[
https://issues.apache.org/jira/browse/FLINK-36696?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17897797#comment-17897797
]
Sai Sharath Dandi commented on FLINK-36696:
-------------------------------------------
[~fanrui] Sure! Please assign the task to me. Also, for anyone interested, we
implemented a thread-local connection proxy to work around this issue
internally before the change to the autoscaler.
> JDBC Autoscaler plugin should use DataSource and not connection
> ---------------------------------------------------------------
>
> Key: FLINK-36696
> URL: https://issues.apache.org/jira/browse/FLINK-36696
> Project: Flink
> Issue Type: Improvement
> Components: Autoscaler
> Reporter: Sai Sharath Dandi
> Priority: Major
>
> The flink-autoscaler-plugin-jdbc uses a single
> [Connection|https://github.com/apache/flink-kubernetes-operator/blob/main/flink-autoscaler-plugin-jdbc/src/main/java/org/apache/flink/autoscaler/jdbc/state/JdbcStateInteractor.java#L38]
> to interact with the Database. It is very suboptimal as multiple threads
> will not be able to query the database concurrently. Instead, we should use a
> datasource and request for connection from connection pool, each time we need
> one.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
