[
https://issues.apache.org/jira/browse/QPID-7974?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16209226#comment-16209226
]
Rob Godfrey commented on QPID-7974:
-----------------------------------
Hi Adel,
I think (and again, I'm going off memories from >4 years ago when I originally
committed this work) that while the tables where created in all uppercase, when
the metadata was brought back the table names were not in the same case as the
broker originally asked them to be created in. Unfortunately I can't remember
(and I should have commented in the code) which DB / OS this issue occurred on.
(Some googling suggests MySQL differs depending on the underlying operating
system and/or has a setting to force the actual table name to be stored all
lower case).
Another alternative would (assuming the table name has not been stored mixed
case) to run two queries, one with the table name forced upper case, and one
with the table name forced lower (maybe a third with the original case if it is
not covered by the prior two cases). That would presumably catch any sane DB
implementation.
> JdbcUtils.TableExists is very slow on big databases such as Oracle
> ------------------------------------------------------------------
>
> Key: QPID-7974
> URL: https://issues.apache.org/jira/browse/QPID-7974
> Project: Qpid
> Issue Type: Improvement
> Components: Java Broker
> Affects Versions: qpid-java-6.1.4
> Reporter: Adel Boutros
>
> The problem is that JdbcUtils.tableExists will actually load all tables and
> then iterate to find the correct one. On Oracle, this can be around 40 000
> tables. So it will take 12 minutes for the broker to start.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
