Streaming breakpoint for data tuples that stay in an operator graph too long

摘要

A streams manager monitors data tuples processed by a streaming application represented by an operator graph. The streams manager includes a tuple breakpoint mechanism that allows defining a tuple breakpoint that fires when a tuple has been in the operator graph too long. What constitutes too long can be defined in a number of different ways, including a time limit, a processing limit for multiple operators, and a processing limit for an individual operator. When the tuple breakpoint fires, one or more operators in the operator graph are halted according to specified halt criteria. Information corresponding to the breakpoint that fired is then displayed. The tuple breakpoint mechanism thus provides a way to debug a streaming application that may have data tuples that stay in the operator graph too long.

主权项

1. An apparatus comprising:
at least one processor; a memory coupled to the at least one processor; a streaming application residing in the memory and executed by the at least one processor, the streaming application comprising an operator graph that includes a plurality of operators that process a plurality of data tuples; a performance monitor residing in the memory and executed by the at least one processor, the performance monitor monitoring each of the plurality of data tuples processed by the plurality of operators in the operator graph to determine time for the plurality of operators to process each of the plurality of data tuples and a number of times each of the plurality of operators process each of the plurality of data tuples; and a tuple breakpoint mechanism residing in the memory and executed by the at least one processor, the tuple breakpoint mechanism including at least one tuple breakpoint that fires when at least one of the plurality of data tuples has been processed more than a specified limit for processing the at least one data tuple by all of the plurality of operators that process the at least one data tuple, and when the tuple breakpoint mechanism determines the tuple breakpoint fires based on information from the performance monitor, the tuple breakpoint mechanism halts at least one operator in the operator graph and displays information regarding a data tuple that caused the tuple breakpoint to fire.