Windows Gateway Services Configuration Guide

For the MSMQ connector running in the Mule Runtime, the service first authenticates the call by inspecting the HTTP Authorization header as specified in RFC 2616.

The authentication is done through this unique security token used by the MSMQ connector. This token is included on every HTTP request to the Gateway using the Mule scheme. The following example shows how MSMQ leverages the Gateway to connect to a specific queue sending the security token as specified in this section:

The authorization token should match on the connector and the Gateway configuration file. The following configuration setting shows how the token is set within the Gateway configuration file Mule.SelfHost.exe.config :

When you configure the connector running in the Mule Runtime, set the authorization token value using the Gateway Access Token configuration setting (accessToken attribute at the connector’s XML configuration).

Note: The installer for the Windows Gateway service automatically generates a cryptographically secure token for use by callers upon first install. This token is displayed and placed upon the clipboard during installation for easy copying into a Mule application.

Users executing the call on behalf of a Gateway-served connector authenticate through two custom HTTP headers, mule-impersonate-username and mule-impersonate-password.

These two headers represent the Windows credentials of an existing user in the Active Directory forest where the Windows Gateway service is running, or a local account on the machine hosting the service. When these HTTP headers are included in an HTTP Request, the Windows Gateway service authenticates and impersonates this user before executing the operation required by the connector. This provides the ability to configure the correct access control list permissions using Windows credentials.

In the following table you can find the configuration settings that are only related to the MSMQ connector:

When your queue is marked to require authentication, you can impersonate the caller user as specified in the configuration settings. Additionally, if you work with a remote queue, the connector has a particular header to override this behavior.

If the Anypoint Gateway service runs on one machine and must access a remote queue on another machine, you must do one of these options (based on your environment configuration):

Windows Gateway Services supports running in a load-balanced configuration to allow for fault tolerance. When running multiple Gateway Services instances, each member must be configured to perform MSMQ background jobs at a non-overlapping interval.

MSMQ background job processing is by default performed every 10 minutes starting at zero minutes past the hour. To prevent multiple gateway instances simultaneously attempting to perform cleanup on the queues when running in a load-balanced configuration, a setting called cleanup-delay must be specified on each gateway instance. The recommended value to use for this on each machine is (10 / instanceCount) * (instanceNumber - 1) where instanceNumber is an integer value 1..n.

For example, a cluster of two machines would use a cleanup-delay of 0 on machine 1 and 5 on machine 2. A cluster of three machines would use a cleanup-delay of 0 on machine 1, 3 on machine 2, and 6 on machine 3. Synchronize the clocks for the machines in your cluster by NTP or an equivalent mechanism to ensure this offset is applied correctly.

The cleanup-delay setting is in the Mule.SelfHost.config file:

Note: When running in a load-balanced configuration, the Gateway Services must configured to run as 'Administrator' when the nodes involved (MSMQ, gateways) are under a WORKGROUP but not joined to a DOMAIN. When joined to the same DOMAIN the permissions for each of the nodes and objects involved (queues) should be correctly set by the domain’s administrator.

Windows Gateway Services is configured to log Information events. This is configured in the <switches> element. If you want to log everything, use the Verbose level by changing it in the configuration element.

Configure the tracing levels at the <switch> level in the configuration file:

Other possible levels are:

To trace or debug the routing service within the Windows Gateway Services, there is a setting to enable to get more details about the error that was generated while connecting to the routing service. To get this information at the tracing listeners, enable it using the includeExceptionDetailInFaults attribute in the serviceDebug element. To do this, set its value to true:

This setting extends the error message returned by the service and adds an internal stack trace of the cause, which in some scenarios may help you understand what the issue or problem is.

A useful way to troubleshoot issues is to enable the console listener (by default it is enabled, but if not then you should add it to the listeners section), and run the Windows Gateway Services from the command line.

Within the console, you can see real-time information being traced, like requests, responses, and some warnings or errors. These are useful to see if the connector is reaching the Gateway properly, or other possible causes that could be generating a fault.

Event Tracing for Windows (ETW) is a very efficient built-in publish and subscribe mechanism for doing event tracing at the kernel level. There is little overhead in using this feature compared to other traditional tracing solutions that rely on I/O for storing the traces in persistence storage such as files or databases. As a built-in mechanism in Windows, many of the operating systems services and components use this feature as well. For that reason, not only can you troubleshoot the application but also many of the OS components involved in the same execution.

In ETW, there are applications publishing events in queues (or providers) and other applications consuming events from those queues in real-time through ETW sessions. When an event is published in a provider, it goes nowhere unless there is a session collecting events on that queue. (The events are not persisted).

The tracing system in .NET includes a trace listener for ETW, EventProviderTraceListener, which you can configure with a session identifier, which ETW uses to collect traces:

In the example, the session is associated with this identifier:

{47EA5BF3-802B-4351-9EED-7A96485323AC}

This command transfers useful information into the dumpfile.xml text file. For more information, see Tracerpt.