On Mon, Nov 19, 2007 at 12:00:42PM -0600, jonathan doklovic wrote: >Hi, > >I have a map task that uses a regex to parse data. >Sometimes the regex matcher throws an unavoidable StackOverflowError. > >I'm catching these errors and want to skip the map/reduce entirely when >encountered. > >If I don't do anything in my catch, the job runs and shows a few failed >tasks. >
That seems like the right approach, first up. Either way, could you please provide more background, logs, error messages etc.? Arun PS: Some info on how to get help from hadoop-user@ - http://wiki.apache.org/lucene-hadoop/Help >If I put a return; within my catch to try to skip, the job goes into an >endless loop at some point with the map and reduce tasks stuck without >making any progress. > >Any suggestions? > >Thanks, > >- Jonathan
