Handling Retries Across Non-Completing States
Retries are only cancelled if a request/response has been marked as completing on the external domain (see Concepts). So you could have the case where you would like to retry a transaction across multiple states which could included non-completing request/responses.
In the following example we have modelled a request/response in an External Domain, of which one is non-completing (Completing flag is not ticked) and the other is completing (completing flag is ticked).
The event behaviour in this example simply moves the state based on the response; either Insufficient Funds or Balance OK, which would complete the flow:
When the Insufficient Funds Response is received, the state is moved to Insufficient Funds and the retries are not stopped because the response is a non-completing response. Retries will continue to occur even when we receive the first OK Response because we are now in state Insufficient Funds but the retries were firing from state Checking Balance.
You can stop this retry occurring once the completing response is received by configuring the retry properties as follows:
ipf.flow.Testflow.CheckingBalance.CheckBalance.initial-retry-interval = 1s
ipf.flow.Testflow.CheckingBalance.CheckBalance.max-retries = 0
ipf.flow.Testflow.InsufficientFunds.CheckBalance.initial-retry-interval = 1s
ipf.flow.Testflow.InsufficientFunds.CheckBalance.max-retries = 3 #or however many attempts
ipf.flow.Testflow.InsufficientFunds.CheckBalance.backoff-type = USER_DEFINED
ipf.flow.Testflow.InsufficientFunds.CheckBalance.jitter-enabled = false
ipf.flow.Testflow.InsufficientFunds.CheckBalance.custom-backoffs = [20s, 40s] #or whatever values you want hereThis configuration effectively moves the retry capabilities to the InsufficientFunds state, and we do not attempt any retries in the CheckingBalance state.