Logging is an important element of the application development process, but not always given the attention it deserves. As a developer, there are various types of logs that you will need to handle in the application such as: logs related to business events, operational conditions, user interaction, exceptions, debug logs, etc. In this blog post I will discuss logging related to Solace messaging and how to handle them effectively in your application.
I will present a general overview of application logging in the context of Solace messaging first, followed by other implicit and external logging facilities available in Solace. I will follow this up with another post with code snippets and config files to handle logging in the application.
This post assumes working familiarity with basic messaging concepts and Solace features. You are encouraged to visit the Solace Developer portal to familiarize yourself with Solace concepts.
As with any other application, the question of what to log, how to log and how much to log in Solace applications is a subjective one. I will outline some recommendations for logging in Solace messaging context below. These should be considered along with existing guidelines in place in your project and application.
Depending on the API, a number of frameworks and engines are available for logging. Java applications for e.g., can use Log4j, Log4j2, Logback, SLF4J, etc. In all likelihood, your team or project may be already using a framework and as a result, your options may be limited. Solace Java API uses Jakarta Commons Logging to support different logging frameworks, such as log4j.
Almost all logging frameworks support logging levels; and use of the right level for logging can’t be stressed enough. Understanding the Solace philosophy to logging will help you align that with the logging you need from your application.
Since each framework defines logging levels of its own, Syslog logging levels are used for comparison. A brief description of each level is listed below along with how Solace SDKs use and map them.
|Log Level||JMS||JCSMP||Java RTO||C||.NET|
As an application developer using Solace SDKs, you should consider logging the following:
Solace events are sent by the API asynchronously to the application as they occur. For e.g., Solace API may be trying to restore lost connectivity between your application and the Solace message router. This would result in connection state events being sent to the application.
Solace API exceptions may be thrown when your application calls the Solace API. Which exceptions are thrown is dependent on the underlying conditions. For e.g., a JMS application may find the queue its trying to use not configured in JNDI store and hence the lookup failed. This would result in ‘javax.naming.NameNotFoundException’ thrown to the application.
An Application state is an application defined condition that you wish to log mainly to help troubleshooting and communicating. For e.g., you may want to identify and log your consumer application state when it is successfully connected, first message is consumed, etc.
Debug logs are, for most part are for consumption by developers and will leave it to your liking.
Not everything that’s available is worth logging. More logging doesn’t necessarily lead to better insights. Here are some potentially “high overhead” logging situations that you should avoid or log in the correct severity so it can be turned off in environments like Production.
Most projects and application teams have some guidelines in place for logging. Here are some additional guidelines related to Solace messaging that you should consider.
The Solace API sends several exceptions and events to publishers that may be useful for the publisher applications to log.
Persistent (guaranteed) publishers should log the publisher events such as the following:
Consumer applications should also handle Exceptions when interacting with Solace.
Application assigned ID: As the name suggests, this is generated by the publisher application and populated in custom message header of the message. This is the method for generating a unique ID since your application has complete control over it.
JMS Message ID: This is automatically generated by the API and populated in the JMS header of the message. In the absence of an application assigned ID, a JMS Message ID can be used as a unique ID. In a mixed API environment (JMS + non JMS) care should be taken to populate the equivalent headers from non JMS APIs.
AD Message ID: This is automatically generated by the Solace guaranteed messaging system, but it’s not guaranteed to be unique per message. For e.g., a retransmitted message on a non-exclusive queue will have different AD Message ID than the original message. Also, this applies only to persistent messages and is not always exposed to the application. For these reasons, AD Message ID is not recommended to be used as unique ID.
In addition to the application instrumented logging discussed above, Solace also provides API logs that detail API activities and Solace message router logs that provide a “Solace message router view” of your application and VPN. These logs are provided to you with no additional coding effort, but can be very valuable during troubleshooting.
All Solace APIs have implicit logging through natively supported language features and extensions. For instance, Solace JCSMP API uses Log4J and C API uses logging event callbacks. These are generated automatically and depending on the application logging configuration, they are logged to console, file etc.
The logs we have discussed so far are all generated on the application host either explicitly (application instrumented) or implicitly (API generated). Apart from these, Solace message router also has its own set of logs such as event logs, command logs, debug logs, etc. These logs are global in scope and usually managed centrally by middleware or system administrators. The Solace message router event log in particular, provides a “Solace message router view” of your VPN activities. This can augment your understanding when used along with the application and API logs during troubleshooting. Please refer to Monitoring events using syslog in the Solace message Bus Management document for details of the event log format and how to parse special events (like client disconnect events).
If the default Solace message router logging timestamp is not precise enough, you can enable millisecond timestamp for logging on the Solace message router. Please check the steps in Configuring Event output and thresholds for steps.
So far, we have covered essential background for application logging, logging facilities available in Solace and considerations for logging in different types of Solace messaging. I will follow this up with another post with details on handling them in specific APIs. I will leave you with the following links to explore until next time …