[
https://issues.apache.org/jira/browse/OAK-9345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17687486#comment-17687486
]
Perini Nicola commented on OAK-9345:
------------------------------------
A similar issue is occurring for the queries introduced
e.g.
{code:java}
select ID, MODIFIED, MODCOUNT, CMODCOUNT, HASBINARY, DELETEDONCE,
VERSION, SDTYPE, SDMAXREVTIME, T_DATA, BDATA from NODES where not (ID
like :1 or ID like :2 or ID like :3 ) and (VERSION is null or
VERSION < :4 ) {code}
> Query performance issue
> -----------------------
>
> Key: OAK-9345
> URL: https://issues.apache.org/jira/browse/OAK-9345
> Project: Jackrabbit Oak
> Issue Type: Bug
> Components: query
> Environment: DocumentNodeStoreService: RDB
> Oracle 18C
> Jackrabbit OAK 1.36
> Reporter: Ivar Suits
> Priority: Critical
> Attachments: repository-config_rdb.json
>
>
> Added over 5 million documents to Jackrabbit OAK after that performance
> problems started with following query. Using Oracle DB for
> DocumentNodeStoreService It s doing table access full scans.
> {code}
> select ID, MODIFIED, MODCOUNT, CMODCOUNT, HASBINARY, DELETEDONCE, VERSION,
> SDTYPE, SDMAXREVTIME, DATA, BDATA
> from NODES where ID > :1
> and ID < :2
> order by ID FETCH FIRST ? ROWS ONLY
> {code}
> Looks like Oracle DB oes not like ID column index and switches to table
> full scan, because ID column is not number. As a solution probably index
> hint need to be added and there should index name for ID column.
>
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
