[ 
https://issues.apache.org/jira/browse/FLINK-40729?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Martijn Visser updated FLINK-40729:
-----------------------------------
    Description: 
FLINK-40249 overtook the assertj, log4j and PostgreSQL bumps Dependabot filed, 
but {{mssql-jdbc}} and {{derby}} are still open and several drivers have 
drifted as far without Dependabot watching them.

{{mssql-jdbc}} is on {{10.2.1.jre8}}, inside the affected range of 
CVE-2025-59250, and so is the {{11.2.0.jre8}} that 
apache/flink-connector-jdbc#208 proposes; that branch is fixed from {{11.2.4}}. 
It goes to {{13.6.0.jre8}}, keeping the jre8 classifier so the module still 
builds to Java 8 bytecode. Also {{mysql-connector-j}} 8.2.0 to 26.7.0, 
{{ojdbc8}} 21.8.0.0 to 23.26.3.0.0, {{trino-jdbc}} 418 and its test image 451 
both to 483, {{oceanbase-client}} 2.4.8 to 2.4.19, {{h2}} 2.2.224 to 2.5.250, 
and {{openlineage}} 1.32.0 to 1.53.0 with {{META-INF/NOTICE}}. One commit each.

{{mockito-core}} is removed rather than bumped: the code style guide asks for 
reusable test implementations, it had three users and one was dead code. That 
also settles what blocked a bump, mockito 5 wanting {{objenesis}} 3.3 against 
the 2.1 we pin for {{kryo}}.

The Trino tests are re-enabled; they were disabled for a Jenkins container 
problem and we run on GitHub Actions. Derby stays at {{10.14.2.0}}

  was:
Dependabot opened apache/flink-connector-jdbc#269 ({{snappy-java}} 1.1.10.4 to 
1.1.10.9) and #270 ({{guava}} 33.0.0-jre to 33.7.1-jre). Both change pins in 
the root {{dependencyManagement}} that only reach the tests and provided scope, 
and 33.7.1-jre is still inside the affected range of GHSA-xxph-c9ww-hj94, which 
is fixed in 33.7.2.

The dependency that does reach users is OpenLineage. A project that depends on 
{{flink-connector-jdbc-core}} resolves {{jackson-core}} and 
{{jackson-databind}} 2.18.9 through {{openlineage-java}} 1.53.0, inside the 
affected ranges of GHSA-7hhh-6rmp-j9qf, GHSA-p6pp-m3f8-5c89, 
GHSA-cxp5-3px4-pw24 and GHSA-wv8q-qhhj-9h54. The {{jackson-bom}} import in our 
pom does not reach consumers. {{openlineage}} 1.55.0 brings 2.18.11, which 
fixes all four.

One commit each:
* {{openlineage}} 1.53.0 to 1.55.0
* {{jackson-bom}} 2.22.2 to 2.22.4, for the same advisories in the tests
* {{guava}} 33.0.0-jre to 33.8.0-jre
* {{google-cloud-spanner-jdbc}} 2.44.0 to 2.49.0, which moves 
{{bcprov-jdk18on}} from 1.80 to 1.85
* Remove the core {{META-INF/NOTICE}}. OpenLineage is a regular dependency, not 
bundled, and Flink's license checker reports both entries as "not bundled, but 
listed".
* Remove the pins for {{snappy-java}}, {{scala-library}}, {{objenesis}}, 
{{kryo}}, {{commons-compress}}, {{commons-codec}}, {{httpclient}}, {{httpcore}} 
and {{checker-qual}}. The convergence check passes without each of them on 
Flink 2.1.3, 2.2.1, 2.3.0 and 2.4-SNAPSHOT. Most held the tests below what 
Flink ships, {{snappy-java}} at 1.1.10.4 against Flink's 1.1.10.7, and the 
{{com.esotericsoftware.kryo}} pin matches nothing since Flink 2.0. This 
supersedes #269.

{{commons-lang3}} stays pinned at 3.18.0, because the two OpenLineage artifacts 
ask for different versions and 3.18.0 is what Flink ships.


> Refresh the JDBC driver and test dependency versions raised by Dependabot
> -------------------------------------------------------------------------
>
>                 Key: FLINK-40729
>                 URL: https://issues.apache.org/jira/browse/FLINK-40729
>             Project: Flink
>          Issue Type: Technical Debt
>          Components: Connectors / JDBC
>            Reporter: Martijn Visser
>            Assignee: Martijn Visser
>            Priority: Major
>              Labels: pull-request-available
>             Fix For: jdbc-5.0.0
>
>
> FLINK-40249 overtook the assertj, log4j and PostgreSQL bumps Dependabot 
> filed, but {{mssql-jdbc}} and {{derby}} are still open and several drivers 
> have drifted as far without Dependabot watching them.
> {{mssql-jdbc}} is on {{10.2.1.jre8}}, inside the affected range of 
> CVE-2025-59250, and so is the {{11.2.0.jre8}} that 
> apache/flink-connector-jdbc#208 proposes; that branch is fixed from 
> {{11.2.4}}. It goes to {{13.6.0.jre8}}, keeping the jre8 classifier so the 
> module still builds to Java 8 bytecode. Also {{mysql-connector-j}} 8.2.0 to 
> 26.7.0, {{ojdbc8}} 21.8.0.0 to 23.26.3.0.0, {{trino-jdbc}} 418 and its test 
> image 451 both to 483, {{oceanbase-client}} 2.4.8 to 2.4.19, {{h2}} 2.2.224 
> to 2.5.250, and {{openlineage}} 1.32.0 to 1.53.0 with {{META-INF/NOTICE}}. 
> One commit each.
> {{mockito-core}} is removed rather than bumped: the code style guide asks for 
> reusable test implementations, it had three users and one was dead code. That 
> also settles what blocked a bump, mockito 5 wanting {{objenesis}} 3.3 against 
> the 2.1 we pin for {{kryo}}.
> The Trino tests are re-enabled; they were disabled for a Jenkins container 
> problem and we run on GitHub Actions. Derby stays at {{10.14.2.0}}



--
This message was sent by Atlassian Jira
(v10.3.26#10030026)

Reply via email to