Follow-up on HSEARCH-1375.
The mass indexer executes many operations and is expected to run for a very long time. For that reason (probably?), the failure of some of these operations (indexing an entity) is not considered critical, and indexing proceeds if such a failure happens.
Since these failures can happen multiple times during this very long process, it can be useful for users to get more information more quickly:
Increment a counter in the MassIndexingMonitor for every entity that failed to load.
Report failed operations (flush, indexing of an entity, ...) to a MassIndexingFailureHandler provided by the user and specific to that mass indexer. This would provide the user with a stream of failures that is scoped to the mass indexer exclusively, which could be used to display the failures on a web interfaces, or more importantly to push the references to entities that failed indexing to a queue, to try again later. This may be a feature for a "niche" audience, but I suspect it wouldn't be hard to implement, since we already report all failing operations to a FailureHandler as of HSEARCH-1375.