v3.11.10

AWS Cloud Formation (HDXCTL) / Kubernetes (HKT) Download : hdxctl-v3.11.10 / hkt-v3.11.10

Release Locations

AWS Cloud Formation

1
wget -O hdxctl https://hdx-infrastructure.s3.amazonaws.com/hdxctl-v3.11.10 && chmod +x hdxctl

Kubernetes (GKE)

1
wget -O hkt https://hdx-infrastructure.s3.amazonaws.com/hkt/hkt-v3.11.10 && chmod +x hkt

Query

  • BugFix: Poor cold query performance due to manifest initialization
  • Enable query tracing

Core

  • BugFix: Postgres Catalog Connection Retry on broken_connection Error
  • BugFix: Non-deterministic Logging Segfault/mutex error
  • Enhancement: Indexer now uses parallel sort to improve throughput
  • BugFix: Non-Primary Indexed DateTime64 < , <= handling

Portal/API

  • Add sql_transform and sql_expression to the transform UI.
  • Add custom function and dictionaries API endpoints
  • Add project name prefix to published functions
  • BugFix: Encoding for # in the query UI
  • BugFix: Table UI stats compression ratio
  • BugFix: Sources UI when the peer-pool is empty

Ingest

  • Split out the indexer as a service for Streaming
  • Autoingest changed from single to multi SQS queues. A new field autoingest.source has been added to the table API endpoint. This is a string field, format "sqs://my-queue-name".

Infrastructure/Command & Control

  • K8s: Improved secret handling on upgrade
  • K8s: Add reconnect support for RabbitMQ JobQueues
  • Default tunable setting USE_HTTPS_WITH_S3 to true
  • Increase postgres max_connections to 1000
  • Add forwarding rules to allow prometheus to scrape indexers
  • K8s: Use premium disk for postgres data volume
  • Add a tunable setting for RDS backup retention period.
  • Replace Streaming dashboard to include new metrics from indexer
  • K8s: Update merge period to 15s and batch size to 1000
  • K8s: Fix traefik autoscale
  • K8s: Adds support for Pub/Sub(GCP) autoingest