Runtime environment
When creating a Yandex Data Processing cluster, you can choose the image version that determines component versions.
Below is a list of current and deprecated Yandex Data Processing images. Each image version includes conda, pip (Python environment managers), and a collection of pre-installed libraries.
Yandex Data Processing has no native mechanism for image version upgrades. To upgrade your image version, create a new cluster. To make sure the version you use is always up-to-date, automate the creation and removal of temporary Yandex Data Processing clusters using Yandex Managed Service for Apache Airflow™. To run jobs automatically, apart from Managed Service for Apache Airflow™ you can also use Yandex DataSphere.
Environment
When creating a cluster, you can choose one of the following environments:
PRODUCTION: For stable versions of your apps.PRESTABLE: For testing purposes. The prestable environment is similar to the production environment and likewise covered by an SLA, but it is the first to get new features, improvements, and bug fixes. In the prestable environment, you can test new versions for compatibility with your application.
When you create a cluster, the environment affects the choice of the image build, giving its version with accuracy down to the minor one. You start using new image builds:
- For
PRODUCTION: At least one week after the release. - For
PRESTABLE: Directly after the release.
Once stabilized, each minor version supports backward compatibility. However, we recommend using a test configuration with the PRESTABLE environment for processes requiring regular creation of clusters. This will allow you so detect likely backward compatibility issues earlier.
Once a cluster is created, the environment does not affect its operation. You cannot change an existing cluster's environment.
Current images
Note
Access to image 2.2 is provided on request. Contact support or your account manager.
| Components | Image 2.11 | Image 2.2 (beta) |
|---|---|---|
| Component versions | ||
| Hadoop | 3.3.2 | 3.3.2 |
| Tez | 0.10.1 | — |
| Spark | 3.3.2 | 3.5.0 |
| Zeppelin | 0.10.0 | — |
| Livy | 0.8.0 | 0.8.0 |
| Versions of Python and machine learning libraries | ||
| Python | 3.8.13 | 3.11.10 |
| PyArrow | 4.0.0 | 14.0.2 |
| ipykernel | 5.3.4 | 6.29.5 |
| PyHive | 0.6.1 | 0.7.0 |
| scikit-learn | 0.24.1 | 1.5.1 |
| pandas | 1.2.4 | 2.2.2 |
| koalas | 1.8.2 | — |
| numpy | 1.20.1 | 1.26.4 |
| boto3 | 1.16.7 | 1.34.154 |
| IPython | 7.22.0 | 8.27.0 |
| Matplotlib | 3.4.2 | 3.9.2 |
1 Stable since 2.1.15.
2 Spark 3.3.2 is supported in Yandex Data Processing images starting from version 2.1.4. Images versions 2.1.1-2.1.3 contain Spark 3.2.1.
Deprecated images
Note
These images are deprecated. We recommend using the latest image versions. Existing clusters will continue running, but you will not be able to create new clusters with deprecated versions.
| Components | Image 1.4 | Image 2.0 |
|---|---|---|
| Component versions | ||
| Hadoop | 2.10.0 | 3.2.2 |
| Tez | 0.9.2 | 0.10.0 |
| Hive | 2.3.6 | 3.1.2 |
| Zookeeper | 3.4.14 | 3.4.14 |
| HBase | 1.3.5 | 2.2.7 |
| Sqoop | 1.4.7 | — |
| Oozie | 5.2.0 | 5.2.1 |
| Spark | 2.4.6 | 3.0.3 |
| Flume | 1.9.0 | — |
| Zeppelin | 0.8.2 | 0.9.0 |
| Livy | 0.7.0 | 0.8.0 |
| Versions of Python and machine learning libraries | ||
| Python | 3.7.9 | 3.8.10 |
| PyArrow | 0.13.0 | 1.0.1 |
| ipykernel | 5.1.3 | 5.3.4 |
| TensorFlow | 1.15.0 | — |
| CatBoost | 0.20.2 | — |
| PyHive | 0.6.1 | 0.6.1 |
| LightGBM | 2.3.0 | — |
| XGBoost | 0.90 | — |
| scikit-learn | 0.21.3 | 0.23.2 |
| pandas | 0.25.3 | 1.1.3 |
| koalas | — | 1.7.0 |
| numpy | — | 1.19.2 |
| boto3 | — | 1.16.7 |
| IPython | 7.9.0 | 7.19.0 |
| Matplotlib | 3.1.1 | 3.2.2 |