.
RPOPLPUSH (or
BRPOPLPUSH for the blocking variant) offers a way to avoid this
problem: the consumer fetches the message and at the same time pushes
it into a processing list. It will use the LREM command in order to
remove the message from the processing list once the message has been
processed.
An additional client may monitor the processing list for
items that remain there for too much time, and will push those timed
out items into the queue again if needed.