0

I am using Kundera as an ORM for cassandra. As part of our nightly job we try insert approximately 60k rows into cassandra daily. During the process if a row with PRIMARY key already exists (as may update primary key too), we do the following

  1. Delete
  2. Flush explicitly
  3. Insert
  4. Flush explicitly

Without calling Flush I noticed some inconsistency in final Total number of rows.

We end up in Heap out of memory as almost every second Full GC tries to happen. During this process as part of heap dump I found there are too many instances of 'com.impetus.kundera.persistence.event.EntityEventDispatcher'

Any thoughts why so ? Please help.

Vinod Jayachandran
  • 3,726
  • 8
  • 51
  • 88

1 Answers1

0

If these objects are not GC'd then its a bug. please raise it over github

vivek mishra
  • 1,162
  • 8
  • 16