env.writeCollectionName(collectionName);
env.writeCollectionPath(collectionPath);
env.writeInputFormat(inputFormat);
env.writeDocnoMappingClass(mappingClass);
env.writeTokenizerClass(tokenizer);
env.writeDocnoOffset(docnoOffset);
conf.set("mapreduce.task.timeout", "6000000"); // needed for stragglers (e.g., very long documents in Wikipedia)
conf.set("mapreduce.map.memory.mb", "2048");
conf.set("mapreduce.map.java.opts", "-Xmx2048m");