Details
-
Bug
-
Status: Closed
-
Minor
-
Resolution: Fixed
-
1.0.0
-
None
-
None
Description
Switch to hadoop's MapWritable broke CrawlDbMerger. Part of the reason is that we reuse the same MapWritable instance during reduce
which apparently is a big no-no for hadoop's MapWritable. Also, hadoop's MapWritable#putAll doesn't work (I think.... see HADOOP-5142),
so we should also work around that.