[ https://issues.apache.org/jira/browse/CASSANDRA-14823?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16702747#comment-16702747 ]
mck edited comment on CASSANDRA-14823 at 11/29/18 6:01 AM: ----------------------------------------------------------- [~bdeggleston], correct me if i'm wrong please… For this to impact a user all the following must be true: - upgrading from 2.1, - clients were performing partition or multi-row deletes, creating range tombstones, - partitions have multiple rows and are {{>64KB}} (ie has IndexInfo column index blocks), - before the post-upgrade {{`nodetool upgradesstables`}} is complete, clients perform reads with an ordering clause (that reverses the clustering column's ordering on disk). That is a data-model that does not do {{DELETEs}} (or only does single-row or column {{DELETEs}}) will not be affected by this bug. Likewise a data-model that never specifies the "{{ORDER BY"}} clause in any {{SELECTs}} will not be affected by this bug. was (Author: michaelsembwever): [~bdeggleston], correct me if i'm wrong please… For this to impact a user all the following must be true: - upgrading from 2.1, - clients were performing partition or multi-row deletes, creating range tombstones, - partitions have multiple rows and are {{>64KB}} (ie has IndexInfo column index blocks), - before the post-upgrade {{`nodetool upgradesstables`}} is complete, clients perform reads with an ordering clause (that reverses the clustering column's ordering on disk). That is a data-model that does not do {{DELETEs}}, or only single-row or column {{DELETEs}}, will not be affected by this bug. Likewise a data-model that never specifies the "{{ORDER BY"}} clause in any {{SELECTs}} will not be affected by this bug. > Legacy sstables with range tombstones spanning multiple index blocks create > invalid bound sequences on 3.0+ > ----------------------------------------------------------------------------------------------------------- > > Key: CASSANDRA-14823 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14823 > Project: Cassandra > Issue Type: Bug > Components: Local Write-Read Paths > Reporter: Blake Eggleston > Assignee: Blake Eggleston > Priority: Major > Fix For: 3.0.18, 3.11.4 > > > During upgrade from 2.1 to 3.0, reading old sstables in reverse order would > generate invalid sequences of range tombstone bounds if their range > tombstones spanned multiple column index blocks. The read fails in different > ways depending on whether the 2.1 tables were produced by a flush or a > compaction. -- 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