How are you identifying that data is missing? How are your nodes configured? Are all nodes master-eligible? If so, have you set minimum_master_nodes to 2 as described here?
I did not experience this issue until now, the system was running like a clock for the past 1 year, suddenly from 2 weeks ago we experience inconsistencies in the data.
I believe the cluster settings are correct
All nodes master eligible and minimum master is set to 2
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.