[jira] [Resolved] (PHOENIX-7865) Set explicit HDFS block size on replication log files

["Tanuj Khurana (Jira)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/PHOENIX-7865?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Tanuj Khurana resolved PHOENIX-7865.
------------------------------------
    Resolution: Fixed

> Set explicit HDFS block size on replication log files
> -----------------------------------------------------
>
>                 Key: PHOENIX-7865
>                 URL: https://issues.apache.org/jira/browse/PHOENIX-7865
>             Project: Phoenix
>          Issue Type: Sub-task
>            Reporter: Tanuj Khurana
>            Assignee: Tanuj Khurana
>            Priority: Major
>
> The ReplicationLog used to clamp the configured max rotation size to the FileSystem default block size to keep each log file in a single HDFS block (so the disruptor consumer thread never makes an addBlock RPC to the NameNode). That clamp silently capped large rotation-size configs. But this coupling was too tight and sometimes could create too many log files for the ReplicationLogProcessor to process.
> Decouple the two concerns: stop clamping, and instead create each log file with an explicit per-file block size.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)