Status
Current state: Under Discussion Accepted
Discussion thread: http://mail-archives.apache.org/mod_mbox/samza-dev/201703.mbox/browser
JIRA: SAMZA-1126
Released: 0.13.0
Problem
In Samza 0.12, we introduced the StreamProcessor API to support Samza as a Library model. That introduced "processorId" that is synonymous to the logical "containerId" that Samza assigns in a fixed-set container model.
However, having an "int" processorId is insufficient to uniquely identify a processor in various runtime environment(s). This proposal is to define proper semantics to processorId and how it will be used in the samza tech stack.
...
Code Block | ||||
---|---|---|---|---|
| ||||
public interface JobCoordinator { ... // Before // int getProcessorId(); String getLocalProcessorIdgetProcessorId(); ... } |
New Interface
Code Block | ||||
---|---|---|---|---|
| ||||
/** * To be used by the JobCoordinator to generator ProcessorId * It can be configured by using a configuration - app.processor-id-generator.class */ public interface ProcessorIdGenerator { /** * Generates a String representation to identify the processor instance * This value can be representative of its current executing environment. It can also be custom-managed by the user. * **Note**: processorId has to be unique among the processors within a job * @return String , identifier for the processor */ String generateProcessorId(); } // Default (after 0.13+) public class UUIDGenerator implements ProcessorIdGenerator { String generateProcessorId() { return UUID.randomUUID(); } } |
...