Are you sure nodes are discovering each other and there are no topology changes in the middle? Sounds like you're sporadically losing some data which can happen when you lose too many nodes at a time. Can you try to change the cache mode to REPLICATED or increase number of backups?
-Val -- View this message in context: http://apache-ignite-users.70518.x6.nabble.com/Running-Spark-app-in-cluster-tp10073p10080.html Sent from the Apache Ignite Users mailing list archive at Nabble.com.