You are viewing a plain text version of this content. The canonical link for it is here.

Posted to notifications@iotdb.apache.org by "Haiming Zhu (Jira)" <ji...@apache.org> on 2022/02/11 05:25:00 UTC

[jira] [Assigned] (IOTDB-2528) MLog corruption and recovery bug after killing system

     [ https://issues.apache.org/jira/browse/IOTDB-2528?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Haiming Zhu reassigned IOTDB-2528:
----------------------------------

    Assignee: Haiming Zhu

> MLog corruption and recovery bug after killing system
> -----------------------------------------------------
>
>                 Key: IOTDB-2528
>                 URL: https://issues.apache.org/jira/browse/IOTDB-2528
>             Project: Apache IoTDB
>          Issue Type: Bug
>          Components: Core/Schema Manager
>    Affects Versions: 0.12.4
>            Reporter: Yukun Zhou
>            Assignee: Haiming Zhu
>            Priority: Major
>              Labels: pull-request-available
>             Fix For: 0.12.5
>
>         Attachments: image-2022-02-10-20-01-00-677.png, image-2022-02-11-11-38-50-593.png
>
>
> There're two problems.
> First, where creating timeseries via session, the usage of double quote in measurementId is not checked, which results in illegalPathException during revocery.
> Second, since mlog is not force sync at every metadata operation, some metadata may be lost after killing system during timeseries creation and recovery. Besides, the corrupted mlog will be used to recording new metadata operation, but the corrupted record will prevent mmanager from reading the following records, which also results in metadata lost.
> !image-2022-02-10-20-01-00-677.png!
>  
> !image-2022-02-11-11-38-50-593.png!



--
This message was sent by Atlassian Jira
(v8.20.1#820001)