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

Posted to issues@phoenix.apache.org by "Istvan Toth (Jira)" <ji...@apache.org> on 2023/03/13 14:56:00 UTC

[jira] [Commented] (PHOENIX-6878) Upserts into tables with ROW_TIMESTAMP are extremely slow

    [ https://issues.apache.org/jira/browse/PHOENIX-6878?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17699684#comment-17699684 ] 

Istvan Toth commented on PHOENIX-6878:
--------------------------------------

I don't quite get why PHOENIX-4178 forces BOTH getting the timestamp from the server AND disables secondary indexes.

IUC disabling secondary indexes already solves the original consistency problem, so getting the server timestamp for each mutation is superfluous. 

HBase and Phoenix generally requires the time in the cluster to by in sync anyway.

> Upserts into tables with ROW_TIMESTAMP are extremely slow
> ---------------------------------------------------------
>
>                 Key: PHOENIX-6878
>                 URL: https://issues.apache.org/jira/browse/PHOENIX-6878
>             Project: Phoenix
>          Issue Type: Bug
>          Components: core
>    Affects Versions: 5.2.0, 5.1.3
>            Reporter: Istvan Toth
>            Priority: Major
>
> PHOENIX-4187 changed upserts into tables with ROW_TIMESTAMP fields to always refresh the table cache before executing the mutation.
> This kills non batched write performance for such tables, as the Table metadata is re-read for every row.



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