18 October 2023 - v3.50.9
12 months ago by Rich Vanderwal
You can download version 3.50.9 here.
Upgrade on GKE:
kubectl apply -f "https://www.hydrolix.io/operator/v3.50.9/operator-resources?namespace=${HDX_KUBERNETES_NAMESPACE}&gcp-storage-sa=${GCP_STORAGE_SA}"
Upgrade on EKS:
kubectl apply -f "https://www.hydrolix.io/operator/v3.50.9/operator-resources?namespace=${HDX_KUBERNETES_NAMESPACE}&aws-storage-role=${AWS_STORAGE_ROLE}"
General
- Service log level can be modified with the
log_level
tunable - Add
stream_concurrency_limit
tunable to adjust stream head concurrent connections - Gracefully fallback to default merge pools if configured pool cannot be found
- Distribution of secrets is limited to pods on a "need-to-know" basis
- Node type targeting now works for Crunchy Data pods
- Keycloak pods have been scaled up in
prod
andmega
profiles - Enabled brute force protection by default on Keycloak realm
- Always 404 when UUIDs in URL are incorrect
- Add primary_key to table API endpoint
- Add support for pushing metrics and integrate it into new merge
- Removed datetime64 as a type on transforms
- Check traefik events as part of operator readiness probe
- Add autovalue for Kinesis stream ARN and name
- Make hydrolix pods run as turbine user
- GOMEMLIMIT environment variable is now applied to all pods for better scaling
- Ingest pods now support the
/config-version
endpoint - Add vector pod options for disabling logging
- New project naming UI allows project name change
- UI now has the ability to delete projects
- Add timerange and limit params to audit trail UI
- Update franz-go kafka library to 1.14.4
- Improved logging for Akamai SIEM
Notable Bug Fixes
- Mitigate dropping of MESSAGE_QUEUE_MAX_ROWS value, defaults to 100
- More carefully handle writing large events to internal message queue
- Fixed merge not reporting raw_size and uncompressed_mem_size
- Merge now considers both mem_size and uncompressed_mem_size when merging summary tables
- Fixed broken CTEs and type matching when timerange filter is on
- Fixed ingest dictionary lookups broken on scale-up: wait for dictionaries and config to be loaded
- Fixed secret and env related problems affecting kinesis and siem on v3.50.x
- Fixed redpanda data races and update default write limit
- Fixed operator crash creating kinesis
- Error gracefully on compound type mismatch
- Fixed UI bug in CSV transform generation
- Fixed UI query reference tab highlighting
- Shard keys can no longer be changed after they are set
- Updates to pools no longer fail on incorrect validation
- Fixed Operator crashes when using target cpu utilization scale setting
- Handle 412 from Akamai SIEM