While pushing documents to an index with dynamic (large) mapping (having a big cluster state), i occasionaly get the following error (very few ones):
ProcessClusterEventTimeoutException[failed to process cluster event (put-mapping [my_data_type]) within 30s]
The master_timeout parameter is a good workaround for all index.* operations.
Is there any way to provide a master_timeout duration for bulk/index operations ? A cluster wide setting ?