[
https://issues.apache.org/jira/browse/STORM-3083?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
ASF GitHub Bot updated STORM-3083:
----------------------------------
Labels: pull-request-available (was: )
> Upgrade HikariCP version to 2.4.7
> ---------------------------------
>
> Key: STORM-3083
> URL: https://issues.apache.org/jira/browse/STORM-3083
> Project: Apache Storm
> Issue Type: Improvement
> Components: storm-jdbc
> Affects Versions: 2.0.0, 1.2.1
> Reporter: Jungtaek Lim
> Assignee: Jungtaek Lim
> Priority: Major
> Labels: pull-request-available
>
> I've encountered the issue which reason was version conflict regarding
> disruptor via Storm and Phoenix 5. I've filed STORM-3077 and resolved the
> root reason, but I also found that HikariCP doesn't give proper error message
> on this. It just complains about timeout while initializing pool, but the
> actual error was NoSuchMethodError.
> The issue in HikariCP version we use is that it leverages default
> implementation of ThreadPoolExecutor even while it checks fail-fast in
> initialization. So fail-fast works anyway, but it says "timed-out", not
> exposing the error.
> One thing to note is that HikariCP moved the main version line to support JDK
> 8 and change artifact name to "HikariCP-java7" on JDK7 compatible maintenance
> version. If we would like to provide same user experience, we should keep
> using "HikariCP" artifact, not "HikariCP-java7", since Maven might not
> recognize they're same.
> Hopefully I found there's another HikariCP version (2.4.7) which all of below
> conditions are met:
> * artifact name is "HikariCP", not "HikariCP-java7"
> * fail-fast works correctly
> * supports Java 7
>
> So upgrading HikariCP to 2.4.7 would resolve the issue in simplest and safest
> way.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
