[ https://issues.apache.org/jira/browse/CASSANDRA-19939?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17884532#comment-17884532 ]
Stefan Miklosovic edited comment on CASSANDRA-19939 at 9/25/24 8:16 AM: ------------------------------------------------------------------------ What is even more interesting is that when I iterate over partitions, then in order to see if there is some tombostone on a row (ttl, cell, range), the way it works is that if I iterate over rows, not only over parititions, in order to see if there is some tombstone or not, and I evaluate that yes, this is a tombstone, so I am going to render it, then it will actually skip it. Imagine that as if by mere looking into a row if it is a tombstone or not, the iterator already looked at that, and that iterator "remembers" what rows it iterated over. It is just one pass. was (Author: smiklosovic): What is even more interesting is that when I iterate over partitions, then in order to see if there is some tombostone on a row (ttl, cell, range), the way it works that if I iterate over rows, not only over parititions, in order to see if there is some tombstone or not, and I evaluate that yes, this is a tombstone, so I am going to render it, then it will actually skip it. Imagine that as if by mere looking into a row if it is a tombstone or not, the iterator already looked at that, and that iterator "remembers" what rows it iterated over. It is just one pass. > Add sstabledump option to display only tombstones > ------------------------------------------------- > > Key: CASSANDRA-19939 > URL: https://issues.apache.org/jira/browse/CASSANDRA-19939 > Project: Cassandra > Issue Type: New Feature > Components: Tool/sstable > Reporter: Brad Schoening > Assignee: Stefan Miklosovic > Priority: Normal > Time Spent: 10m > Remaining Estimate: 0h > > The utility sstabledump would be more useful if it had an option to display > just tombsones, much like the -e option to display only partition keys. > When tombstones exist, it should display the partition key, clustering > columns and the range, row or cell tombstone. > When investigating tombstone issues, a common way is to run sstabledump. > However, the output can be very large, on the order of gigabytes and scanning > the JSON output for tombstone entries is a hunt-and-peck experience or > requires using a tool like 'jq'. Being able to just view the tombstones > would eliminate the issue of the output being too large and not require > extraneous tools. -- This message was sent by Atlassian Jira (v8.20.10#820010) --------------------------------------------------------------------- To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org