Async

The purpose of this page is to describe how the chuck-stack supports asynchronous events.

Summary

The database goes to great lengths to execute event/trigger functions in a synchronous transaction; however, there are times when these function's execution time can be unpredictable or long-running.

chuck-stack provides several ways to execute event functionality without blocking a transaction and creating a slow user experience.

Considerations

Here are our needs from our asynchronous solution:

audit - audit/prove that an event occured
replay - in case of error, provide a way to remedy a situation and re-execute the event
batch - provide a way to process multiple events in a single execution
notify - send pg_notify signal to provide a PostgreSQL specific signal (see Notify)
webhook - send a signal to resources outside of PostgreSQL

Usage

Here are the steps to perform an asynchronous operation:

Write a record into stk_async with the appropriate stk_async_type indicating that you need an operation performed
Determine how you wish to process the asynchronous operation:
- Create a function that batch processes the unprocessed records on a timer
- Create an application that uses the existing pg_notify() message that gets generated after insert to perform a more real-time execution (see Notify)
- Extend the chuck-stack to perform an http webhook to notify external systems of the pending operation

Details

Table: stk_async;stk_async_type