SDK version support status

This version will be deprecated on August 12, 2026.

This version will be deprecated on July 1, 2026.

This version will be deprecated on May 12, 2026.

This version will be deprecated on March 31, 2026.

This version will be deprecated on February 18, 2026.

This version will be deprecated on January 21, 2026.

Known issue: On Dataflow Streaming Appliance jobs, commits can fail with KeyCommitTooLargeException when a key outputs >180MB of results. Bug affects versions 2.60.x to 2.62.x, fix will be released with 2.63.0. To resolve this issue, downgrade to Apache Beam 2.59.0 or upgrade to Apache Beam 2.63.0 or enable Streaming Engine . See Apache Beam GitHub issue #33588 .

This version will be deprecated on November 25, 2025.

Known issue: On Dataflow Streaming Appliance jobs, commits can fail with KeyCommitTooLargeException when a key outputs >180MB of results. Bug affects versions 2.60.x to 2.62.x, fix will be released with 2.63.0. To resolve this issue, downgrade to Apache Beam 2.59.0 or upgrade to Apache Beam 2.63.0 or enable Streaming Engine . See Apache Beam GitHub issue #33588 .

This version will be deprecated on October 17, 2025.

Known issue: On Dataflow Streaming Appliance jobs, commits can fail with KeyCommitTooLargeException when a key outputs >180MB of results. Bug affects versions 2.60.x to 2.62.x, fix will be released with 2.63.0. To resolve this issue, downgrade to Apache Beam 2.59.0 or upgrade to Apache Beam 2.63.0 or enable Streaming Engine . See Apache Beam GitHub issue #33588 .

Deprecated on March 25, 2025.

Known issue: The WriteToJson transform in cross-language pipelines fails with an error. To resolve this issue, upgrade to Apache Beam 2.55.1. For more information, see Issue #30776: Beam YAML WriteToJson fails .

Deprecated on April 29, 2022.

Known issue: If you use Runner V2, by default Dataflow will use a Java 8 SDK container regardless of user Java version. To work around for Java 11, you can override the default SDK image by providing pipeline option `--worker_harness_container_image=gcr.io/cloud-dataflow/v1beta3/beam_java11_sdk:2.29.0`. See Apache Beam issue tracker for more information.

Deprecated on February 22, 2022.

Deprecated on January 7, 2022.

This release adds support for JDK 11.

Deprecated on December 11, 2021.

Deprecated on October 23, 2021.

Deprecated on September 18, 2021.

Deprecated on July 29, 2021.

Deprecated on June 8, 2021.

Deprecated on May 27, 2021.

Deprecated on April 15, 2021.

Deprecated on February 4, 2021.

Deprecated on January 23, 2021.

Deprecated on January 6, 2021.

Deprecated on October 7, 2020.

Deprecated on August 23, 2020.

• Dataflow users who use schema features (including SQL transforms) shouldn't upgrade to 2.15.0 due to a known issue. See the Apache Beam issue tracker for more information.

Deprecated on August 1, 2020.

Deprecated on June 6, 2020.

This release adds experimental support for JDK 9 and later versions. See the Apache Beam issue tracker for more information.

Deprecated on April 25, 2020.

• The Dataflow runner has an incorrect logging configuration that might cause all logs to be missing. To work around this issue, add slf4j-jdk14 to your runtime dependencies.

Deprecated as of March 1, 2020.

• The Dataflow runner has an incorrect logging configuration that might cause all logs to be missing. To work around this issue, add slf4j-jdk14 to your runtime dependencies.

Deprecated as of February 11, 2020.

• SDK 2.10.0 depends on gcsio client library version 1.9.13, which has known issues:
  • Reading side inputs can result in sending many requests to Cloud Storage. As a result, Dataflow jobs can fail with HTTP 429 errors from Cloud Storage.
  • Apache Beam ParquetIO fails when reading files from Cloud Storage.

  To work around these issues, either upgrade to SDK 2.11.0, or override the gcsio client library version to 1.9.16 or later.

• The Dataflow runner has an incorrect logging configuration that might cause all logs to be missing. To work around this issue, add slf4j-jdk14 to your runtime dependencies.

Deprecated as of December 13, 2019.

• If you enable the Streaming Engine (Beta) experiment , don't upgrade to SDK 2.9.0 due to a known issue. If you choose to use SDK 2.9.0, you must also set the enable_conscrypt_security_provider experimental flag to enable Conscrypt, which has known stability issues.
• In a specific case, users of Apache Beam Java SDKs 2.9.0 and earlier might experience data duplication when reading files from Cloud Storage. Duplication might occur when all of the following conditions are true:
  • You're reading files with the content-encoding set to gzip, and the files are dynamically decompressive transcoded by Cloud Storage.
  • The file size (decompressed) is larger than 2.14 GB.
  • The input stream runs into an error (and is recreated) after 2.14 GB is read.
  As a workaround, don't set the content-encoding header, and store compressed files in Cloud Storage with the proper extension (for example, gz for gzip). For existing files, you can update the content-encoding header and filename with the gcloud storage commands .

Deprecated as of October 25, 2019.

• Pipelines might become stuck due to an issue with the Conscrypt library. If you see errors in Stackdriver Logging with stack traces that include Conscrypt related calls, you might be affected by this issue. To resolve the issue, upgrade to SDK 2.9.0 or higher.
• In a specific case, users of Apache Beam Java SDKs 2.9.0 and earlier might experience data duplication when reading files from Cloud Storage. Duplication might occur when all of the following conditions are true:
  • You're reading files with the content-encoding set to gzip, and the files are dynamically decompressive transcoded by Cloud Storage.
  • The file size (decompressed) is larger than 2.14 GB.
  • The input stream runs into an error (and is recreated) after 2.14 GB is read.
  As a workaround, don't set the content-encoding header, and store compressed files in Cloud Storage with the proper extension (for example, gz for gzip). For existing files, you can update the content-encoding header and filename with the gcloud storage commands .

Deprecated as of October 3, 2019.

• Pipelines might become stuck due to an issue with the Conscrypt library. If you see errors in Stackdriver Logging with stack traces that include Conscrypt related calls, you might be affected by this issue. To resolve the issue, upgrade to SDK 2.9.0 or higher.
• In a specific case, users of Apache Beam Java SDKs 2.9.0 and earlier might experience data duplication when reading files from Cloud Storage. Duplication might occur when all of the following conditions are true:
  • You're reading files with the content-encoding set to gzip, and the files are dynamically decompressive transcoded by Cloud Storage.
  • The file size (decompressed) is larger than 2.14 GB.
  • The input stream runs into an error (and is recreated) after 2.14 GB is read.
  As a workaround, don't set the content-encoding header, and store compressed files in Cloud Storage with the proper extension (for example, gz for gzip). For existing files, you can update the content-encoding header and filename with the gcloud storage commands .

Deprecated as of August 8, 2019.

• Pipelines might become stuck due to an issue with the Conscrypt library. If you see errors in Stackdriver Logging with stack traces that include Conscrypt related calls, you might be affected by this issue. To resolve the issue, upgrade to SDK 2.9.0 or higher.
• In a specific case, users of Apache Beam Java SDKs 2.9.0 and earlier might experience data duplication when reading files from Cloud Storage. Duplication might occur when all of the following conditions are true:
  • You're reading files with the content-encoding set to gzip, and the files are dynamically decompressive transcoded by Cloud Storage.
  • The file size (decompressed) is larger than 2.14 GB.
  • The input stream runs into an error (and is recreated) after 2.14 GB is read.
  As a workaround, don't set the content-encoding header, and store compressed files in Cloud Storage with the proper extension (for example, gz for gzip). For existing files, you can update the content-encoding header and filename with the gcloud storage commands .

Deprecated as of June 6, 2019.

• Pipelines might become stuck due to an issue with the Conscrypt library. If you see errors in Stackdriver Logging with stack traces that include Conscrypt related calls, you might be affected by this issue. To resolve the issue, upgrade to SDK 2.9.0 or higher.
• In a specific case, users of Apache Beam Java SDKs 2.9.0 and earlier might experience data duplication when reading files from Cloud Storage. Duplication might occur when all of the following conditions are true:
  • You're reading files with the content-encoding set to gzip, and the files are dynamically decompressive transcoded by Cloud Storage.
  • The file size (decompressed) is larger than 2.14 GB.
  • The input stream runs into an error (and is recreated) after 2.14 GB is read.
  As a workaround, don't set the content-encoding header, and store compressed files in Cloud Storage with the proper extension (for example, gz for gzip). For existing files, you can update the content-encoding header and filename with the gcloud storage commands .

Decommissioned as of Aug 12, 2020.

This version is decommissioned due to the discontinuation of support for JSON-RPC and Global HTTP Batch Endpoints .

• In a specific case, users of Apache Beam Java SDKs 2.9.0 and earlier might experience data duplication when reading files from Cloud Storage. Duplication might occur when all of the following conditions are true:
  • You're reading files with the content-encoding set to gzip, and the files are dynamically decompressive transcoded by Cloud Storage.
  • The file size (decompressed) is larger than 2.14 GB.
  • The input stream runs into an error (and is recreated) after 2.14 GB is read.
  As a workaround, don't set the content-encoding header, and store compressed files in Cloud Storage with the proper extension (for example, gz for gzip). For existing files, you can update the content-encoding header and filename with the gcloud storage commands .

Decommissioned as of Aug 12, 2020.

This version is decommissioned due to the discontinuation of support for JSON-RPC and Global HTTP Batch Endpoints .

• In a specific case, users of Apache Beam Java SDKs 2.9.0 and earlier might experience data duplication when reading files from Cloud Storage. Duplication might occur when all of the following conditions are true:
  • You're reading files with the content-encoding set to gzip, and the files are dynamically decompressive transcoded by Cloud Storage.
  • The file size (decompressed) is larger than 2.14 GB.
  • The input stream runs into an error (and is recreated) after 2.14 GB is read.
  As a workaround, don't set the content-encoding header, and store compressed files in Cloud Storage with the proper extension (for example, gz for gzip). For existing files, you can update the content-encoding header and filename with the gcloud storage commands .

Decommissioned as of Aug 12, 2020.

This version is decommissioned due to the discontinuation of support for JSON-RPC and Global HTTP Batch Endpoints .

• In a specific case, users of Apache Beam Java SDKs 2.9.0 and earlier might experience data duplication when reading files from Cloud Storage. Duplication might occur when all of the following conditions are true:
  • You're reading files with the content-encoding set to gzip, and the files are dynamically decompressive transcoded by Cloud Storage.
  • The file size (decompressed) is larger than 2.14 GB.
  • The input stream runs into an error (and is recreated) after 2.14 GB is read.
  As a workaround, don't set the content-encoding header, and store compressed files in Cloud Storage with the proper extension (for example, gz for gzip). For existing files, you can update the content-encoding header and filename with the gcloud storage commands .

Decommissioned as of Aug 12, 2020.

This version is decommissioned due to the discontinuation of support for JSON-RPC and Global HTTP Batch Endpoints .

• In a specific case, users of Apache Beam Java SDKs 2.9.0 and earlier might experience data duplication when reading files from Cloud Storage. Duplication might occur when all of the following conditions are true:
  • You're reading files with the content-encoding set to gzip, and the files are dynamically decompressive transcoded by Cloud Storage.
  • The file size (decompressed) is larger than 2.14 GB.
  • The input stream runs into an error (and is recreated) after 2.14 GB is read.
  As a workaround, don't set the content-encoding header, and store compressed files in Cloud Storage with the proper extension (for example, gz for gzip). For existing files, you can update the content-encoding header and filename with the gcloud storage commands .

Decommissioned as of Aug 12, 2020.

This version is decommissioned due to the discontinuation of support for JSON-RPC and Global HTTP Batch Endpoints .

• In a specific case, users of Apache Beam Java SDKs 2.9.0 and earlier might experience data duplication when reading files from Cloud Storage. Duplication might occur when all of the following conditions are true:
  • You're reading files with the content-encoding set to gzip, and the files are dynamically decompressive transcoded by Cloud Storage.
  • The file size (decompressed) is larger than 2.14 GB.
  • The input stream runs into an error (and is recreated) after 2.14 GB is read.
  As a workaround, don't set the content-encoding header, and store compressed files in Cloud Storage with the proper extension (for example, gz for gzip). For existing files, you can update the content-encoding header and filename with the gcloud storage commands .

This version will be deprecated on August 12, 2026.

This version will be deprecated on July 1, 2026.

This version will be deprecated on May 12, 2026.

This version will be deprecated on March 31, 2026.

This version will be deprecated on February 18, 2026.

This version will be deprecated on January 21, 2026.

This version will be deprecated on November 25, 2025.

This version will be deprecated on October 17, 2025.

This version will be deprecated on September 11, 2025.

Deprecated on August 16, 2025.

Known issue: Pipelines that read data from Cloud Storage by using GcsIO, either directly or indirectly, such as with BigQueryIO, might experience data corruption on the Apache Beam Python SDK versions 2.53.0 to 2.58.1. For more information, see the Apache Beam GitHub issue #32169 .

Deprecated on August 6, 2025.

Known issue: Pipelines that read data from Cloud Storage by using GcsIO, either directly or indirectly, such as with BigQueryIO, might experience data corruption on the Apache Beam Python SDK versions 2.53.0 to 2.58.1. For more information, see the Apache Beam GitHub issue #32169 .

Deprecated on June 26, 2025.

Known issue: Pipelines that read data from Cloud Storage by using GcsIO, either directly or indirectly, such as with BigQueryIO, might experience data corruption on the Apache Beam Python SDK versions 2.53.0 to 2.58.1. For more information, see the Apache Beam GitHub issue #32169 .

Deprecated on May 1, 2025.

Known issue: Pipelines that read data from Cloud Storage by using GcsIO, either directly or indirectly, such as with BigQueryIO, might experience data corruption on the Apache Beam Python SDK versions 2.53.0 to 2.58.1. For more information, see the Apache Beam GitHub issue #32169 .

Deprecated on April 8, 2025.

Known issue: Python pipelines that use this SDK might experience occasional stuckness. For more information, see the Apache Beam GitHub issue #30867 .

Known issue: Pipelines that read data from Cloud Storage by using GcsIO, either directly or indirectly, such as with BigQueryIO, might experience data corruption on the Apache Beam Python SDK versions 2.53.0 to 2.58.1. For more information, see the Apache Beam GitHub issue #32169 .

Deprecated on March 25, 2025.

Known issue: The WriteToJson transform in cross-language pipelines fails with an error. To resolve this issue, upgrade to Apache Beam 2.55.1. For more information, see Issue #30776: Beam YAML WriteToJson fails .

Known issue: Python pipelines that use this SDK might experience occasional stuckness. For more information, see the Apache Beam GitHub issue #30867 .

Known issue: Pipelines that read data from Cloud Storage by using GcsIO, either directly or indirectly, such as with BigQueryIO, might experience data corruption on the Apache Beam Python SDK versions 2.53.0 to 2.58.1. For more information, see the Apache Beam GitHub issue #32169 .

Deprecated on February 14, 2025.

Known issue: Pipelines that read data from Cloud Storage by using GcsIO, either directly or indirectly, such as with BigQueryIO, might experience data corruption on the Apache Beam Python SDK versions 2.53.0 to 2.58.1. For more information, see the Apache Beam GitHub issue #32169 .

Deprecated on January 4, 2025.

Known issue: Pipelines that read data from Cloud Storage by using GcsIO, either directly or indirectly, such as with BigQueryIO, might experience data corruption on the Apache Beam Python SDK versions 2.53.0 to 2.58.1. For more information, see the Apache Beam GitHub issue #32169 .

Deprecated on November 17, 2024.

Deprecated on October 12, 2024.

Deprecated on August 30, 2024.

Deprecated on July 17, 2024.

Deprecated on May 31, 2024.

Deprecated on May 10, 2024.

Deprecated on March 10, 2025.

Deprecated on February 15, 2024.

Deprecated on January 13, 2024.

Deprecated on November 17, 2023.

Deprecated on October 17, 2023.

Deprecated on August 23, 2023.

Deprecated on June 27, 2023.

Deprecated on May 25, 2023.

Deprecated on April 20, 2023.

Deprecated on March 4, 2023.

• When you use WriteToBigQuery with the ignore_unknown_columns parameter set to true , the parameter value is ignored. This issue is fixed in version 2.38.0. See BEAM-14039 for details.

Deprecated on February 7, 2023.

• When you use WriteToBigQuery with the ignore_unknown_columns parameter set to true , the parameter value is ignored. This issue is fixed in version 2.38.0. See BEAM-14039 for details.

Deprecated on December 29, 2022.

Deprecated on November 11, 2022.

Deprecated on October 7, 2022.

Deprecated on August 25, 2022.

Deprecated on July 8, 2022.

Deprecated on June 10, 2022.

Deprecated on April 29, 2022.

Deprecated on February 22, 2022.

Deprecated on January 7, 2022.

Deprecated on December 11, 2021.

Deprecated on October 23, 2021.

Deprecated on September 18, 2021. Also, the Apache Beam SDK 2.24.0 version is the last to support Python 2 and Python 3.5.

Dataflow no longer supports pipelines using Python 2. Read more information on the Python 2 support on Google Cloud page.

Deprecated on July 29, 2021.

Dataflow no longer supports pipelines using Python 2. Read more information on the Python 2 support on Google Cloud page.

Deprecated on June 8, 2021.

Dataflow no longer supports pipelines using Python 2. Read more information on the Python 2 support on Google Cloud page.

Deprecated on May 27, 2021.

Dataflow no longer supports pipelines using Python 2. Read more information on the Python 2 support on Google Cloud page.

Deprecated on April 15, 2021.

Dataflow no longer supports pipelines using Python 2. Read more information on the Python 2 support on Google Cloud page.

Deprecated on February 4, 2021.

Dataflow no longer supports pipelines using Python 2. Read more information on the Python 2 support on Google Cloud page.

Deprecated on January 23, 2021.

Dataflow no longer supports pipelines using Python 2. Read more information on the Python 2 support on Google Cloud page.

Deprecated on January 6, 2021.

Dataflow no longer supports pipelines using Python 2. Read more information on the Python 2 support on Google Cloud page.

Deprecated on August 1, 2020.

• The MongoDB source added in this release has a known issue that can result in data loss. See BEAM-7866 for details.

Deprecated as of October 3, 2019.