[ https://issues.apache.org/jira/browse/CASSANDRA-14838?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16665006#comment-16665006 ]
Sam Tunnicliffe commented on CASSANDRA-14838: --------------------------------------------- +1. Only nit is the unnecessary throws clauses in the test setup method declarations. > Dropped columns can cause reverse sstable iteration to return prematurely > ------------------------------------------------------------------------- > > Key: CASSANDRA-14838 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14838 > Project: Cassandra > Issue Type: Bug > Reporter: Blake Eggleston > Assignee: Blake Eggleston > Priority: Major > Fix For: 3.0.18, 3.11.4, 4.0 > > > CASSANDRA-14803 fixed an issue where reading legacy sstables in reverse could > return early in certain cases. It's also possible to get into this state with > current version sstables if there are 2 or more indexed blocks in a row that > only contain data for a dropped column. Post 14803, this will throw an > exception instead of returning an incomplete response, but it should just > continue reading like it does for legacy sstables -- This message was sent by Atlassian JIRA (v7.6.3#76005) --------------------------------------------------------------------- To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org