robgruen's WebLog : .NET Remoting

Sticky Sessions, NLB, and .NET Remoting

robgruen — Mon, 25 Jul 2005 14:48:00 GMT

What is Network Load Balancing (NLB) ? http://www.microsoft.com/technet/prodtechnol/windowsserver2003/library/TechRef/6ac4a6ba-1c0c-46be-8c6a-2c2e0e567e98.mspx

What configurations does MS support in NLB at the time of this writing (Short Answer: SingleCall)? http://support.microsoft.com/default.aspx?scid=kb;en-us;830217

A lot of current implementations use Singletons in NLB (this configuration is possible with the framework but is not supported by Product Support Services). What does this mean? Well, you run into this error: System.Net.WebException, "the underlying connection was closed." Typically this occurs because the NLB solution sends a reset on the port of the server. Generally you can fix this by configuring sticky sessions on your NLB hardware/software. If your NLB solution doesn't offer this take your singleton off of NLB or switch to using webservices.

I took some notes on some remoting questions I recently answered...

robgruen — Fri, 08 Jul 2005 00:01:00 GMT

Server activated objects (SAOs both singleton and singlecall) use threadpool magic to execute remoting calls on thread pool threads. See ThreadPool.QueueUserWorkItem in the docs. Also note that you can change the # of thread pool threads…the default is 25 per proc.

It is possible to share data with Singlecall object instances by having static members in the shared class. I would only recommend this for read-only scenarios and use Singletons for any other scenario. Keep in mind that you are responsible for thread safety.

Both the TCP and HTTP channels were designed to assume that there is network connectivity. Using these channels between processes on the same box that’s disconnected from the network won’t work out of the box. To get this to work you have to bind the channel to the loopback address (either 127.0.0.1 or 0.0.0.0). In .NET 2.0 you can simply use the IPC channel to do remoting sans network.

</channel>

</channels>

Help, I've lost my remote object!?

robgruen — Wed, 22 Jun 2005 22:14:00 GMT

So this is the 1st time I've seen this but I guess it makes sense since I only deal with so many things. I was helping someone setup IIS hosting of a remote object but for the life of me I couldn't get it working on this W2k3 box. I kept getting 404 (not found) when attempting to invoke the remote object. We'd switch to XP and everything's kosher...back to W2k3 with the same code and it wouldn't work. I checked the IIS logs and found my entry but still no luck. I even ran filemon to see if the config file for the object was getting touched by IIS/ASP.NET but nada.

IIS Log:

2005-06-22 17:10:43 XX.XX.XX.XX GET /Remoting/RemoteObj.rem - 80 - XX.XX.XX.XX HTTP/1.1 Mozilla/4.0+(compatible;+MSIE+6.0;+Windows+5.1.2600.0;+MS+.NET+Remoting;+MS+.NET+CLR+1.1.4322.2032+) webserver98 404 0 2

Well, after consulting with Adam it turns out this is a "feature" of IIS 6:

“HTTP Error 404” – File or Directory no found” error message when you request dynamic content with IIS 6.0 - http://support.microsoft.com/?id=315122

and for reference purposes:

Web service extension lockdown policy prevents this request.

IIS Status Codes - http://support.microsoft.com/?id=318380

Caspol -s Behavioral changes for Whidbey

robgruen — Mon, 13 Jun 2005 17:30:00 GMT

We generally use "caspol.exe -s off" as a quick test to determine if .NET framework security plays a part in problems with managed components. It used to be that sometimes people would forget to run "caspol.exe -s on" to turn framework security on. Imagine all the nasties that could infect your box then!

Well, in Whidbey things are different. Now when you run caspol to turn off .NET framework security you see this in the console window:

c:\Program Files\Microsoft Visual Studio 8\VC>caspol -s off
Microsoft (R) .NET Framework CasPol 2.0.50603.0
c Microsoft Corporation. All rights reserved.

CAS enforcement is being turned off temporarily. Press <enter> when you want to
restore the setting back on.

So, now there's chance that security will be comprimised by you forgetting to turn security off since it's temporary. You may ask what happens if caspol terminates unexpectedly? Good question, I tested it!

If caspol.exe terminates before you hit <enter> to turn security back on it will revert to the ON state. Here's another side note, running caspol -s on doesn't do anything. You'll see this for output when you try turn security on:

c:\Program Files\Microsoft Visual Studio 8\VC>caspol -s on
Microsoft (R) .NET Framework CasPol 2.0.50603.0
c Microsoft Corporation. All rights reserved.

Because security can no longer be turned off, turning on security does not have
any effect.
Success

The evading solution to large remoting messages

robgruen — Mon, 02 May 2005 21:57:00 GMT

Thanks to rick for the title!

We are seeing more and more people encountering:

Q322975 BUG: Passing large quantities of data in .NET Remoting calls causes an unexpected exception to occur - http://support.microsoft.com/?id=322975.

As you can tell from the title of the KB article, this is indeed a bug. Unfortunately the behavior addressed by this knowledgebase article is more of a “by design” problem than a coding error. Remoting was never designed for large messages and to my knowledge no testing was done on the types of large messages that cause the errors mentioned in the KB article. From what I have seen with remoting a “typical” remoting capsule is in the neighborhood of 16k – 32k (This is a very generous estimate).

There’s a HUGE difference between 32k and > ~5MB. Taking the development stance you can see why this behavior is by design. However, taking the implementer’s stance, we can see that when using datasets and other complex objects real world scenarios can easily exceed 32k. Developers don’t like fixing code that has an “easy” work around, and implementers don’t want to compromise their design or already coded applications. What can be done?

Well, the last time I checked both you and I lived in the real world, so here are some real world things that can be done to mitigate seeing the SocketException from Q322975.

The source of this exception is actually within the networking stack and is courtesy of the implementation of one or another socket library. Since I’m not familiar with that code I’ve had to speculate about a few things and can tell you this: The error is somewhat hardware bound and throwing hardware at it has mildly successful results. For example, I’d see this problem sooner and more often on a PIII 600 Mhz with 256MB Ram and lest often on a P4 3.6 Ghz with 2GB of Ram. I’m under the impression that the buffering of the incoming data is somewhat dependent on the system hardware. That means that this exception can be alleviated to some extent but not eliminated. This is kind of like taking some cold medication; the medicine suppresses the symptoms but you still have the cold…it just may or may not be there when the medicine wears off.

If you’ve already maxed out the hardware and want the least impact to your existing codebase possible, you can create a new remoting channel that implements some level of data compression. Think of this as zipping up all of your remoting messages. Of course this requires the minor change of modifying your remoting configuration. It might cost you time and money depending on your ability/willingness to write your own channel implementation. You can certainly save both time and money by using a 3^rd party channel implementation such as the one provided by GenuineChannels.

A more involved solution includes actually modifying your code to a slightly larger extent. The benefit of this option however is that it’s not as involved as writing your own channel and does not involve the acquisition of a 3^rd party product. Another type of compression that can be implemented is actually shrinking the remoting message by excluding that data which is not needed by the client. Please consider the following KB article: Q829740 Improving DataSet Serialization and Remoting Performance – http://support.microsoft.com/?id=829740. Simply put, you modify the serialization of types you pass via remoting in such a way that you are sending only data that is necessary on the other side of the remoting call.

The last option is indicated in the original KB article: chunk your data into smaller pieces. This work around involves modifying those calls that generate huge messages into several smaller pieces. I recommend that messages probably shouldn’t exceed 2 MB as I’ve never seen this exception on messages smaller than twice that size.

Silly me!

robgruen — Fri, 29 Apr 2005 00:19:00 GMT

I'm working on this managed application that's doing some remoting. The app's got all sorts of heterogenous channels being mixed and matched....at one point I was getting an error with this text: "Tcp channel protocol violation: expecting preamble."

I found some very interesting causes for this exception except of course in my particular scenario. Turns out that I was trying to make a tcp:// request to a http:// server channel. Whoops! Anyway, I figured I'd share since I didn't see any other info out there for this error with this cause.

Read this if you're getting an InvalidCastException when referencing a type created with AppDomain.CreateInstanceFrom

robgruen — Wed, 06 Apr 2005 05:43:00 GMT

I was helping someone out with an issue the other day. Basically we were attempting to create an instance of an object within an AppDomain and then use that object. Unfortunately when casting the object from the object returned from CreateInstance to the desired type we were getting an InvalidCastException. Looking at the debugger showed that both types were identical (base folder, version, key, etc.).

The code looks something like this:

object obj = domain.CreateInstanceAndUnwrap(@"C:\CustomAssembly.dll", "CustomObjects.Class1");

Type t = obj.GetType();

CustomObjects.Class1 obj;

obj = (CustomObjects.Class1)obj;

To make the problem more hard to diagnose if I created a new AppDomain and then created the type in the new AppDomain: what do you know, no problem! After careful inspection I noticed two instances of CustomAssembly.dll loaded at different offsets within the domain. Turns out the reference in my project for CustomAssembly and the path I was passing into CreateInstance pointed to two different physical files. Change the call to CreateInstance so it passes in the path that is referenced in the assembly and you're problem's solved.

obj.Equals(obj) returns FALSE? For remoting objects this is TRUE!!! Confused yet?

robgruen — Mon, 28 Feb 2005 23:12:00 GMT

RemoteObj obj = (RemoteObj)Activator.GetObject(typeof(RemoteObj), "http://localhost:2222/remoteobj");

// Prints "does not work"

Console.WriteLine("Remote: Equals " + (obj.Equals(obj) ? "works" : "does not work"));

There's a very logical explination for this behavior:

When you call Equals() on the __TransparentProxy for your remoting object the call is actually getting forwarded to your remoted object. Inside of your remoted object Equals is called with the parameter that was passed in which in this case is a __TransparentProxy. So, Equals compares your RemoteObject with __TransparentProxy and determines that the two objects are not the same.

You can verify that this is the case simply by overriding Equals in your remote object, put a breakpoint there and see that this = RemotingObject and obj = __TransparentProxy. When comparing these objects they are not the same and Equals returns false.

I need to see more clearly….”Marshal BYref object”s in .NET Remoting?

robgruen — Wed, 26 Jan 2005 21:20:00 GMT

I hope everyone’s a bit familiar with .NET remoting because I think I’ve confused myself. J This post talks about the behavior of ByRef (VB.NET) and ref (C#) keywords for remoting calls.

Let’s say I have a remoting object that exposes the following method and I would like to call it from a remote client.

public void PassDataSet(ref System.Data.DataSet ds)

{

ds.Tables.Add();

}

Does this even work? Why, yes, yes it does! And even better, does the remote client see the changes that were made on the server side…it sure does. Pretty crazy hunh! “What the heck is going on here?” you may ask. How is it possible to pass pointer to an object to a remote server?

Well, this isn’t what’s happening. When you specify ref or ByRef on a remote function the .net framework actually serializes your parameter and adds it to the remoting message. So, in this instance when I call PassDataSet the data set is serialized on the client, added to the remoting message and passed to the server. The server then deseralizes the remoting message, including the dataset. The code on the server then adds a table to the dataset, creates the return remoting message, serializes the updated dataset, and adds it to the return message. Then, on the client side the framework deserializes the message and dataset and then updates your copy of the dataset with the version returned in the remoting message.

See the SOAP message of this call:

POST /RemoteObject.rem HTTP/1.1

User-Agent: Mozilla/4.0+(compatible; MSIE 6.0; Windows 5.1.2600.0; MS .NET Remoting; MS .NET CLR 1.1.4322.2032 )

Content-Type: text/xml; charset="utf-8"

SOAPAction: "http://schemas.microsoft.com/clr/nsassem/RemoteObjectNamespace.RemoteObject/RemoteObject#PassDataSet"

Content-Length: 1516

Expect: 100-continue

Connection: Keep-Alive

Host: localhost:8888

<SOAP-ENV:Envelope xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xsd="http://www.w3.org/2001/XMLSchema" xmlns:SOAP-ENC="http://schemas.xmlsoap.org/soap/encoding/" xmlns:SOAP-ENV="http://schemas.xmlsoap.org/soap/envelope/" xmlns:clr="http://schemas.microsoft.com/soap/encoding/clr/1.0" SOAP-ENV:encodingStyle="http://schemas.xmlsoap.org/soap/encoding/">

<SOAP-ENV:Body>

<i2:PassDataSet id="ref-1" xmlns:i2="http://schemas.microsoft.com/clr/nsassem/RemoteObjectNamespace.RemoteObject/RemoteObject">

</i2:PassDataSet>

<a1:DataSet id="ref-4" xmlns:a1="http://schemas.microsoft.com/clr/nsassem/System.Data/System.Data%2C%20Version%3D1.0.5000.0%2C%20Culture%3Dneutral%2C%20PublicKeyToken%3Db77a5c561934e089">

<XmlSchema id="ref-5"><?xml version="1.0" encoding="utf-16"?>

<xs:schema id="emptydataset" xmlns="" xmlns:xs="http://www.w3.org/2001/XMLSchema" xmlns:msdata="urn:schemas-microsoft-com:xml-msdata">

<xs:element name="emptydataset" msdata:IsDataSet="true">

<xs:complexType>

<xs:choice maxOccurs="unbounded" />

</xs:complexType>

</xs:element>

</xs:schema></XmlSchema>

<XmlDiffGram id="ref-6"><diffgr:diffgram xmlns:msdata="urn:schemas-microsoft-com:xml-msdata" xmlns:diffgr="urn:schemas-microsoft-com:xml-diffgram-v1" /></XmlDiffGram>

</a1:DataSet>

</SOAP-ENV:Body>

</SOAP-ENV:Envelope>

And the return message from the server (note the added table to the dataset):

HTTP/1.1 100 Continue

HTTP/1.1 200 OK

Content-Type: text/xml; charset="utf-8"

Server: MS .NET Remoting, MS .NET CLR 1.1.4322.2032

Content-Length: 1734

<SOAP-ENV:Body>

<i2:PassDataSetResponse id="ref-1" xmlns:i2="http://schemas.microsoft.com/clr/nsassem/RemoteObjectNamespace.RemoteObject/RemoteObject">

</i2:PassDataSetResponse>

<a1:DataSet id="ref-4" xmlns:a1="http://schemas.microsoft.com/clr/nsassem/System.Data/System.Data%2C%20Version%3D1.0.5000.0%2C%20Culture%3Dneutral%2C%20PublicKeyToken%3Db77a5c561934e089">

<xs:schema id="emptydataset" xmlns="" xmlns:xs="http://www.w3.org/2001/XMLSchema" xmlns:msdata="urn:schemas-microsoft-com:xml-msdata">

<xs:element name="emptydataset" msdata:IsDataSet="true">

<xs:complexType>

<xs:choice maxOccurs="unbounded">

<xs:element name="Table1">

<xs:complexType>

</xs:complexType>

</xs:element>

</xs:choice>

</xs:complexType>

</xs:element>

</xs:schema></XmlSchema>

</a1:DataSet>

</SOAP-ENV:Body>

</SOAP-ENV:Envelope>

If you modify the method definition so that the dataset is not passed “by reference” the incoming remoting message will remain the same. The return message from the server will not have the <a1:DataSet…></a1:DataSet> element.

As you can imagine, this could severely impact your applications performance depending on the size of your dataset. There’s a KB article that talks about improving dataset Serialization that may help here: Q892740 Improving DataSet Serialization and Remoting Performance. However, if you’re not passing in a DataSet but rather a complex object type this won’t help. What to do? The best thing would to create your own custom serializer or only serialize those properties of your object that will be modified on the other side of your remoting boundarie.

One quick note: This marshaling behavior works for the DataSet since it is marked as serializable. If the parameter you are trying to pass as “ref” or “byref” isn’t serializeable you’ll get the following exception:

An unhandled exception of type 'System.Runtime.Serialization.SerializationException' occurred in mscorlib.dll

Additional information: The type RemoteObjectNamespace.RemoteThrowException in Assembly RemoteObject, Version=1.0.1851.26880, Culture=neutral, PublicKeyToken=null is not marked as serializable.

So, where does that leave us? If you truly want to marshal an object across remoting boundaries via reference your object has to inherit from MarshalByRefObject or implement the necessary interfaces. If that isn’t an option you have to make some performance tradeoffs. It’s all in a days work for you architects!

Getting a NullReference Exception Generating WSDL for your Remoted Object?

robgruen — Thu, 16 Dec 2004 23:11:00 GMT

Maybe you didn’t know, but you can generate a WSDL document for your remote object fairly easily. You can generate the contract XML directly or you can use soapsuds.exe to generate a proxy dll that you can share with your consumers. All you have to do is make a request to your remote object URI and append ?WSDL and you’re done.

So let’s say I was hosing a remote object called RemOBj on my local machine on port 8081 and I specified it’s URI as RemOBJ.rem. To activate this object I’d make remote requests to: http://localhost:8081/RemoOBJ.rem. If I wanted to get the WSDL for this I’d request http://localhost:8081/RemoOBJ.rem?wsdl. You can make this request in IE for your remote object to see the output.

There are some cases however where this will fail and in IE you’ll see the following information:

System.NullReferenceException: Object reference not set to an instance of an object. at System.Runtime.Remoting.MetadataServices.RealSchemaType.Resolve(StringBuilder sb) at System.Runtime.Remoting.MetadataServices.XMLNamespace.Resolve() at System.Runtime.Remoting.MetadataServices.WsdlGenerator.Resolve() at System.Runtime.Remoting.MetadataServices.WsdlGenerator.Generate() at System.Runtime.Remoting.MetadataServices.SUDSGenerator.Generate() at System.Runtime.Remoting.MetadataServices.MetaData.ConvertTypesToSchemaToStream(ServiceType[] serviceTypes, SdlType sdlType, Stream outputStream) at System.Runtime.Remoting.MetadataServices.SdlChannelSink.GenerateSdl(SdlType sdlType, IServerResponseChannelSinkStack sinkStack, ITransportHeaders requestHeaders, ITransportHeaders responseHeaders, Stream& outputStream) at System.Runtime.Remoting.MetadataServices.SdlChannelSink.ProcessMessage(IServerChannelSinkStack sinkStack, IMessage requestMsg, ITransportHeaders requestHeaders, Stream requestStream, IMessage& responseMsg, ITransportHeaders& responseHeaders, Stream& responseStream) at System.Runtime.Remoting.Channels.Http.HttpServerTransportSink.ServiceRequest(Object state) at System.Runtime.Remoting.Channels.SocketHandler.ProcessRequestNow()

If you tried to create a proxy with soapsuds you’d see this exeception:

Error: Unable to retrieve schema from url: http://localhost:8081/RemOBJ.rem?wsdl, The remote server returned an error: (500) Internal Server Error.

If you are seeing either of these and your remote object implementation and remote object interface are defined in separate assemblies you can try one of the following:

1. Place the interface definition and implementation in the same assembly.

2. Ensure that the interface and implementation are rooted in different namespaces.