In 2003, Andy Longshaw and Eoin Woods had an initial discussion about distributed error handling. Andy had been looking for some general advice or patterns related to this topic and both were slightly surprised that there was a lack of information as many designers and developers are engaged in creating distributed business systems. Because of this, they started to collate their experience of what they had found to work (and just as importantly what had not worked) in the area of error handling in the distributed systems they had been involved with.
Once they had captured a set of patterns and proto-patterns, they took this material to OT2004 where it was workshopped under the title of "What Went Wrong: Error handling in a distributed environment". This workshop led to refinements of some patterns, confirmation of some proto-patterns and a general tightening up of the concepts. The resultant paper can be downloaded here.
Andy delivered a session at OOP2005 on this topic - the slides from which are here.
Two additional patterns were workshopped at EuroPLoP 2005
We would welcome any comments or feedback you have from your experiences in this area.