The specification of ReadableByteChannel.read() shows -1 as result value for end-of-stream. Moreover it specifies ClosedByInterruptExceptionas possible result if the thread is interrupted.
Now I thought that would be all - and it is most of the time. However, now and then I get the following:
java.io.IOException: Eine vorhandene Verbindung wurde vom Remotehost geschlossen
at sun.nio.ch.SocketDispatcher.read0(Native Method)
at sun.nio.ch.SocketDispatcher.read(SocketDispatcher.java:25)
at sun.nio.ch.IOUtil.readIntoNativeBuffer(IOUtil.java:233)
at sun.nio.ch.IOUtil.read(IOUtil.java:206)
at sun.nio.ch.SocketChannelImpl.read(SocketChannelImpl.java:236)
at ...
I do not unterstand why I don't get -1 in this case. Also this is not a clean exception, as I cannot catch it without catching any possible IOException.
So here are my questions:
Why is this exception thrown in the first place?
Is it safe to assume that ANY exception thrown by read are about the socket being closed?
Is all this the same for write()?
And by the way: If I call SocketChannel.close() do I have to call SocketChannel.socket().close() as well or is this implied by the earlier?
Thanks, Steffen
Funny, but somebody today already posted a link to Fallacies of Distributed Computing.
In your case, as a fine German to English translator tells me An existing connection was forcibly closed by remote host.
When you deal with I/O, and specifically Socket I/O, you have to be prepared that any IOException would be thrown at you.
Some of them, like ClosedByInterruptException you may handle intelligently. Others, you may
Add throws declarations to you method and let callers deal with IOExceptions
Wrap IOExceptions into checked exception specific to your subsystem
Wrap IOExceptions into RuntimeException specific or not to your subsystem
Log IOException and continue.
In any case, once you get IOException, you probably cannot do much to communicate with that channel. So you may just close your connection and retry later.
BTW, read will only return -1 to you if you SUCCESSFULLY reached end-of-stream. In your case, I am sure, the connection is closed midstream.
EDIT in reply to OP comments
Can I prevent a channel to be
interruptible?
No, this is a property of specific channel that you are using. If it implements InterruptibleChannel, then it IS interruptible, and will be closed upon receipt of thread interrupt. java.nio.channels.SocketChannel is interruptible
The docs for
ClosedByInterruptException state:
"Checked exception received by a
thread when another thread interrupts
it while it is blocked in an I/O
operation upon a channel." Still how
can it be blocking, if it is
NON-blocking IO?
If you look at specification of ReadableByteChannel.read you will see that the method is declared to only throw IOException. Then in JavaDoc it hints what kind of specific IOExceptions and undre what conditions may be thrown by a standard java.nio implementation of this interface. This is one of the example of cooperative programming. The declarer of the interface tells the implementer how they should implement the method and what exceptions should they throw under what conditions.
For example, if I implement the read method in my exotic channel I may choose to totally ignore the specification and not throw any of exception in the declaration. The user of my class though will probably pay a heavy price when encountering unexpected behavior and will probably dump my implementation in favor of more robust one.
In any case, I think you are confused with how you should react to different exceptions declared in read method.
First of all, not every exception in read method specification may be thrown. Case in point ClodesByInterruptException most likely will NOT get thrown in case you are using a NON-blocking I/O. On the other hand some implementers may choose to close the channel upon receipt of the interrupt and throw this exception when you attempt to read even if you are in NON-blocking I/O. But you really should not get concerned with that decision, because:
You probably control whether current
thread is interrupted or not
This is just the other kind of
IOException, and by default should
be treated as fatal
And yes, the example IOEception is
fatal, but my question is: Are they
ALL?
Remember, that your framework should function normally, or shutdown gracefully when any kind of exception is thrown. For example your code, or library code at any point can throw unchecked ( Runtime ) exception. Only testing and more intimate knowledge of your environment can tell you which exceptions are truly fatal and which can be handled safely.
Because frameworks are written by people they also have bugs, deficiencies, etc. So there may be a case when IOException is thrown, but the underlying channel is still OK. Unfortunately these conditions can only be found with blood and guts in real production environment. Here is one of the example where IOException thrown by a socket can be totally ignored: http://bugs.sun.com/view_bug.do?bug_id=4516760.
So, start by writing generic handler for all IOExceptions and treat them all as fatal, relax it for specific conditions that you find out in production.
Per the documentation here:
Checked exception received by a thread
when another thread interrupts it
while it is blocked in an I/O
operation upon a channel. Before this
exception is thrown the channel will
have been closed and the interrupt
status of the previously-blocked
thread will have been set.
This suggests that the thread that owns the ReadableByteChannel is being interrupted by another thread in your application... is this the case?
If another thread was closing the channel, I would expect you would see a AsynchronousCloseException.
I don't see a reason why you couldn't catch the ClosedByInterruptException explicitly and allow any other exceptions to be handled higher up the stack:
try
{
rbc.read(dst);
}
catch ( ClosedByInterruptException cbie)
{
/* handler */
}
As for write(), there is no write() method for ReadableByteChannel, however, the WritableByteChannel, does have a write() method and it can throw the same exceptions.
Related
This question already has answers here:
Java: Global Exception Handler
(6 answers)
Closed 8 years ago.
I know I can set my own UncaughtExceptionHandler to execute code on uncaught exceptions, but is there a way to have some code execute on all caught exception without having to call that method in every single try-catch block?
My custom code sends me an email with the stack trace on uncaught exceptions. I also want that email for caught exception but I don't want to track down every single try-catch statement and add it.
I'm not looking for a global exception handler (I mentioned UncaughtExceptionHandler), I'm looking for a global CAUGHT exception handler.
You can do it by altering the Exception class to record every Exception. This can be done by changing the boot class path, runtime code injection or using the JVMTI.
I suspect it's a bad idea as the system will catch more exception than you might expect. You could get thousands of emails a second if something goes wrong.
I suggest you revisit your exceptions handling policy and strategy. When an exception happens it generally means something went wrong, i.e., something exceptional (as in unexpected) happened in the code (database connection lost, file not found, network unreachable, etc.) or a bug in the code causes an unexpected problem (NPE comes to mind...). In these scenarios the normal program flow can't continue.
It's usually considered best practice to fail fast in such cases, unless you are expecting some exception to be thrown and know how to recover, i.e., the exceptional condition wasn't that exceptional after all.
If you need to fail fast, you don't need to explicitly handle the exceptions as your thread or program will just die and you can just use the UncaughtExceptionHandler to log the stack trace, send emails or notifications etc.
Situation in which you can successfully recover and resume the normal program flow are far less common in my opinion and usually do not require developers/ops to be notified as it is part of the normal functioning of the system (otherwise you would not be able to handle the exception). So simply logging the specific message and useful data associated to this scenario is usually enough, e.g., "WARNING: Transient error x was recovered. State was y and z.".
The java.io.InputStream.close() method is declared to throw an IOException. Under what circumstances would such an exception actually be thrown?
Edit: Yes I have read the javadoc. Can anybody be more specific than "when an I/O error occurs"? What I/O error can occur while closing an InputStream?
In the case of input stream reading from a file system, an error can be raised while the file system itself updates the last access time metadata, or some other metadata, on a file during closure. Anyway, this happens almost never in practice.
In the case of an input stream reading from a network connection, an error on closure is a bit more conceivable. A normal closure of a network socket actually involves a closure request (TCP/IP FIN packet) being sent over the connection and waiting for the other end to acknowledge this closure request. (In fact, the other end of the connection then in turn sends a closure request, which the closing end acknowledges.) So in the case of a socket input stream, a closure operation actually involves sending traffic over the connection and the closure can thus fail with an error.
Note that in many implementations, close() generally doesn't throw an IOException if the stream is already closed; it simply fails silently to close the stream again.
I'm looking through the Java source code, and have found something interesting which is indicative of the IOException reason. InputStream is an abstract class. We therefore can't predict the kind of input which will be closed, so it's good to keep information flowing.
Whatever code uses that input stream needs to be able to throw an IOException, because there is a chance that closing the input stream may fail. If it fails, whatever's using the implementation needs to know about it, because there's a good chance it needs to be handled.
It's important to understand the layout of the Exception structure in Java. Every exception, of course, extends Exception. However, there are also broader categories of exceptions: java.lang.IOException is one of these, and covers all possible input/output exceptions. When we say there has been an I/O error, we're referencing anything which falls under IOException. As a result, many exceptions extends this one, e.g. FileNotFoundException, EOFException, etc. as it's important to have a broad, overarching exception to manage these.
As a result, any IO class will need to be able to throw any of the various IOExceptions on close. close() therefore must throw IOException - this gives its implementation the ability to throw any of the extensions of IOException as well. This is why close() throws IOException - exceptions are inherited, and you need to be able to any of the IOExceptions when closing a stream.
Here are a couple scenarios worth noting:
You can't close an IOStream twice, though this generally doesn't throw an exception if you do
The content becomes inaccessible (e.g. a disk was unmounted) (The close() is actually critical for the operating system, as it needs to have an indicator of when the file is no longer busy)
The generic source has closed
Any generic failure not covered by all other subclasses of IOException (e.g. FileNotFoundException)
You can check what caused an IOException by running Exception.getMessage().
The underlying close system call will have to be made eventually, e.g. on linux http://linux.die.net/man/2/close. That call is documented to fail with EIO: "An I/O error occurred." So the reason is, that the underlying file system close call can fail.
I have wondered about this myself and have done a little research on this topic few years ago. This is what I know....
If you look at the javadoc in the link you provided, it clearly states that "The close method of InputStream does nothing", so there is no way it would throw an exception, correct? But then looking at all of the subclasses of IOException you will see that there are many situations where subclasses of inputstream may not be able to close the stream. So my best guess is that this exception is designed for subclasses to make use of it.
http://docs.oracle.com/javase/6/docs/api/java/io/IOException.html
In some cases, it is nothing more than a nuisance, and in others it clearly indicates that something went wrong. It all depends on what type of inputstream implementation you are making use of.
I have been reading the JLS and I encountered the section 11.1.3. Asynchronous Exceptions from which I quote:
Most exceptions occur synchronously as a result of an action by the
thread in which they occur, and at a point in the program that is
specified to possibly result in such an exception. An asynchronous
exception is, by contrast, an exception that can potentially occur at
any point in the execution of a program.
And
Asynchronous exceptions occur only as a result of:
[...]
An internal error or resource limitation in the Java virtual machine that prevents it from implementing the semantics of the
Java programming language. In this case, the asynchronous exception
that is thrown is an instance of a subclass of VirtualMachineError.
Is it possible to catch such exceptions for logging purposes or notification (because I believe such thing is unrecoverable)? How can I achieve such thing?
You can catch such exceptions just like any other exception. The only problem is that they may occur at any place in your program, so catching them reliably is hard. You would basically have to wrap the run method of all threads and the main method in a try..catch block, but you can't do that for threads you don't control (like the Swing EDT, or threads for timers etc.).
Also catching any subclass of Error is usually not recommended, because the JVM may be in an unstable state, which might lead to a further failure (for example in the case of OutOfMemoryError, you might not even have enough memory to to the exception handling). However, logging would be a valid reason for catching Errors in my eyes.
My suggested solution would be to use an uncaught exception handler for this by setting it as the default exception handler. In this handler you will get all exceptions and errors, if they are not caught anywhere in the code, and you can try to log them.
There is no point of catching these exceptions (Subclasses of VirtualMachineError) as you have no indecattion in which state the pogram is at the point, the Doc saies about Virtual Machine Errors:
A Java virtual machine implementation throws an object that is an
instance of a subclass of the class VirtualMethodError when an
internal error or resource limitation prevents it from implementing
the semantics described in this chapter. This specification cannot
predict where internal errors or resource limitations may be
encountered and does not mandate precisely when they can be reported.
so assuming you get in an OutOfMemoryError or an UnknownError there isnt much you can do about it, and once your vritualmashine doesnt work properly you cant provide the user anyhelp as your program isnt working properly as well, besides you have no idea at what time, point, and reason it happends since its not a code error that was caused from your program.
Is there a way to get some details regarding exception safety aspects of Java's standard classes? Mainly working with C++ and C#, I'm confused with Java exception specifications, so I need to understand the proper way of working with exceptions.
To be more specific, let's consider ServerSocket. It starts listening for incoming connections as soon as its object is constructed. Then, you should use accept() to accept the connection (if someone tries to connect).
In case you've previously configured your server socket with setSoTimeout(), there's a change that accept() will throw SocketTimeoutException because nobody tried to connect in a specified period of time. That's fine, server socket is still usable, so you just call accept() once again.
But SocketTimeoutException is not the only thing that accept() may throw. What does all the other exceptions mean? If I wrap call to accept() with 2 catch clauses: for SocketTimeoutException and IOException, can I still safely use the related ServerSocket instance after I got into IOException clause?
I'd really appreciate both Java-wide and ServerSocket-specific answers.
It is not safe to reuse the object. For such a question I would always look into a source, that is the reason it is open.
So if you look into that one: http://kickjava.com/src/java/net/ServerSocket.java.htm you notice that in accept() a SocketException (inherits from IOException) is thrown if the socket is closed or not bound anymore. Both states indicate that the ServerSocket is not valid anymore.
So for this class, generally, if you fail with an exception, always try to gracefully close the ServerSocket in a finally block and then recreate it.
Additionally on your Java-wide question scope: Always look into the source and understand what the interface is doing. If it is mission-critical write tests that reproduce the behaviour (should not be easy at all with such a low-level api).
Finally - is Java consistently doing such things that way? No. Some classes are stateless, others are not (like ServerSocket), some are thread-safe, others not. And you need to understand - either from the documentation or (mostly) from the code - what state they build in order to understand what to do when an Exception knocks you off from the main path.
Most people curse those checked Java exceptions, because most of them (as with most of the IOExceptions) are not really recoverable in a meaningful way. Most of the time, they argue, you cannot understand each and every fine corner case. Which is the reason why many complex frameworks may retry twice or thrice if they think in this case they might, but finally throw a RuntimeException to a top framework layer. There they make something useful out of it (a meaningful error providing context) and log all the details they have, which is a huge stack trace most of the time. A great resource of knowledge, feared by many developers.
So what can you do if you could not recover from an untested corner-case problem? Throw up (probably with a some subclass of RuntimeException) the most meaningful stacktrace annotated with all the context you have. Setup monitoring. And if you run into a frequent problem, fix it.
Yes. The object still exists - May be in an error state in which case it will throw other exceptions until that is rectified.
If you read the specification for ServerSocket, you will see that it throws an IOException "if an I/O error occurs when waiting for a connection." Is it safe to accept() on the socket again? Yes. Are you going to get the same Exception thrown again? Likely so.
I have not yet found an easy way to see if an object is still in a usable state. Each object makes its own rules.
In your specific case with ServerSocket I would try one of two different things:
Run netstat or some other utility to see what the OS thinks that the socket is doing. If the OS doesn't think it is listening, then something happened.
or
Write a test program that will throw the exception and see what it does. I do this all the time (especially with proprietary software). It would be harder in the ServerSocket case you picked, since all of the scenarios I can think of (e.g. address in use, insufficient privileges, etc.) would never result in an object being still valid.
But SocketTimeoutException is not the only thing that accept() may throw. What does all the other exceptions mean?
According to the javadoc, the declared exceptions are IOException, SecurityException, SocketTimeoutException and IllegalBlockingModeException. The SecurityException and IllegalBlockingModeException only occur in specific contexts and you should not attempt to catch and handle them. (They are not problems you want to try to recover from!) The IOException case occurs when "some other I/O error" occurs. The javadoc does not specify what those I/O errors might be, but possibilities might include such things as:
the address to which you have bound is no longer valid
a transport protocol error has occurred
some error (resource issue, bug ...) occurred in the OS protocol stack
(The fact that the javadoc doesn't say which IOException subclasses might be thrown is a hint that you shouldn't try to do clever things to try to recover. If you do, your code is likely to be platform dependent.)
If I wrap call to accept() with 2 catch clauses: for SocketTimeoutException and IOException, can I still safely use the related ServerSocket instance after I got into IOException clause?
Yes and no. It is safe in the sense that you won't put your application into a worse state than it is already in. On the other hand, there is no guarantee that the next accept call won't fail with the same problem.
If your application is intended to run as an unattended server, I don't think you have much choice but to log the IOException and try again ... hoping that the problem is transient.
You can find your answer to the javadoc of setsoTimeout, it says :
Enable/disable SO_TIMEOUT with the specified timeout, in milliseconds. With this option set to a non-zero timeout, a call to accept() for this ServerSocket will block for only this amount of time. If the timeout expires, a java.net.SocketTimeoutException is raised, though the ServerSocket is still valid. The option must be enabled prior to entering the blocking operation to have effect. The timeout must be > 0. A timeout of zero is interpreted as an infinite timeout.
A little more context: I have a listener class that listens on network connections. Its listen method should not be called on it twice. What should I throw when it is?
If you want to use a predefined exception, you might throw IllegalStateException, since the object is already in the listening state.
If you do this, though, there should be a way for the client to find out if listen has already been called. E.g. an isListening() method.
You might also create your own exception, possibly extended from IOException, since this is an IO Problem. Presumably, attempting to listen more than once fails since the network resource has already been used on the first listen. (Similar to sockets binding.)
Since it is a logical error that can only occur when the client of your library is using it incorrectly, the exception you throw should be an unchecked exception (e.g. java.lang.RuntimeException or a derivative thereof) as opposed to a checked exception.
java.lang.IllegalStateException is intended for that purpose:
"Signals that a method has been invoked at an illegal or inappropriate time"
But it could also be an exception in your own custom made exception hierarchy.
In this case I would throw a IllegalStateException. It is normally used to indicate that the user tried to make something which is not legal in the current state of the object.