[
https://issues.apache.org/jira/browse/IOTDB-5859?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17721679#comment-17721679
]
周沛辰 commented on IOTDB-5859:
*Description*
Starting from version 1.0, using the load function to import data and then
compact them will result in overlapped data in sequential files.
*Reason*
After version 1.0, using the load function may cause files with the same
timestamp to appear in the system, and the version corresponding to the
sequential file with the same timestamp is incremented, and the version of the
file under a time partition is unique. Compaction use {{version}} as first sort
dimension, resulting in an incorrect order.
*Solution*
Use {{timestamp}} as first sort dimension and {{version}} as second sort
dimension.
> Compaction error when using Version as first sort dimension
> ---
>
> Key: IOTDB-5859
> URL: https://issues.apache.org/jira/browse/IOTDB-5859
> Project: Apache IoTDB
> Issue Type: Improvement
>Reporter: Jinrui Zhang
>Assignee: 周沛辰
>Priority: Major
> Labels: pull-request-available
> Fix For: 1.1.1
>
>
> In current implementation of compaction, the default sort dimension is file
> version when selecting compaction tasks. It will lead to compaction error in
> tsfile load scenario. It is because the TsFile with higher version may not
> has greater timestamp when the TsFile is loaded by tools.
>
> Solution:
> change the sort dimension from file version to timestamp.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)