[ https://issues.apache.org/jira/browse/CASSANDRA-9472?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15115010#comment-15115010 ]
Benedict commented on CASSANDRA-9472: ------------------------------------- Thanks. bq. Did you mean to override deletionInfo() in AtomicBTreePartition? I've removed it in the nits below. I did, I just forgot to put corresponding functionality into {{EnsureOnHeap}}, I've reintroduced it to my copy of the branch, with it added back bq. Looking at the last commit, is it necessary to make a local copy of this.peer? It's a final field initialized in the constructor so why the copy? I assumed that (as it used to be) the peer could be set to zero on free, and that this was how you were encountering the problem. I see now it is only the fake NativeCell that causes problems; I do wonder if we shouldn't simply allocate it for real, to avoid the check for every single call to a cell. It's also highly unlikely to ever appear in anyone's debugger if it's only that specific cell, so we could simply ignore the problem entirely. bq. How do you run the tests given that you've changed cassandra.yaml to use heap_buffers again You're right. I ran the first batch with it offheap, and then changed it back to restore it as I don't like having that modified, but it was premature. I've restored offheap_objects to the base cassandra.yaml > Reintroduce off heap memtables > ------------------------------ > > Key: CASSANDRA-9472 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9472 > Project: Cassandra > Issue Type: Improvement > Reporter: Benedict > Assignee: Benedict > Fix For: 3.x > > > CASSANDRA-8099 removes off heap memtables. We should reintroduce them ASAP. -- This message was sent by Atlassian JIRA (v6.3.4#6332)