You are viewing a plain text version of this content. The canonical link for it is here.
Posted to jira@kafka.apache.org by "Randall Hauch (JIRA)" <ji...@apache.org> on 2017/10/04 23:22:00 UTC

[jira] [Issue Comment Deleted] (KAFKA-5953) Connect classloader isolation may be broken for JDBC drivers

     [ https://issues.apache.org/jira/browse/KAFKA-5953?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Randall Hauch updated KAFKA-5953:
---------------------------------
    Comment: was deleted

(was: A potential workaround is to put all of the JDBC driver JARs on the classpath. Obviously this is not ideal, and we're looking at a proper fix that hopefully will make 1.0.)

> Connect classloader isolation may be broken for JDBC drivers
> ------------------------------------------------------------
>
>                 Key: KAFKA-5953
>                 URL: https://issues.apache.org/jira/browse/KAFKA-5953
>             Project: Kafka
>          Issue Type: Bug
>          Components: KafkaConnect
>    Affects Versions: 0.11.0.0
>            Reporter: Jiri Pechanec
>            Priority: Critical
>
> Let's suppose there are two connectors deployed
> # using JDBC driver (Debezium MySQL connector)
> # using PostgreSQL JDBC driver (JDBC sink).
> Connector 1 is started first - it executes a statement
> {code:java}
> Connection conn = DriverManager.getConnection(url, props);
> {code}
> As a result a {{DriverManager}} calls {{ServiceLoader}} and searches for all JDBC drivers. The postgres driver from connector 2) is found associated with classloader from connector 1).
> Connector 2 is started after that - it executes a statement
> {code:java}
> connection = DriverManager.getConnection(url, username, password);
> {code}
> DriverManager finds the connector that was loaded in step before but becuase the classloader is different - now we use classloader 2) so it refuses to load the class and no JDBC driver is found.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)