17 April 2023 - v3.36.10

You can download version 3.36.10 here.

Upgrade on GKE:

kubectl apply -f "https://www.hydrolix.io/operator/v3.36.10/operator-resources?namespace=${HDX_KUBERNETES_NAMESPACE}&gcp-storage-sa=${GCP_STORAGE_SA}"

Upgrade on EKS:

kubectl apply -f "https://www.hydrolix.io/operator/v3.36.10/operator-resources?namespace=${HDX_KUBERNETES_NAMESPACE}&aws-storage-role=${AWS_STORAGE_ROLE}"


  • Added support for ARN (Amazon Resource Name) roles for Kinesis stream ingestion.
  • Introduced the ability to ingest Cloudwatch logs as individual messages (instead of batches of messages) with the Cloudwatch subtype.
  • Added datastore_project_id configuration field to assist ingestion from Kinesis streams with a [checkpointer hosted in GCP](https://docs.hydrolix.io/docs/ingest-kinesis#specify-an-alternative-checkpoint-datastore.
  • Added support for summary tables for Kinesis data sources via the settings.summary.sql Kinesis configuration.
  • Renamed "credentials" configuration to "access_details".
  • Added configuration options for the names of SIEM credential environment variables.
  • Added configuration to limit internal event queue write size with message_queue_max_rows.
  • Added configuration for concurrency in internal queue reads; first, the KAFKA_READ_CONCURRENCY environment variable; second, the CONTAINER_CPU_LIMIT environment variable; finally, the detected CPU count.
  • Increased timeout for Kinesis stream ingestion from 100 seconds to 200 seconds.
  • Updated the layout of the System Health UI.
  • Added HYDROLIX_LICENSE environment variable, which contains a copy of the Hydrolix license accessible from all pods of your Hydrolix cluster.
  • Added support for token authentication in TCP requests.
  • Increased Linode load balancer throttle floor.
  • Added DISABLE_DISK_CACHE tunable, which lets query peers immediately delete partition metadata from disk after use.

Notable Bug Fixes

  • Added redirect when System Health returns a query.
  • Disabled "retrieve logs" button when System Health fetches a query.
  • Fixed incorrect endtime selection flow in System Health.
  • Added clearing indicators to dropdowns in System Health.
  • Added a tooltip when the System Health "copy SQL" button is clicked.
  • Downgraded log level from "critical" to "warn" for message indicating LRU cache "redzone reached".
  • Upgraded Turbine Authorizor log level from "trace" to "info".
  • Fixed the operator setting the storage class to "gp3" when running in Linode but using an AWS S3 bucket.
  • Fixed ingestion peer caching partition metadata to disk unnecessarily.