Stopping write-behind retries in specific instances?

Question

I was wondering if it's possible to somehow stop particular types of write-behinds from being retried?
Every once in a while an app will submit a badly formatted object to our grid. We're fixing these things as we go and improving the validation (although the object format we inherited from our predecessors is pretty flexible and open to this kind of problem) but it does mean that from time to time we end up with an object that can't possibly be stored.
For example, a recent case involved a number being sent up as a string. When the sql was generated to try and store this object the error "Cannot implicitly convert varchar to float" was thown back.
I love the Coherence retry functionality but in this case it's a real pain as it keeps retrying a hopeless case.
Is there any way to intervene here?

Answer 1

Hi,
according to the documentation:
"If this operation fails (by throwing an exception) after a partial success, the convention is that entries which have been stored successfully are to be removed from the specified mapEntries, indicating that the store operation for the entries left in the map has failed or has not been attempted."
In other words,
1. You need to implement the storeAll method in a way that:
1.1 If there are entries to be retried, then you remove the particular entry which you do not want to retry from the map you received in the storeAll method, and Coherence will not attempt to retry that entry, and ultimately you throw an exception to indicate that there are entries to retry.
1.2 If there are no entries to be retried, then you swallow the exception.
2. You need to implement the store method in a way that if the single erroneous entry you do not want to retry, then you swallow the exception.
Best regards,
Robert