[
https://issues.apache.org/jira/browse/CASSANDRA-8561?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14289357#comment-14289357
]
Jonathan Ellis commented on CASSANDRA-8561:
-------------------------------------------
It's not, because if you query for N rows of M columns each you can have a
maximum of N * M cells shadowed. (And no extra memory is used.) Tombstones
are evil because there's no upper limit to how many you'd have to scan, but
they all have to be sent to the coordinator for read repair, consuming memory
on the read thread.
> Tombstone log warning does not log partition key
> ------------------------------------------------
>
> Key: CASSANDRA-8561
> URL: https://issues.apache.org/jira/browse/CASSANDRA-8561
> Project: Cassandra
> Issue Type: Improvement
> Components: Core
> Environment: Datastax DSE 4.5
> Reporter: Jens Rantil
> Labels: logging
> Fix For: 2.1.3, 2.0.13
>
>
> AFAIK, the tombstone warning in system.log does not contain the primary key.
> See: https://gist.github.com/JensRantil/44204676f4dbea79ea3a
> Including it would help a lot in diagnosing why the (CQL) row has so many
> tombstones.
> Let me know if I have misunderstood something.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
