Ververica Platform 2.15.14
On this page
Release Date: 2026-11-20
Changelog
Apache Flink®
Ververica Platform 2.15.14 supports the following versions:
- Apache Flink® 1.20
- Apache Flink® 1.19
- Apache Flink® 1.18
Ververica Platform 2.15.14 supports Apache Flink® 1.20, Apache Flink® 1.19, and Apache Flink® 1.18 under SLA.
For Stream Edition:
- 1.18.1-stream12-scala_2.12-java8
- 1.18.1-stream12-scala_2.12-java11
- 1.18.1-stream12-scala_2.12-java17
- 1.19.3-stream9-scala_2.12-java8
- 1.19.3-stream9-scala_2.12-java11
- 1.19.3-stream9-scala_2.12-java17
- 1.20.5-stream3-scala_2.12-java8
- 1.20.5-stream3-scala_2.12-java11
- 1.20.5-stream3-scala_2.12-java17
For Spring Edition, the following archives are available:
This is a patch release with improvements and security fixes.
Improvements
- Namespace roles with multiple bindings. When a user matches more than one role binding in a namespace, for example Viewer through one group and Editor through another, Ververica Platform now grants the strongest role: Owner, then Editor, then Viewer. Previously, the effective role depended on the order in which the identity provider listed the user's groups, so it could change between sessions.
Behavior change: a user binding no longer lowers a user below a role they hold through a group. To give a user a lower role, remove the group binding or bind that group to a lower role.
NumPy 2 in the Apache Flink® 1.19 and 1.20 images
The PyFlink environment in the Stream Edition images for Apache Flink® 1.19 and 1.20 (1.19.3-stream9-* and 1.20.5-stream3-*) now ships NumPy 2.4.6 instead of NumPy 1.24. This is required by pyarrow 26.0.1, which fixes CVE-2026-111497.
Python libraries with compiled extensions are built against a specific NumPy major version. A library compiled against NumPy 1.x cannot be loaded with NumPy 2. If a PyFlink job ships or installs such a library, the job fails to start.
Pure-Python jobs, and jobs whose dependencies are already NumPy 2 compatible, are not affected. Deployments can keep using the previous images (1.19.3-stream8-* and 1.20.5-stream2-*) until their dependencies are updated.
Bug Fixes
This release contains no bug fixes.
Vulnerability Fixes (Inside Apache Flink®)
Apache Flink® 1.19 and 1.20
- Updated dnspython to 2.9.2 to address CVE-2026-128089
- Updated msgpack to 1.2.4 to address CVE-2026-114613, GHSA-6v7p-g79w-8964
- Updated pyarrow to 26.0.1 to address CVE-2026-111497. pyarrow 26.0.1 requires NumPy 2, so NumPy is updated to 2.4.6.
- Updated pymongo to 4.18.3 to address CVE-2026-106435
- Updated urllib3 to 2.8.0 to address CVE-2026-97687, CVE-2026-97688, CVE-2026-97689
Apache Flink® 1.18
- Updated msgpack to 1.2.4 to address CVE-2026-114613, GHSA-6v7p-g79w-8964
- Updated pymongo to 4.18.3 to address CVE-2026-106435
- Updated urllib3 to 2.8.0 to address CVE-2026-97687, CVE-2026-97688, CVE-2026-97689
Vulnerability Fixes (Outside of Apache Flink®)
- Updated com.google.guava:guava to 33.7.2-jre to address CVE-2026-102554
- Updated dnspython to 2.9.2 to address CVE-2026-128089
- Updated ognl:ognl to 3.4.16 to address CVE-2026-113730
- Updated org.apache.kerby:kerby-asn1 to 2.1.2 to address CVE-2026-57914
- Updated org.json:json to 20260719 to address CVE-2026-59171
Upgrade
We recommend upgrading with Helm using the following commands. Replace <release-name> with the name of your Ververica Platform Helm release:
1$ helm repo add ververica https://charts.ververica.com
2$ helm repo update
3$ helm upgrade <release-name> ververica/ververica-platform --version 5.11.14 --values custom-values.yaml