HRI Releases

This page lists the releases with notes for the HRI with information about how to upgrade from one version to the next. If upgrading multiple versions, check the upgrade notes of all versions in between. Note that the HRI has an overall release number, which is included here. Individual components of the HRI use the overall major and minor release number as a prefix to their release number, in order to make it easy to identify which versions of components are compatible with each other. For example, v2.1-2.1.5 of the Management API is compatible with v2.1-1.0.1 of Flink validation FHIR, because they both have the v2.1 prefix.

Unless stated otherwise in the release notes of a specific version, upgrading the HRI should be achievable without downtime. If you are only upgrading to a new patch version, simply upgrade your existing deployment with the patched version. Otherwise, the new version can be deployed and configured separately in a different namespace while the old HRI version is still active, and the old HRI version can be deleted when migration is complete. In this case, be sure to use the same Elasticsearch and Event Streams instances for both of the HRI versions. Again, please see the upgrade notes for all versions between your current and target versions for any additional requirements.

v3.x

Version 3.x is the latest version focused on making the HRI a cloud-portable service. The Management API was moved from IBM Functions to a standard REST Web server and is packaged into a docker image for Kubernetes based deployments.

v3.2.4

Release notes

An HRI patch release to update Golang packages to their latest versions to address vulnerabilities.

v3.2.3

Release notes

An HRI patch release to upgrade Flink to version 1.14.3 and Log4j libraries to 2.17.1, fixing several vulnerabilities.

v3.2.0

Release notes

An HRI minor release upgrading Apache Flink to 1.14. There were also the following changes:

Management API

• Upgraded to Golang 1.17
• Upgraded the New Relic Echo agent library

Validation(Flink)

• Upgraded to Flink 1.14

Upgrading

Since the Flink validation projects were upgraded to a new version of Flink, the Flink cluster running the job will also have to be upgraded. Savepoints from the prior version of the validation jobs can be used with the newer version. See the Flink upgrade documentation for more information.

v3.1.0

Release notes

An HRI minor release using public GitHub Alvearie resources. There were also the following changes:

Management API

• Switched Kafka client library to confluent-kafka-go in order to support other Kafka authentication methods. This also caused the configuration options to switch to a list of Kafka connection properties.
• Upgraded several dependencies

Validation(Flink)

• Upgraded to Flink 1.10.3

Upgrading

This release does not contain breaking changes, but there are some changes to the Management API configuration properties. -kafka-properties takes a list of all connection properties. See https://github.com/edenhill/librdkafka/blob/master/CONFIGURATION.md for the full list.

v3.0.0

Release notes

Initial open source release of v3.x. This major HRI release ports the Management API from IBM Cloud Functions to a standard REST Web server and is packaged into a docker image for Kubernetes based deployments. This release supports the goal of making the HRI a cloud-portable service.

Upgrading

This release does not contain breaking changes to the API specification, but there are steps required to migrate the Management API to a Kubernetes based deployment. The same IBM Cloud services (Elasticsearch, Event Streams, AppID, etc.) can be shared between a v2.x deployment and a new v3.x deployment. Below are the migration steps:

1. Deploy v3.0.0 to Kubernetes.
   1. The same namespace of an older deployment can be used.
   2. If using validation, also update the Flink and Zookeeper deployments. They are backwards compatible with v2.x.
   3. Use the same OIDC settings, so that existing configurations are used in the new Management API Kubernetes deployment.
2. If using validation, migrate existing Flink jobs to the latest validation job jars.
   1. Stop existing validation jobs with a savepoint.
   2. Restart each job from the savepoint using the latest validation job jar and change the Management API URL to the new Kubernetes deployment. All other parameters should remain the same.
3. Have Data Integrators and Consumers migrate from the old Management API (URL) to the new Kubernetes deployment endpoints.
4. When all clients have migrated, delete the IBM Functions namespace that held the old Management API.

v2.x

Version 2.x uses IBM Functions to deploy the Management API and includes validation processing. It is deprecated as of March 2022. Please upgrade to the latest version at your earliest convenience.

v2.2.1

Release notes

An HRI patch release to update Golang packages to their latest versions to address vulnerabilities.

v2.2.0

Release notes

Validation processing was removed due to Log4j vulnerabilities in the older Flink version. To use validation update to the latest HRI version. This version is scheduled to be deprecated on March 01, 2022.

v2.1.6

Release notes

First Alvearie release using new GitHub Actions workflow. Upgraded to Flink 1.10.3, other minor dependency updates.

v2.1.5

Release notes

Initial open source release of v2.x.

Upgrading

This release does not contain breaking changes. Data Integrators and Consumers will maintain access to old batches. However, please note that there are changes to the Elasticsearch mapping template, which must be applied to all existing indices (one for every tenant). Below is a series of upgrade tasks.

1. The Elasticsearch index mappings need to be updated to account for the new expectedRecordCount, actualRecordCount, invalidThreshold, and invalidRecordCount fields. recordCount is considered deprecated as of v2.0.0 and expectedRecordCount should be used instead.
2. Download the new index template from GitHub
3. Follow these instructions to upgrade all the existing Elasticsearch indices with the batch.json index template downloaded in the prior step.

The expectedRecordCount, actualRecordCount, invalidThreshold, and invalidRecordCount fields have also been added to the Batch API and Notification Message model, so Data Integrators and Consumers may need to update their integration tools to prevent parsing errors. Additional Batch Notification messages and status values have also been added. For more information, refer to the Batch Status Transitions documentation.

v1.x

Version 1.x uses IBM Functions to deploy the Management API and does not include validation processing. It is deprecated as of March 2022. Please upgrade to the latest version at your earliest convenience.

v1.2.7

Release notes

An HRI patch release to update Golang packages to their latest versions to address vulnerabilities.

v1.2.6

Release notes

First Alvearie release using new GitHub Actions workflow. There were no substantive changes to the API or deployed code.

v1.2.5

Release notes

Initial open source release of v1.x.