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

Posted to dev@phoenix.apache.org by "Tanuj Khurana (Jira)" <ji...@apache.org> on 2025/02/16 00:29:00 UTC

[jira] [Resolved] (PHOENIX-7501) GC issues in TTLRegionScanner when gap is more than TTL

     [ https://issues.apache.org/jira/browse/PHOENIX-7501?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Tanuj Khurana resolved PHOENIX-7501.
------------------------------------
    Fix Version/s: 5.2.2
                   5.3
       Resolution: Fixed

> GC issues in TTLRegionScanner when gap is more than TTL
> -------------------------------------------------------
>
>                 Key: PHOENIX-7501
>                 URL: https://issues.apache.org/jira/browse/PHOENIX-7501
>             Project: Phoenix
>          Issue Type: Bug
>    Affects Versions: 5.2.0, 5.2.1
>            Reporter: Tanuj Khurana
>            Assignee: Tanuj Khurana
>            Priority: Major
>             Fix For: 5.2.2, 5.3
>
>
> TTLRegionScanner masks the expired rows before compaction purges the expired rows. The masking logic calculates the max and min timestamp of the latest cells of the row and if the difference between max and min is greater than TTL it does a gap analysis. In the gap analysis it does a raw scan and fetches all the versions of the row. In our production, we saw this gap analysis causing long GC pauses when the row has many versions.



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