Redis Data Integration release notes 2.2.1 (November 2026)
Fixes for Snowflake credentials, pipeline stop status, and named pipeline ingress.
What's New in 2.2.1
Breaking Changes
- New pipeline names have a maximum of 20 characters: The limit is reduced from 22 characters. Update provisioning scripts that create pipelines with longer names. Updating an existing pipeline does not revalidate its name.
- Resource names change for pipelines other than
default: Any scripts, monitoring, and custom ingress rules that refer to collector Kubernetes resources by name need to be updated. The table shows examples for theorderspipeline. Resource names for thedefaultpipeline stay the same.
| Collector | Previous resource name | New resource name |
|---|---|---|
Debezium (source mysql) |
orders--collector-mysql |
orders--col-mysql |
Flink JobManager (source spanner) |
orders--flink-col-spanner-jm |
orders--fcol-spanner-jm |
Flink TaskManager (source spanner) |
orders--flink-col-spanner-tm |
orders--fcol-spanner-tm |
RIOTX (source snowflake) |
orders--riotx-collector-snowflake |
orders--rcol-snowflake |
Bug Fixes
- Use the Snowflake source's referenced secret provider: With multiple Amazon Web Services (AWS) secret providers, the RIOTX collector now reads credentials and private keys from the provider referenced by its source instead of the first one.
- Wait for stopped components to shut down: Stopping a pipeline or a source no longer reports completion while its pods are still terminating, unless passing
--wait=false. - Support ingress path rewriting for named pipelines: Ingress annotations now substitute
$SERVICEin annotation keys and$PATH_PREFIXin annotation values with the backend service name and effective path prefix. This lets an AWS Application Load Balancer strip the pipeline prefix when routing requests. The substitutions are available for the Debezium collector, Flink collector, Flink processor, and classic processor metrics exporter.