[
https://issues.apache.org/jira/browse/CASSANDRA-10775?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15028493#comment-15028493
]
Benjamin Lerer commented on CASSANDRA-10775:
--------------------------------------------
The problem comes from the fact that in
{{BigTableWriter.TransactionalProxy.doPrepare()}} the Metadata were computed
before the {{CompressedSequentialWriter}} could add the compression ratio to
the {{MetadataCollector}}.
> SSTable compression ratio is not serialized properly
> ----------------------------------------------------
>
> Key: CASSANDRA-10775
> URL: https://issues.apache.org/jira/browse/CASSANDRA-10775
> Project: Cassandra
> Issue Type: Bug
> Components: Local Write-Read Paths
> Reporter: Benjamin Lerer
> Assignee: Benjamin Lerer
> Fix For: 3.0.x, 3.x
>
>
> While reviewing CASSANDRA-10225, I found out that the compression ratio
> returned by the {{StatsMetadata}} for some compressed sstables was - 1
> (NO_COMPRESSION_RATIO).
> After investigation it seems that the wrong compression ratio was serialized
> to disk.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
