You can trace service rules across a cluster. Cluster-wide tracing makes it easier for you to troubleshoot services in clustered configurations where a load balancer dynamically assigns service requests to nodes in the cluster. This functionality is especially useful for debugging mobile applications, RESTful Web Services, and other stateless service executions.
When tracing across the cluster, the trace feature only captures one trace at a time, which might result in some concurrent service executions not being traced. All other concurrent service executions in the cluster continue unimpeded. When the trace in progress completes, the next new service execution in the cluster is traced. For example, you might see traces for invocation one on node A, invocation two on node C, and invocation four on node B. This selective tracing approach ensures that tracing with a breakpoint will not stop all service executions across the cluster.
You can selectively disable cluster-wide tracing based on production level to eliminate
performance impact in production by using the
trace/cluster/ServiceRuleWatchMaxProductionLevel dynamic system setting.
Set the value to the maximum production level for which you want to allow RuleWatch tracing
for Rule-Service- data instances. The Trace option will not be shown in the
Actions menu for the rule form on systems with a production level
higher than the dynamic system setting's value.