I'm relatively new to Akka & Scala, but I would like to use Akka as a generic framework to pull together information from various web tools, and cli commands.
I understand the general principal that in an Actor model, it is highly desirable not to have the actors block. And in the case of the http requests, there are async http clients (such as Spray) that means that I can handle the requests asynchronously within the Actor framework.
However, I'm unsure what is the best approach when combining actors with existing blocking API calls such as the scala ProcessBuilder/ProcessIO libraries. In terms of issuing these CLI commands I expect a relatively small amount of concurrency, e.g. perhaps executing a max of 10 concurrent CLI invocations on a 12 core machine.
Is it better to have a single actor managing these CLI commands, farming the actual work off to Futures that are created as needed? Or would it be cleaner just to maintain a set of separate actors backed by a PinnedDispatcher? Or something else?
From the Akka documentation ( http://doc.akka.io/docs/akka/snapshot/general/actor-systems.html#Blocking_Needs_Careful_Management ):
"
Blocking Needs Careful Management
In some cases it is unavoidable to do blocking operations, i.e. to put a thread to sleep for an indeterminate time, waiting for an external event to occur. Examples are legacy RDBMS drivers or messaging APIs, and the underlying reason in typically that (network) I/O occurs under the covers. When facing this, you may be tempted to just wrap the blocking call inside a Future and work with that instead, but this strategy is too simple: you are quite likely to find bottle-necks or run out of memory or threads when the application runs under increased load.
The non-exhaustive list of adequate solutions to the “blocking problem” includes the following suggestions:
Do the blocking call within an actor (or a set of actors managed by a router [Java, Scala]), making sure to configure a thread pool which is either dedicated for this purpose or sufficiently sized.
Do the blocking call within a Future, ensuring an upper bound on the number of such calls at any point in time (submitting an unbounded number of tasks of this nature will exhaust your memory or thread limits).
Do the blocking call within a Future, providing a thread pool with an upper limit on the number of threads which is appropriate for the hardware on which the application runs.
Dedicate a single thread to manage a set of blocking resources (e.g. a NIO selector driving multiple channels) and dispatch events as they occur as actor messages.
The first possibility is especially well-suited for resources which are single-threaded in nature, like database handles which traditionally can only execute one outstanding query at a time and use internal synchronization to ensure this. A common pattern is to create a router for N actors, each of which wraps a single DB connection and handles queries as sent to the router. The number N must then be tuned for maximum throughput, which will vary depending on which DBMS is deployed on what hardware."
Related
I am asking myself the question: "When should you use scala.concurrent.blocking?"
If I understood correctly, the blocking {} only makes sense to be used in conjunction with the ForkJoinPool. In addition docs.scala-lang.org highlights, that blocking shouldn't be used for long running executions:
Last but not least, you must remember that the ForkJoinPool is not designed for long-lasting blocking operations.
I assume a long running execution is a database call or some kind of external IO. In this case a separate thread pools should be used, e.g. CachedThreadPool. Most IO related frameworks, like sttp, doobie, cats can make use of a provided IO thread pool.
So I am asking myself, which use-case still exists for the blocking statement? Is this only useful, when working with locking and waiting operations, like semaphores?
Consider the problem of thread pool starvation. Say you have a fixed size thread pool of 10 available threads, something like so:
implicit val myFixedThreadPool =
ExecutionContext.fromExecutor(Executors.newFixedThreadPool(10))
If for some reason all 10 threads are tied up, and a new request comes in which requires an 11th thread to do its work, then this 11th request will hang until one of the threads becomes available.
blocking { Future { ... } } construct can be interpreted as saying please do not consume a thread from myFixedThreadPool but instead spin up a new thread outside myFixedThreadPool.
One practical use case for this is if your application can conceptually be considered to be in two parts, one part which say in 90% of cases is talking to proper async APIs, but there is another part which in few special cases has to talk to say a very slow external API which takes many seconds to respond and which we have no control over. Using the fixed thread pool for the true async part is relatively safe from thread pool starvation, however also using the same fixed thread pool for the second part presents the danger of the situation where suddenly 10 requests are made to the slow external API, which now causes 90% of other requests to hang waiting for those slow requests to finish. Wrapping those slow requests in blocking would help minimise the chances of 90% of other requests from hanging.
Another way of achieving this kind of "swimlaning" of true async request from blocking requests is by offloading the blocking request to a separate dedicated thread pool to be used just for the blocking calls, something like so
implicit val myDefaultPool =
ExecutionContext.fromExecutor(Executors.newFixedThreadPool(10))
val myPoolForBlockingRequests =
ExecutionContext.fromExecutor(Executors.newFixedThreadPool(20))
Future {
callAsyncApi
} // consume thread from myDefaultPool
...
Future {
callBlockingApi
}(myPoolForBlockingRequests) // consume thread from myPoolForBlockingRequests
I am asking myself the question: "When should you use scala.concurrent.blocking?"
Well, since that is mostly useful for Future and Future should never be used for serious business logic then never.
Now, "jokes" aside, when using Futures then you should always use blocking when wrapping blocking operations, AND receive a custom ExecutionContext; instead of hardcoding the global one. Note, this should always be the case, even for non-blocking operations, but IME most folks using Future don't do this... but that is another discussion.
Then, callers of those blocking operations may decide if they will use their compute EC or a blocking one.
When the docs mention long-lasting they don't mean anything specific, mostly because is too hard to be specific about that; is context / application specific. What you need to understand is that blocking by default (note the actual EC may do whatever they want) will just create a new thread, and if you create a lot of threads and they take too long to be released you will saturate your memory and kill the program with an OOM error.
For those situations, the recommendation is to control the back pressure of your app to avoid creating too many threads. One way to do that is to create a fixed thread pool for the maximum number of blocking operations you will support and just enqueue all other pending tasks; such EC should just ignore blocking calls. You may also just have an unbound number of threads but manage the back pressure manually in other parts of your code; e.g. with an explicit Queue, this was common advice before: https://gist.github.com/djspiewak/46b543800958cf61af6efa8e072bfd5c
However, having blocked threads is always hurtful for the performance of your app, even if the compute EC is not blocked. The latest talks by Daniel explain those in detail: "The case for effect systems" & "Threads at scale".
So the ecosystem is pushing hard the state of the art to avoid that at all costs but is not a simple task. Still, runtimes like the ones provided by cats-effect or ZIO are optimized to handle blocking tasks the best they can as of today, and will probably improve during this and next years.
I am working with Play Framework (Scala) version 2.3. From the docs:
You can’t magically turn synchronous IO into asynchronous by wrapping it in a Future. If you can’t change the application’s architecture to avoid blocking operations, at some point that operation will have to be executed, and that thread is going to block. So in addition to enclosing the operation in a Future, it’s necessary to configure it to run in a separate execution context that has been configured with enough threads to deal with the expected concurrency.
This has me a bit confused on how to tune my webapp. Specifically, since my app has a good amount of blocking calls: a mix of JDBC calls, and calls to 3rd party services using blocking SDKs, what is the strategy for configuring the execution context and determining the number of threads to provide? Do I need a separate execution context? Why can't I simply configure the default pool to have a sufficient amount of threads (and if I do this, why would I still need to wrap the calls in a Future?)?
I know this ultimately will depend on the specifics of my app, but I'm looking for some guidance on the strategy and approach. The play docs preach the use of non-blocking operations everywhere but in reality the typical web-app hitting a sql database has many blocking calls, and I got the impression from reading the docs that this type of app will perform far from optimally with the default configurations.
[...] what is the strategy for configuring the execution context and
determining the number of threads to provide
Well, that's the tricky part which depends on your individual requirements.
First of all, you probably should choose a basic profile from the docs (pure asynchronous, highly synchronous or many specific thread pools)
The second step is to fine-tune your setup by profiling and benchmarking your application
Do I need a separate execution context?
Not necessarily. But it makes sense to use separate execution contexts if you want to trigger all your blocking IO-calls at once and not in a sequential way (so database call B does not have to wait until database call A is finished).
Why can't I simply configure the default pool to have a sufficient
amount of threads (and if I do this, why would I still need to wrap
the calls in a Future?)?
You can, check the docs:
play {
akka {
akka.loggers = ["akka.event.slf4j.Slf4jLogger"]
loglevel = WARNING
actor {
default-dispatcher = {
fork-join-executor {
parallelism-min = 300
parallelism-max = 300
}
}
}
}
}
With this approach, you basically are turning Play into a one-thread-per-request-model. This is not the idea behind Play, but if you're doing a lot of blocking IO calls, it's the simplest approach. In this case, you don't need to wrap your database calls in a Future.
To put it in a nutshell, you basically have three ways to go:
Only use (IO-)technologies whose API calls are non-blocking and asynchronous. This allows you to use a small threadpool / default execution context which suits the nature of Play
Turn Play into a one-thread-per-request Framework by drastically increasing the default execution context. No futures needed, just call your blocking database as always
Create specific execution contexts for your blocking IO-calls and gain fine-grained control of what you are doing
Firstly, before diving in and refactoring your app, you should determine whether this is actually a problem for you. Run some benchmarks (gatling is superb) and do a few profiles with something like JProfiler. If you can live with the current performance then happy days.
The ideal is to use a reactive driver which would return you a future that then gets passed all the way back to your controller. Unfortunately async is still an Open ticket for slick. Interacting with REST APIs can be made reactive using the PlayWS library, but if you have to go via a library that your 3rd party provides then you're stuck.
So, assuming that none of these are feasible and that you do need to improve performance, the question is what benefit would Play's suggestion have? I think what they're getting at here is that it's useful to partition your threads into those that block and those that can make use of asynchronous techniques.
If, for instance, only some proportion of your requests are long and blocking then with a single thread pool you risk all threads being used for the blocking operations. Your controller would then not be able to handle any new requests, irrespective of whether that request needs to call a blocking service. If you can allocate enough threads that this never happens then no problem.
If, on the other hand, you are hitting your limit for threads then by using two pools you can keep your fast, non-blocking requests snappy. You would have one pool servicing requests in your controller and calling into services which return futures. Some of these futures would actually be performing work using a separate pool of threads, but only for the blocking operations. If there is any portion of your app which could be made reactive, then your controller could take advantage of this while isolating the controller from the blocking operations.
I am starting to learn Scala and functional programming. I was reading the book !Programming scala: Tackle Multi-Core Complexity on the Java Virtual Machine". Upon the first chapter I've seen the word Event-Driven concurrency and Actor model. Before I continue reading this book I want to have an idea about Event-Driven concurrency or Actor Model.
What is Event-Driven concurrency, and how is it related to Actor Model?
An Event Driven programming model involves registering code to be run when a given event fires. An example is, instead of calling a method that returns some data from a database:
val user = db.getUser(1)
println(user.name)
You could instead register a callback to be run when the data is ready:
db.getUser(1, u => println(u.name))
In the first example, no concurrency was happening; The current thread would block until db.getUser(1) returned data from the database. In the second example db.getUser would return immediately and carry on executing the next code in the program. In parallel to this, the callback u => println(u.name) will be executed at some point in the future.
Some people prefer the second approach as it doesn't mean memory hungry Threads are needlessly sat around waiting for slow I/O to return.
The Actor Model is an example of how Event-Driven concepts can be used to help the programmer easily write concurrent programs.
From a super high level, Actors are objects that define a series of Event Driven message handlers that get fired when the Actor receives messages. In Akka, each instance of an Actor is single Threaded, however when many of these Actors are put together they create a system with concurrency.
For example, Actor A could send messages to Actor B and C in parallel. Actor B and C could fire messages back to Actor A. Actor A would have message handlers to receive these messages and behave as desired.
To learn more about the Actor model I would recommend reading the Akka documentation. It is really well written: http://doc.akka.io/docs/akka/2.1.4/
There is also lot's of good documentation around the web about Event Driven Concurrency that us much more detailed than what I've written here. http://berb.github.io/diploma-thesis/original/055_events.html
Theon's answer provides a good modern overview. I'd like to add some historical perspective.
Tony Hoare and Robert Milner both developed mathematical algebra for analysing concurrent systems (Communicating Sequential Processes, CSP, and Communicating Concurrent Systems, CCS). Both of these look like heavy mathematics to most of us but the practical application is relatively straightforward. CSP led directly to the Occam programming language amongst others, with Go being the newest example. CCS led to Pi calculus and the mobility of communicating channel ends, a feature that is part of Go and was added to Occam in the last decade or so.
CSP models concurrency purely by considering automomous entities ('processes', v.lightweight things like green threads) interacting simply by event exchange. The medium for passing events is along channels. Processes may have to deal with several inputs or outputs and they do this by selecting the event that is ready first. The events usually carry data from the sender to the receiver.
A principle feature of the CSP model is that a pair of processes engage in communication only when both are ready - in practical terms this leads to what is usually called 'synchronous' communication. However, the actual implementations (Go, Occam, Akka) allow channels to be buffered (the normal state in Akka) so that the lock-step exchange of events is often actually decoupled instead.
So in summary, an event-driven CSP-based system is really a data-flow network of processes connected by channels.
Besides the CSP interpretation of event-driven, there have been others. An important example is the 'event-wheel' approach, once popular for modelling concurrent systems whilst actually having a single processing thread. Such systems handle events by putting them into a processing queue and dealing with them due course, usually via a callback. Java Swing's event processing engine is a good example. There were others, e.g. for time-based simulation engines. One might think of the Javascript / NodeJS model as fitting into this category as well.
So in summary, an event-wheel was a way to express concurrency but without parallelism.
The irony of this is that the two approaches I've described above are both described as event driven but what they mean by event driven is different in each case. In one case, hardware-like entities are wired together; in the other, almost all actions are executed by callbacks. The CSP approach claims to be scalable because it's fully composable; it's naturally adept at parallel execution also. If there are any reasons to favour one over the other, these are probably it.
To understand the answer to this you have to look at event concurrency from the OS layer up. First you start with threads which are the smallest section of code that can be run by the OS and eventually deal with I/O, timing and other kinds of events.
The OS groups threads into a process in which they share the same memory, protection and security permissions. Above that layer you have user programs which typically make I/O requests that are handled by user libraries.
The I/O libraries handle these requests in one of two ways. Unix-like systems use a "reactor" model in which the library registers I/O handlers for all the different types of I/O and events in the system. These handlers are activated when I/O is ready on a specific device. Windows-like systems use an I/O completion model in which I/O requests are made and a callback is triggered when the request is complete.
Both of these models require a significant amount of overhead to manage overall program state if you were to use them directly. However some programming tasks (web apps / services) lend themselves to a seemingly more direct implementation if you use an event model directly, but you still need to manage all of that program state. In order to track program logic across dispatches of several related events you have to manually track state and pass it around to the callbacks. This tracking structure is usually called a state context or baton. As you might imagine passing batons around all over the place to numerous seemingly unrelated handlers makes for some extremely hard to read and spaghetti-like code. It's also a pain to write and debug -- especially when you're trying to handle the synchronization of various concurrent paths of execution. You start getting into Futures and then the code becomes really difficult to read.
One well-known event processing library is call libuv. It's a portable event loop that integrates Unix's reactor model with Windows' completion model into a single model usually called a "proactor". Its the event handler that drives NodeJS.
Which brings us to communicating sequential processes.
https://en.wikipedia.org/wiki/Communicating_sequential_processes
Rather than writing asynchronous I/O dispatch and synchronization code using one or more concurrency models (and their often competing conventions), we flip the problem on its head. We use a "coroutine" which looks like normal sequential code.
A simple example is a coroutine that receives a single byte over an event channel from another coroutine that sends a single byte. This effectively synchronizes I/O producer and consumer because the writer/sender has to wait for a reader/receiver and vice-versa. While either process is waiting they explicitly yield execution to other processes. When a coroutine yields, its scoped program state is saved on a stack frame thus saving you from the confusion of managing multi-layered baton state in an event loop.
Using applications built on these event channels we can construct arbitrary, reusable, concurrent logic and the algorithms no longer look like spaghetti code. In pure CSP systems if you write to a channel and there is no reader, you will be blocked. The channel endpoints are known via handles internally to the program.
Actor systems are different in a couple of ways. First, the endpoints are the actor threads and they are named and known external to the mainline program. The second difference is that sends and receives on these channels are buffered. In other words if you send a message to an actor and there isn't one listening or its busy you aren't blocked until one reads from their input channel. Other differences exist like one actor can publish to two different actors concurrently.
As you might guess Actor systems can easily be built from CSP systems. There are other details like waiting for specific event patterns and selecting from them, but that's the basics.
I hope that clarifies things a bit.
Other constructs can be built from these ideas. Various programming systems (Go, Erlang, etc) include CSP implementations within them. Operating systems like Inferno and Node9 use CSPs and Channels as the basis of their distributed computing model.
Go: https://en.wikipedia.org/wiki/Go_(programming_language)
Erlang: https://en.wikipedia.org/wiki/Erlang_(programming_language)
Inferno: https://en.wikipedia.org/wiki/Inferno_(operating_system)
Node9: https://github.com/jvburnes/node9
How does the actor model (in Akka) work when you need to perform I/O (ie. a database operation)?
It is my understanding that a blocking operation will throw an exception (and essentially ruin all concurrency due to the evented nature of Netty, which Akka uses). Hence I would have to use a Future or something similar - however I don't understand the concurrency model.
Can 1 actor be processing multiple message simultaneously?
If an actor makes a blocking call in a future (ie. future.get()) does that block only the current actor's execution; or will it prevent execution on all actors until the blocking call has completed?
If it blocks all execution, how does using a future assist concurrency (ie. wouldn't invoking blocking calls in a future still amount to creating an actor and executing the blocking call)?
What is the best way to deal with a multi-staged process (ie. read from the database; call a blocking webservice; read from the database; write to the database) where each step is dependent on the last?
The basic context is this:
I'm using a Websocket server which will maintain thousands of sessions.
Each session has some state (ie. authentication details, etc);
The Javascript client will send a JSON-RPC message to the server, which will pass it to the appropriate session actor, which will execute it and return a result.
Execution of the RPC call will involve some I/O and blocking calls.
There will be a large number of concurrent requests (each user will be making a significant amount of requests over the WebSocket connection and there will be a lot of users).
Is there a better way to achieve this?
Blocking operations do not throw exceptions in Akka. You can do blocking calls from an Actor (which you probably want to minimize, but thats another story).
no, 1 actor instance cannot.
It will not block any other actors. You can influence this by using a specific Dispatcher. Futures use the default dispatcher (the global event driven one normally) so it runs on a thread in a pool. You can choose which dispatcher you want to use for your actors (per actor, or for all). I guess if you really wanted to create a problem you might be able to pass exactly the same (thread based) dispatcher to futures and actors, but that would take some intent from your part. I guess if you have a huge number of futures blocking indefinitely and the executorservice has been configured to a fixed amount of threads, you could blow up the executorservice. So a lot of 'ifs'. a f.get blocks only if the Future has not completed yet. It will block the 'current thread' of the Actor from which you call it (if you call it from an Actor, which is not necessary by the way)
you do not necessarily have to block. you can use a callback instead of f.get. You can even compose Futures without blocking. check out talk by Viktor on 'the promising future of akka' for more details: http://skillsmatter.com/podcast/scala/talk-by-viktor-klang
I would use async communication between the steps (if the steps are meaningful processes on their own), so use an actor for every step, where every actor sends a oneway message to the next, possibly also oneway messages to some other actor that will not block which can supervise the process. This way you could create chains of actors, of which you could make many, in front of it you could put a load balancing actor, so that if one actor blocks in one chain another of the same type might not in the other chain. That would also work for your 'context' question, pass of workload to local actors, chain them up behind a load balancing actor.
As for netty (and I assume you mean Remote Actors, because this is the only thing that netty is used for in Akka), pass of your work as soon as possible to a local actor or a future (with callback) if you are worried about timing or preventing netty to do it's job in some way.
Blocking operations will generally not throw exceptions, but waiting on a future (for example by using !! or !!! send methods) can throw a time out exception. That's why you should stick with fire-and-forget as much as possible, use a meaningful time-out value and prefer callbacks when possible.
An akka actor cannot explicitly process several messages in a row, but you can play with the throughput value via the config file. The actor will then process several message (i.e. its receive method will be called several times sequentially) if its message queue it's not empty: http://akka.io/docs/akka/1.1.3/scala/dispatchers.html#id5
Blocking operations inside an actor will not "block" all actors, but if you share threads among actors (recommended usage), one of the threads of the dispatcher will be blocked until operations resume. So try composing futures as much as possible and beware of the time-out value).
3 and 4. I agree with Raymond answers.
What Raymond and paradigmatic said, but also, if you want to avoid starving the thread pool, you should wrap any blocking operations in scala.concurrent.blocking.
It's of course best to avoid blocking operations, but sometimes you need to use a library that blocks. If you wrap said code in blocking, it will let the execution context know you may be blocking this thread so it can allocate another one if needed.
The problem is worse than paradigmatic describes since if you have several blocking operations you may end up blocking all threads in the thread pool and have no free threads. You could end up with deadlock if all your threads are blocked on something that won't happen until another actor/future gets scheduled.
Here's an example:
import scala.concurrent.blocking
...
Future {
val image = blocking { load_image_from_potentially_slow_media() }
val enhanced = image.enhance()
blocking {
if (oracle.queryBetter(image, enhanced)) {
write_new_image(enhanced)
}
}
enhanced
}
Documentation is here.
I'm coming from Java, where I'd submit Runnables to an ExecutorService backed by a thread pool. It's very clear in Java how to set limits to the size of the thread pool.
I'm interested in using Scala actors, but I'm unclear on how to limit concurrency.
Let's just say, hypothetically, that I'm creating a web service which accepts "jobs". A job is submitted with POST requests, and I want my service to enqueue the job then immediately return 202 Accepted — i.e. the jobs are handled asynchronously.
If I'm using actors to process the jobs in the queue, how can I limit the number of simultaneous jobs that are processed?
I can think of a few different ways to approach this; I'm wondering if there's a community best practice, or at least, some clearly established approaches that are somewhat standard in the Scala world.
One approach I've thought of is having a single coordinator actor which would manage the job queue and the job-processing actors; I suppose it could use a simple int field to track how many jobs are currently being processed. I'm sure there'd be some gotchyas with that approach, however, such as making sure to track when an error occurs so as to decrement the number. That's why I'm wondering if Scala already provides a simpler or more encapsulated approach to this.
BTW I tried to ask this question a while ago but I asked it badly.
Thanks!
I'd really encourage you to have a look at Akka, an alternative Actor implementation for Scala.
http://www.akkasource.org
Akka already has a JAX-RS[1] integration and you could use that in concert with a LoadBalancer[2] to throttle how many actions can be done in parallell:
[1] http://doc.akkasource.org/rest
[2] http://github.com/jboner/akka/blob/master/akka-patterns/src/main/scala/Patterns.scala
You can override the system properties actors.maxPoolSize and actors.corePoolSize which limit the size of the actor thread pool and then throw as many jobs at the pool as your actors can handle. Why do you think you need to throttle your reactions?
You really have two problems here.
The first is keeping the thread pool used by actors under control. That can be done by setting the system property actors.maxPoolSize.
The second is runaway growth in the number of tasks that have been submitted to the pool. You may or may not be concerned with this one, however it is fully possible to trigger failure conditions such as out of memory errors and in some cases potentially more subtle problems by generating too many tasks too fast.
Each worker thread maintains a dequeue of tasks. The dequeue is implemented as an array that the worker thread will dynamically enlarge up to some maximum size. In 2.7.x the queue can grow itself quite large and I've seen that trigger out of memory errors when combined with lots of concurrent threads. The max dequeue size is smaller 2.8. The dequeue can also fill up.
Addressing this problem requires you control how many tasks you generate, which probably means some sort of coordinator as you've outlined. I've encountered this problem when the actors that initiate a kind of data processing pipeline are much faster than ones later in the pipeline. In order control the process I usually have the actors later in the chain ping back actors earlier in the chain every X messages, and have the ones earlier in the chain stop after X messages and wait for the ping back. You could also do it with a more centralized coordinator.