See #63
All the bolts are set to have as many instances as there are workers, apart from the WARC Bolt. There is no risk of collision in the outputs as the instance number is used in the filename so there is no reason not to use more than 1 instance, especially given that it could save on serialization and sending heavy tuples across the cluster.
See #63
All the bolts are set to have as many instances as there are workers, apart from the WARC Bolt. There is no risk of collision in the outputs as the instance number is used in the filename so there is no reason not to use more than 1 instance, especially given that it could save on serialization and sending heavy tuples across the cluster.