Develop Reference

C# WebAPI Garbage Collection

I just delivered my first C# WebAPI application to the first customer. Under normal load, performance initially is even better than I expected. Initially.
Everything worked fine until, at some point, memory was up and garbage collection started running riot (as in "It collects objects that are not yet garbage"). At that point, there were multiple W3WP threads with some ten gigs of ram altogether, and single-digit gigs per worker. After a restart of the IIS everything was back to normal, but of course the memory usage is rising again.
Please correct me if I am wrong, but
Shouldn't C# have automatic garbage collection?
Shouldn't it be easy for GC to collect the garbage of a WebAPI application?
And please help me out:
How can I explicitly state what GC should collect, thus preventing memory leaks? Is someBigList = null; the way to go?
How can I detect where the memory leaks are?
EDIT: Let me clarify some things.
My .NET WebAPI application is mostly a bunch of
public class MyApiController:ApiController
{
[HttpGet]
public MyObjectClass[] MyApi(string someParam) {
List<MyObjectClass> list = new List<MyObjectClass>();
...
for/while/foreach {
MyObjectClass obj = new MyObjectClass();
obj.firstStringAttribute = xyz;
...
list.Add(obj);
}
return list.ToArray();
}
}
Under such conditions, GC should be easy: after "return", all local variables should be garbage. Yet with every single call the used memory increases.
I initially thought that C# WebAPI programs behave similar to (pre-compiled) PHP: IIS calls the program, it is executed, returns the value and is then completely disposed off.
But this is not the case. For instance, I found static variables to keep their data between runs, and now I disposed of all static variables.
Because I found static variables to be a problem for GC:
internal class Helper
{
private static List<string> someVar = new List<string>();
internal Helper() {
someVar=new List<string>();
}
internal void someFunc(string str) {
someVar.Add(str);
}
internal string[] someOtherFunc(string str) {
string[] s = someVar.ToArray();
someVar=new List<string>();
return s;
}
}
Here, under low-memory conditions, someVar threw a null pointer error, which in my opinion can only be caused by GC, since I did not find any code where someVar is actively nullified by me.
I think the memory increase slowed down since I actively set the biggest array variables in the most often used Controllers to null, but this is only a gut feeling and not even nearly a complete solution.
I will now do some profiling using the link you provided, and get back with some results.

Shouldn't C# have automatic garbage collection?
C# is a programming language for the .NET runtime, and .NET brings the automatic garbage collection to the table. So, yes, although technically C# isn't the piece that brings it.
Shouldn't it be easy for GC to collect the garbage of a WebAPI application?
Sure, it should be just as easy as for any other type of .NET application.
The common theme here is garbage. How does .NET determine that something is garbage? By verifying that there are no more live references to the object. To be honest I think it is far more likely that you have verified one of your assumptions wrongly, compared to there being a serious bug in the garbage collector in such a way that "It collects objects that are not yet garbage".
To find leaks, you need to figure out what objects are currently held in memory, make a determination whether that is correct or not, and if not, figure out what is holding them there. A memory profiler application would help with that, there are numerous available, such as the Red-Gate ANTS Memory Profiler.
For your other questions, how to make something eligible for garbage collection? By turning it into garbage (see definition above). Note that setting a local variable to null may not necessarily help or be needed. Setting a static variable to null, however, might. But the correct way to determine that is to use a profiler.
Here are some shot-in-the-dark type of tips you might look into:
Look at static classes, static fields, and static properties. Are you storing data there that is accumulating?
How about static events? Do you have this? Do you remember to unsubscribe the event when you no longer need it?
And by "static fields, properties, and events", I also mean normal instance fields, properties and events that are held in objects that directly or indirectly are stored in static fields or properties. Basically, anything that will keep the objects in memory.
Are you remembering to Dispose of all your IDisposable objects? If not, then the memory being used could be unmanaged. Typically, however, when the garbage collector collects the managed object, the finalizer of that object should clean up the unmanaged memory as well, however you might allocate memory that the GC algorithm isn't aware of, and thus thinks it isn't a big problem to wait with collection. See the GC.AddMemoryPressure method for more on this.

How does Monitor.Enter work? [duplicate]

I got a question how Monitor.Enter works. I investigated .net framework source code, and it shows this only:
[System.Security.SecurityCritical] // auto-generated
[ResourceExposure(ResourceScope.None)]
[MethodImplAttribute(MethodImplOptions.InternalCall)]
private static extern void ReliableEnter(Object obj, ref bool lockTaken);
I guess Monitor.Enter implementation is platform dependent, so I browsed Mono source code and I gave up :(
Yes, a critical section assigned for each System.Object instance may solve, but, I don't think the actual Monitor.Lock is written like this, because creating a critical section for each System.Object will cost unlimitedly. (Win32 does not allow billions of critical section objects in a process!)
Does anybody know how Monitor.Enter works? Please reply. Thanks in advance.

Every object in .NET has two extra (hidden—you can't see them) overhead members:
A "type object pointer". This is just a reference to the Type instance of the object. In fact, you can "access" this by calling GetType().
A "sync block index". This is a native WORD size integral type which is an index into the CLR's internal array of "Sync Blocks".
This is how it keeps track of which objects are locked.
The Sync Block structure contains a field that can be marked for locking. Basically, when you lock an object, this field is switched on. When the lock is released, it is switched off (basically - I haven't looked at the SSCLI for long enough to delve deeper into how that sort of operation works - I believe it is based on EnterCriticalSection though..).
The MethodImplOptions.InternalCall arguments you've passed to the attribute above means that the actual implementation of that method resides in the CLR.. which is why you can't browse any further through the code.

Looking at the Mono source code, it seems that they create a Semaphore (using CreateSemaphore or a similar platform-specific function) when the object is first locked, and store it in the object. There also appears to be some object pooling going on with the semaphores and their associated MonoThreadsSync structures.
The relevant function is static inline gint32 mono_monitor_try_enter_internal (MonoObject *obj, guint32 ms, gboolean allow_interruption) in the file mono/metadata/monitor.c, in case you're interested.
I expect that Microsoft .Net does something similar.

Microsoft .NET - whenever possible - tries a spinlock on the thin lock structure in the object header. (Notice how one can "lock" on any object.)
Only if there is a need for it will an event handle be used from the pool or a new one allocated.

How to make the JIT extend stack variables to the end of scope (GC is too quick)

We're dealing with the GC being too quick in a .Net program.
Because we use a class with native resources and we do not call GC.KeepAlive(), the GC collects the object before the Native access ends. As a result the program crashes.
We have exactly the problem as described here:
Does the .NET garbage collector perform predictive analysis of code?
Like so:
{ var img = new ImageWithNativePtr();
IntPtr p = img.GetData();
// DANGER!
ProcessData(p);
}
The point is: The JIT generates information that shows the GC that img is not used at the point when GetData() runs. If a GC-Thread comes at the right time, it collects img and the program crashes. One can solve this by appending GC.KeepAlive(img);
Unfortunately there is already too much code written (at too many places) to rectify the issue easily.
Therefore: Is there for example an Attribute (i.e. for ImageWithNativePtr) to make the JIT behave like in a Debug build? In a Debug build, the variable img will remain valid until the end of the scope ( } ), while in Release it looses validity at the comment DANGER.

To the best of my knowledge there is no way to control jitter's behavior based on what types a method references. You might be able to attribute the method itself, but that won't fill your order. This being so, you should bite the bullet and rewrite the code. GC.KeepAlive is one option. Another is to make GetData return a safe handle which will contain a reference to the object, and have ProcessData accept the handle instead of IntPtr — that's good practice anyway. GC will then keep the safe handle around until the method returns. If most of your code has var instead of IntPtr as in your code fragment, you might even get away without modifying each method.

You have a few options.
(Requires work, more correct) - Implement IDisposable on your ImageWithNativePtr class as it compiles down to try { ... } finally { object.Dispose() }, which will keep the object alive provided you update your code with usings. You can ease the pain of doing this by installing something like CodeRush (even the free Xpress supports this) - which supports creating using blocks.
(Easier, not correct, more complicated build) - Use Post Sharp or Mono.Cecil and create your own attribute. Typically this attribute would cause GC.KeepAlive() to be inserted into these methods.
The CLR has nothing built in for this functionality.

I believe you can emulate what you want with a container that implements IDispose, and a using statement. The using statement allows for defining the scope and you can place anything you want in it that needs to be alive over that scope. This can be a helpful mechanism if you have no control over the implementation of ImageWithNativePtr.
The container of things to dispose is a useful idiom. Particularly when you really should be disposing of something ... which is probably the case with an image.
using(var keepAliveContainer = new KeepAliveContainer())
{
var img = new ImageWithNativePtr();
keepAliveContainer.Add(img);
IntPtr p = img.GetData();
ProcessData(p);
// anything added to the container is still referenced here.
}

Where did variable = null as "object destroying" come from?

Working on a number of legacy systems written in various versions of .NET, across many different companies, I keep finding examples of the following pattern:
public void FooBar()
{
object foo = null;
object bar = null;
try
{
foo = new object();
bar = new object();
// Code which throws exception.
}
finally
{
// Destroying objects
foo = null;
bar = null;
}
}
To anybody that knows how memory management works in .NET, this kind of code is painfully unnecessary; the garbage collector does not need you to manually assign null to tell that the old object can be collected, nor does assigning null instructs the GC to immediately collected the object.
This pattern is just noise, making it harder to understand what the code is trying to achieve.
Why, then, do I keep finding this pattern? Is there a school that teaches this practice? Is there a language in which assigning null values to locally scoped variables is required to correctly manage memory? Is there some additional value in explicitly assigning null that I haven't percieved?

It's FUDcargo cult programming (thanks to Daniel Earwicker) by developers who are used to "free" resources, bad GC implementations and bad API.
Some GCs didn't cope well with circular references. To get rid of them, you had to break the cycle "somewhere". Where? Well, if in doubt, then everywhere. Do that for a year and it's moved into your fingertips.
Also setting the field to null gives you the idea of "doing something" because as developers, we always fear "to forget something".
Lastly, we have APIs which must be closed explicitly because there is no real language support to say "close this when I'm done with it" and let the computer figure it out just like with GC. So you have an API where you have to call cleanup code and API where you don't. This sucks and encourages patterns like the above.

It is possible that it came from VB which used a reference counting strategy for memory management and object lifetime. Setting a reference to Nothing (equivalent to null) would decrement the reference count. Once that count became zero then the object was destroyed synchronously. The count would be decremented automatically upon leaving the scope of a method so even in VB this technique was mostly useless, however there were special situations where you would want to greedily destroy an object as illustrated by the following code.
Public Sub Main()
Dim big As Variant
Set big = GetReallyBigObject()
Call big.DoSomething
Set big = Nothing
Call TimeConsumingOperation
Call ConsumeMoreMemory
End Sub
In the above code the object referenced by big would have lingered until the end without the call to Set big = Nothing. That may be undesirable if the other stuff in the method was a time consuming operation or generated more memory pressure.

It comes from C/C++ where explicitly made setting your pointers to null was the norm (to eliminate dangling pointers)
After calling free():
#include <stdlib.h>
{
char *dp = malloc ( A_CONST );
// Now that we're freeing dp, it is a dangling pointer because it's pointing
// to freed memory
free ( dp );
// Set dp to NULL so it is no longer dangling
dp = NULL;
}
Classic VB developers also did the same thing when writing their COM components to prevent memory leaks.

It is more common in languages with deterministic garbage collection and without RAII, such as the old Visual Basic, but even there it's unnecessary and there it was often necessary to break cyclic references. So possibly it really stems from bad C++ programmers who use dumb pointers all over the place. In C++, it makes sense to set dumb pointers to 0 after deleting them to prevent double deletion.

I've seen this a lot in VBScript code (classic ASP) and I think it comes from there.

I think it used to be a common misunderstanding among former C/C++ developers. They knew that the GC will free their memory, but they didn't really understand when and how. Just clean it and carry on :)

I suspect that this pattern comes from translating C++ code to C# without pausing to understand the differences between C# finalization and C++ finalization. In C++ I often null things out in the destructor, either for debugging purposes (so that you can see in the debugger that the reference is no longer valid) or, rarely, because I want a smart object to be released. (If that's the meaning I'd rather call Release on it and make the meaning of the code crystal-clear to the maintainers.) As you note, this is pretty much senseless in C#.
You see this pattern in VB/VBScript all the time too, for different reasons. I mused a bit about what might cause that here:
Link

May be the convention of assigning null originated from the fact that had foo been an instance variable instead of a local variable, you should remove the reference before GC can collect it. Someone slept during the first sentence and started nullifying all their variables; the crowd followed.

It comes from C/C++ where doing a free()/delete on an already released pointer could result in a crash while releasing a NULL-pointer simply did nothing.
This means that this construct (C++) will cause problems
void foo()
{
myclass *mc = new myclass(); // lets assume you really need new here
if (foo == bar)
{
delete mc;
}
delete mc;
}
while this will work
void foo()
{
myclass *mc = new myclass(); // lets assume you really need new here
if (foo == bar)
{
delete mc;
mc = NULL;
}
delete mc;
}
Conclusion: IT's totally unneccessary in C#, Java and just about any other garbage-collecting language.

Consider a slight modification:
public void FooBar()
{
object foo = null;
object bar = null;
try
{
foo = new object();
bar = new object();
// Code which throws exception.
}
finally
{
// Destroying objects
foo = null;
bar = null;
}
vavoom(foo,bar);
}
The author(s) may have wanted to ensure that the great Vavoom (*) did not get pointers to malformed objects if an exception was previously thrown and caught. Paranoia, resulting in defensive coding, is not necessarily a bad thing in this business.
(*) If you know who he is, you know.

VB developers had to dispose all of their objects, to try and mitigate the chance of a Memory leak. I can imagine this is where it has come from as VB devs migrated over to .NEt / c#

I can see it coming from either a misunderstanding of how the garbage collection works, or an effort to force the GC to kick in immediately - perhaps because the objects foo and bar are quite large.

I've seen this in some Java code before. It was used on a static variable to signal that the object should be destroyed.
It probably didn't originate from Java though, as using it for anything other than a static variable would also not make sense in Java.

It comes from C++ code especially smart pointers. In that case it's rougly equivalent to a .Dispose() in C#.
It's not a good practice, at most a developer's instinct. There is no real value by assigning null in C#, except may be helping the GC to break a circular reference.

Setting Objects to Null/Nothing after use in .NET

Should you set all the objects to null (Nothing in VB.NET) once you have finished with them?
I understand that in .NET it is essential to dispose of any instances of objects that implement the IDisposable interface to release some resources although the object can still be something after it is disposed (hence the isDisposed property in forms), so I assume it can still reside in memory or at least in part?
I also know that when an object goes out of scope it is then marked for collection ready for the next pass of the garbage collector (although this may take time).
So with this in mind will setting it to null speed up the system releasing the memory as it does not have to work out that it is no longer in scope and are they any bad side effects?
MSDN articles never do this in examples and currently I do this as I cannot
see the harm. However I have come across a mixture of opinions so any comments are useful.

Karl is absolutely correct, there is no need to set objects to null after use. If an object implements IDisposable, just make sure you call IDisposable.Dispose() when you're done with that object (wrapped in a try..finally, or, a using() block). But even if you don't remember to call Dispose(), the finaliser method on the object should be calling Dispose() for you.
I thought this was a good treatment:
Digging into IDisposable
and this
Understanding IDisposable
There isn't any point in trying to second guess the GC and its management strategies because it's self tuning and opaque. There was a good discussion about the inner workings with Jeffrey Richter on Dot Net Rocks here: Jeffrey Richter on the Windows Memory Model and
Richters book CLR via C# chapter 20 has a great treatment:

Another reason to avoid setting objects to null when you are done with them is that it can actually keep them alive for longer.
e.g.
void foo()
{
var someType = new SomeType();
someType.DoSomething();
// someType is now eligible for garbage collection
// ... rest of method not using 'someType' ...
}
will allow the object referred by someType to be GC'd after the call to "DoSomething" but
void foo()
{
var someType = new SomeType();
someType.DoSomething();
// someType is NOT eligible for garbage collection yet
// because that variable is used at the end of the method
// ... rest of method not using 'someType' ...
someType = null;
}
may sometimes keep the object alive until the end of the method. The JIT will usually optimized away the assignment to null, so both bits of code end up being the same.

No don't null objects. You can check out https://web.archive.org/web/20160325050833/http://codebetter.com/karlseguin/2008/04/28/foundations-of-programming-pt-7-back-to-basics-memory/ for more information, but setting things to null won't do anything, except dirty your code.

Also:
using(SomeObject object = new SomeObject())
{
// do stuff with the object
}
// the object will be disposed of

In general, there's no need to null objects after use, but in some cases I find it's a good practice.
If an object implements IDisposable and is stored in a field, I think it's good to null it, just to avoid using the disposed object. The bugs of the following sort can be painful:
this.myField.Dispose();
// ... at some later time
this.myField.DoSomething();
It's good to null the field after disposing it, and get a NullPtrEx right at the line where the field is used again. Otherwise, you might run into some cryptic bug down the line (depending on exactly what DoSomething does).

Chances are that your code is not structured tightly enough if you feel the need to null variables.
There are a number of ways to limit the scope of a variable:
As mentioned by Steve Tranby
using(SomeObject object = new SomeObject())
{
// do stuff with the object
}
// the object will be disposed of
Similarly, you can simply use curly brackets:
{
// Declare the variable and use it
SomeObject object = new SomeObject()
}
// The variable is no longer available
I find that using curly brackets without any "heading" to really clean out the code and help make it more understandable.

In general no need to set to null. But suppose you have a Reset functionality in your class.
Then you might do, because you do not want to call dispose twice, since some of the Dispose may not be implemented correctly and throw System.ObjectDisposed exception.
private void Reset()
{
if(_dataset != null)
{
_dataset.Dispose();
_dataset = null;
}
//..More such member variables like oracle connection etc. _oraConnection
}

The only time you should set a variable to null is when the variable does not go out of scope and you no longer need the data associated with it. Otherwise there is no need.

this kind of "there is no need to set objects to null after use" is not entirely accurate. There are times you need to NULL the variable after disposing it.
Yes, you should ALWAYS call .Dispose() or .Close() on anything that has it when you are done. Be it file handles, database connections or disposable objects.
Separate from that is the very practical pattern of LazyLoad.
Say I have and instantiated ObjA of class A. Class A has a public property called PropB of class B.
Internally, PropB uses the private variable of _B and defaults to null. When PropB.Get() is used, it checks to see if _PropB is null and if it is, opens the resources needed to instantiate a B into _PropB. It then returns _PropB.
To my experience, this is a really useful trick.
Where the need to null comes in is if you reset or change A in some way that the contents of _PropB were the child of the previous values of A, you will need to Dispose AND null out _PropB so LazyLoad can reset to fetch the right value IF the code requires it.
If you only do _PropB.Dispose() and shortly after expect the null check for LazyLoad to succeed, it won't be null, and you'll be looking at stale data. In effect, you must null it after Dispose() just to be sure.
I sure wish it were otherwise, but I've got code right now exhibiting this behavior after a Dispose() on a _PropB and outside of the calling function that did the Dispose (and thus almost out of scope), the private prop still isn't null, and the stale data is still there.
Eventually, the disposed property will null out, but that's been non-deterministic from my perspective.
The core reason, as dbkk alludes is that the parent container (ObjA with PropB) is keeping the instance of _PropB in scope, despite the Dispose().

Stephen Cleary explains very well in this post: Should I Set Variables to Null to Assist Garbage Collection?
Says:
The Short Answer, for the Impatient
Yes, if the variable is a static field, or if you are writing an enumerable method (using yield return) or an asynchronous method (using async and await). Otherwise, no.
This means that in regular methods (non-enumerable and non-asynchronous), you do not set local variables, method parameters, or instance fields to null.
(Even if you’re implementing IDisposable.Dispose, you still should not set variables to null).
The important thing that we should consider is Static Fields.
Static fields are always root objects, so they are always considered “alive” by the garbage collector. If a static field references an object that is no longer needed, it should be set to null so that the garbage collector will treat it as eligible for collection.
Setting static fields to null is meaningless if the entire process is shutting down. The entire heap is about to be garbage collected at that point, including all the root objects.
Conclusion:
Static fields; that’s about it. Anything else is a waste of time.

There are some cases where it makes sense to null references. For instance, when you're writing a collection--like a priority queue--and by your contract, you shouldn't be keeping those objects alive for the client after the client has removed them from the queue.
But this sort of thing only matters in long lived collections. If the queue's not going to survive the end of the function it was created in, then it matters a whole lot less.
On a whole, you really shouldn't bother. Let the compiler and GC do their jobs so you can do yours.

Take a look at this article as well: http://www.codeproject.com/KB/cs/idisposable.aspx
For the most part, setting an object to null has no effect. The only time you should be sure to do so is if you are working with a "large object", which is one larger than 84K in size (such as bitmaps).

I believe by design of the GC implementors, you can't speed up GC with nullification. I'm sure they'd prefer you not worry yourself with how/when GC runs -- treat it like this ubiquitous Being protecting and watching over and out for you...(bows head down, raises fist to the sky)...
Personally, I often explicitly set variables to null when I'm done with them as a form of self documentation. I don't declare, use, then set to null later -- I null immediately after they're no longer needed. I'm saying, explicitly, "I'm officially done with you...be gone..."
Is nullifying necessary in a GC'd language? No. Is it helpful for the GC? Maybe yes, maybe no, don't know for certain, by design I really can't control it, and regardless of today's answer with this version or that, future GC implementations could change the answer beyond my control. Plus if/when nulling is optimized out it's little more than a fancy comment if you will.
I figure if it makes my intent clearer to the next poor fool who follows in my footsteps, and if it "might" potentially help GC sometimes, then it's worth it to me. Mostly it makes me feel tidy and clear, and Mongo likes to feel tidy and clear. :)
I look at it like this: Programming languages exist to let people give other people an idea of intent and a compiler a job request of what to do -- the compiler converts that request into a different language (sometimes several) for a CPU -- the CPU(s) could give a hoot what language you used, your tab settings, comments, stylistic emphases, variable names, etc. -- a CPU's all about the bit stream that tells it what registers and opcodes and memory locations to twiddle. Many things written in code don't convert into what's consumed by the CPU in the sequence we specified. Our C, C++, C#, Lisp, Babel, assembler or whatever is theory rather than reality, written as a statement of work. What you see is not what you get, yes, even in assembler language.
I do understand the mindset of "unnecessary things" (like blank lines) "are nothing but noise and clutter up code." That was me earlier in my career; I totally get that. At this juncture I lean toward that which makes code clearer. It's not like I'm adding even 50 lines of "noise" to my programs -- it's a few lines here or there.
There are exceptions to any rule. In scenarios with volatile memory, static memory, race conditions, singletons, usage of "stale" data and all that kind of rot, that's different: you NEED to manage your own memory, locking and nullifying as apropos because the memory is not part of the GC'd Universe -- hopefully everyone understands that. The rest of the time with GC'd languages it's a matter of style rather than necessity or a guaranteed performance boost.
At the end of the day make sure you understand what is eligible for GC and what's not; lock, dispose, and nullify appropriately; wax on, wax off; breathe in, breathe out; and for everything else I say: If it feels good, do it. Your mileage may vary...as it should...

I think setting something back to null is messy. Imagine a scenario where the item being set to now is exposed say via property. Now is somehow some piece of code accidentally uses this property after the item is disposed you will get a null reference exception which requires some investigation to figure out exactly what is going on.
I believe framework disposables will allows throw ObjectDisposedException which is more meaningful. Not setting these back to null would be better then for that reason.

Some object suppose the .dispose() method which forces the resource to be removed from memory.