You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@phoenix.apache.org by "Gokcen Iskender (Jira)" <ji...@apache.org> on 2020/11/11 21:30:00 UTC
[jira] [Updated] (PHOENIX-6220) CREATE INDEX shouldn't ignore
IMMUTABLE_STORAGE_SCHEME and COLUMN_ENDCODED_BYTES
[ https://issues.apache.org/jira/browse/PHOENIX-6220?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Gokcen Iskender updated PHOENIX-6220:
-------------------------------------
Description: IMMUTABLE_STORAGE_SCHEME and COLUMN_ENCODED_BYTES that is specified in CREATE INDEX needs to be respected. Today, it is ignored and value is just gotten from the parent. Parent table can still use ONE_CELL_PER_COLUMN storage scheme. If the parent table is using COLUMN_ENCODED_BYTES=0 (none), we recommend using COLUMN_ENCODED_BYTES=2 for the index table by default (can be overriden) when SINGLE_CELL storage scheme is used. 2-byte column mapping enables to define 65535 columns which should be enough. (was: GlobalIndexChecker has repairIndexRows method, that builds a scan for data table and goes over indexMaintainer.getAllColumns to get columns from that. But the issue is there are only 2 columns for the SINGLE_CELL index. That part needs to change.)
> CREATE INDEX shouldn't ignore IMMUTABLE_STORAGE_SCHEME and COLUMN_ENDCODED_BYTES
> ----------------------------------------------------------------------------------
>
> Key: PHOENIX-6220
> URL: https://issues.apache.org/jira/browse/PHOENIX-6220
> Project: Phoenix
> Issue Type: Sub-task
> Reporter: Gokcen Iskender
> Priority: Major
>
> IMMUTABLE_STORAGE_SCHEME and COLUMN_ENCODED_BYTES that is specified in CREATE INDEX needs to be respected. Today, it is ignored and value is just gotten from the parent. Parent table can still use ONE_CELL_PER_COLUMN storage scheme. If the parent table is using COLUMN_ENCODED_BYTES=0 (none), we recommend using COLUMN_ENCODED_BYTES=2 for the index table by default (can be overriden) when SINGLE_CELL storage scheme is used. 2-byte column mapping enables to define 65535 columns which should be enough.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)