Version 4.1.0

Release Date: 24th March, 2026

New Features

New Service Observability

• JupyterHub Observability: This release introduces real-time visualization of JupyterHub metrics in Pulse, enabling monitoring of user activity, resource usage, and performance through interactive dashboards and charts. For more details, see JupyterHub.
• Observability for Airflow on K8s:
  • This release provides observability support for Airflow deployed in the Kubernetes cluster. For details about the configuration, see Configure Airflow and Pulse to Monitor Standalone Airflow on Kubernetes.
  • A complete revamp of the Airflow documentation as shown in the Pulse UI. For details, see Airflow.

Other Features

• Pulse Sense (Beta): This release introduces Pulse Sense, an agentic capability within Pulse designed to simplify troubleshooting and operational analysis. Pulse Sense helps users quickly diagnose issues such as slow-running applications or unhealthy components without the need to manually navigate across multiple dashboards. For more information, see Pulse Sense (Beta).

• Kubernetes Cluster Optimization (Beta):

  • This release introduces Kubernetes Cluster Optimization in Pulse, enabling safe reuse of unused CPU and memory requests without changing how users define resources.
  • Pulse detects the gap between requested and actual usage, exposes reclaimable capacity as pulse-cpu and pulse-memory, and improves workload packing and utilization.
  • The update also adds Optimization and Savings Potential views to help teams increase efficiency and reduce cluster costs. For details, see Optimize K8s Cluster Using Kubernetes Optimizer (Beta).

• WCAG 2.1 AA Accessibility: Pulse UI now includes WCAG 2.1 AA accessibility improvements, with better color contrast, visible focus states, full keyboard navigation, correct ARIA support, accessible forms, and screen reader compatibility for all content. These updates improve usability for all users and support enterprise accessibility and VPAT compliance. For details, see WCAG 2.1 AA Accessibility Compliance.

• Gluten Velox Metrics in Pulse: When Spark jobs are executed with Gluten enabled and the Velox library in use, the associated metrics are displayed in the Pulse UI. Pulse also provides DAG-level visibility for both Velox and Spark operators, enabling you to analyze execution flow and performance at the operator level. For details, see Gluten SQL/DataFrame.

• Pulse Alerts Notification on Netcool: Pulse now supports alert notifications on Netcool, enabling you to forward alerts directly to your Netcool environment for centralized event management and faster incident response. For more information, see:

  • Notifications > check the Netcool row from the table.
  • Netcool Notifications

• Disaster Recovery Plugins for Hadoop, Hive, and HBase:

  • Introduced three Disaster Recovery action plugins: Hadoop DR Cluster Sync, Hive Cluster Replication, and HBase Cluster Replication.
  • These enable cross-cluster replication for HDFS, Hive, and HBase using snapshots, DistCp, and REPL workflows, with support for incremental sync, automated scheduling, and configurable execution parameters.
    • For details, see Set Configuration for Action Runbooks > Hadoop DR Cluster Sync, Hive Cluster Replication, and HBASE Cluster Replication.

New Alerts

• NS and DS Quota Usage Alerting: Pulse now provides a predefined alert, HDFS_NS_DS_QUOTA_ USAGE, to monitor Namespace (NS) and Disk Space (DS) quota usage. The alert is triggered when the configured usage thresholds are exceeded. For more information, see:

  • Configure DS and NS Quota Paths for Creating an Alert
  • HDFS Alerts, check the row HDFS_NS_DS_QUOTA_USAGE.

• Alert for HDFS Small Files: Pulse now enables you to create an alert for detecting high small-file counts in HDFS directories using FS Analytics. Users can configure threshold-based alerts grouped by path to proactively identify directories with excessive small files and prevent HDFS performance issues.

  • For details on configuring the HDFS small files threshold, see Configure HDFS Small File Threshold.
  • For details about creating alerts when small files breach the threshold defined, see Create Alerts for Small Files.

• Coordinator-Specific Impala Alert Metric: You can create custom alerts for tracking in-flight queries on the Impala coordinator by selecting the Impala.Impala_server_backend_num_queries_executing metric from the Metrics drop-down on the Alert Creation page.

• Kafka Custom Alerts: This release enables you to create custom alerts to detect partitions under minimum ISR, under-replicated partitions, and low broker count conditions that can lead to producer failures or data availability risks. For configuration details, see Kafka Custom Alerts.

Enhancements

Pulse on Kubernetes Architecture and Observability Improvements

• This release introduces architectural improvements to the Pulse operator deployment model in Kubernetes. The Custom Resource (CR) now maintains the complete desired state by appending add-on configurations instead of replacing core service configurations. This ensures the CR remains the single source of truth and improves overall resiliency.
• In addition, Kubernetes metrics collection no longer requires cluster-admin (cluster-role) access. Pulse now supports namespace-scoped RBAC for metrics retrieval. When cluster-level permissions are not granted, only namespace-level metrics are collected. For details, see Configure Kubernetes Metrics Collection Using the PulseNode Agent.

Deployment

• Pulse Server Monitoring on Kubernetes: Pulse now supports enhanced monitoring of Kubernetes deployments through the Admin UI. Administrators can view real-time pod health, container resource usage, and live service logs for both Core and Add-on services, improving operational visibility and troubleshooting efficiency. For more information, see Monitor Pulse Server on Kubernetes.

• PulseNode Agent Deployment on Kubernetes: This release adds support for deploying and removing the PulseNode Agent on Kubernetes using RBAC configuration and Manager API endpoints. It enables streamlined deployment, cluster registration, and clean removal of PulseNode in containerized environments. For details, see Deploy PulseNode Agent on Kubernetes.

• Introduced healthcheck commands for NATs Object Store: This release introduces new healthcheck commands for NATs Object Store to manage, update, and watch events for NATS object stores associated with Pulse. For details, see Object Store Commands.

• Pulse Home-Support for Custom Installation Directories: Pulse now lets you set a custom Pulse Home directory instead of using the default /opt/pulse path. This flexibility lets your agents in directories outside /opt/pulse. For more information, see Install Agents in a Custom Directory.

• Updated: NATS Subject Configuration for ClickHouse > Pulse Streaming: In this release, the nats_subjects value has been updated from separate subjects — clickhouse.query_log and clickhouse.spans — to a unified, cluster-aware subject: clickhouse_events_ <cluster_name>.0

  • The queries and spans now stream through a single NATS subject.
  • Action Required: You must update the nats_subjects configuration to ensure continued event ingestion in Pulse. For details, see Create Materialized Views for NATS Integration.

• New Hook Version Introduced: This release introduces a new Hive hook version: ad-hive-hook_odp_3.1.4.3.2.3.5-3.

  • If you are using ODP version 3.2.3.5-3, you must use this hook version.
  • For details, see Supported Hook Versions and Place Pulse hook JARs and Set Properties (Hive and Tez).

• Deprecated ad-director (old actions): This release deprecates the ad-director service (old actions). You can now migrate to New Actions (Manage and Configure Actions) by upgrading to Pulse 4.1.0. For details, see Upgrade to Version 4.1.0.

Observability

HDFS

• Real-Time Sync Enhancement for HDFS File Explorer: This release enhances the HDFS File Explorer in Pulse to support real-time synchronization of HDFS metadata. Previously, HDFS insights were generated once every 24 hours. With this enhancement, administrators can monitor HDFS files whenever needed (near real-time), enabling improved visibility into active workloads and faster operational decision-making. For details, see below:
  • Update Real-time Monitoring Parameters of HDFS
  • Real-Time Data Support Without Compromising System Stability

Dashplots

• New Charts in Databases > File Analytics: The Hive Table Temperature chart in the File Analytics tab has been moved to the new Table Insights tab. This change has been made due to migration from Splashboard to Dashplot, which imposed certain limitations.

  • The Table Insights tab now includes the following new charts:

    • Hive Files by Database
    • Hive Table File Count Distribution

  • For more details, please refer to: Table Instights.

• New Visualization Library: Pulse now introduces a centralized Visualization Library, allowing visualizations to be created once and reused across multiple dashboards.

  • Visualizations can now be shared across dashboards without duplication.
  • Editing a shared visualization updates it everywhere it is used.
  • Common filters used by multiple charts are automatically promoted to dashboard-level filters.
  • Existing dashboards are automatically migrated only if you are upgrading from 4.0.x — no user action required. If you are on earlier versions, upgrade to 4.0.x, and then migrate to 4.1.0.
  • Export/import and scheduled reports continue to work seamlessly.
  • For details about adding visualizations to a dashboard, see Add Visualizations to a Dashboard.

Impala

• Enhanced Impala Query Observability: Improved the stability and throughput of the Impala connector under high-load conditions.

  • Improved the reliability of Impala query tracking under high-load conditions.
  • Ensured that queries from both Hue and the Impala shell are consistently captured.
  • Resolved issues that caused missing or inconsistent query states in the Pulse UI.
  • Enhanced overall stability to prevent query data delays or loss.
  • For details about the Agent and Connector, see Configure Impala Connector and Configure Impala Agent.

• Enhancements to Impala Canary Action: The following improvements have been introduced in this release:

  • Removed the “krb5.conf File Path” field from the Perform Impala Canary Action configuration and internally defaulted to Kerberos action using /etc/krb5.conf.
  • Added support for secure and coordinator-specific connectivity through the new fields: Impala Coordinator Host, Enable Impala SSL, and Impala CA Certificate Path. For details, see Perform Impala Canary Operations.
  • Improved logging, error messaging, and observability for Pulse Canary actions, including:
    • Enhanced timeout handling with better visibility into where and why a failure occurred

• Enhanced the Impala Kill Query Action: The following improvements have been introduced in this release:

  • The reason for termination (for example, query duration or read threshold breach) is now displayed under Actions > Executions.

  • To view details:

    • Click an action name to open the execution logs
    • On the action logs page, click Show Output

  • Additional visibility has been added for Impala queries (Impala, Tez, and YARN applications):

    • The user who killed the query
    • The query ID associated with the terminated query
    • Resource Pool in case of Impala

  • Additional visibility has been added for terminated queries (Impala, Tez, and YARN applications):

    • On the Actions > Executions page, select View Killed Queries
    • You are redirected to the YARN Application Explorer to view the details, such as query ID, user, etc.

• Impala Dashboard XLS Download Support: All dashboards on the Pulse Impala UI page can now be downloaded in .xlsx format.

• Flexible Column Widths: Added flexible columns on some of the dashboards, such as Impala, Spark, Tez, and MR, etc. query dashboards, to improve readability of long hostnames and coordinator details.

Other Updates

• Downloading Records:

  • The option to download records for components such as Tez queries, YARN applications, Spark Jobs, Impala queries, etc., from the Pulse UI in .xlsx format has been removed due to high memory consumption, which could impact the Pulse UI service performance.
  • You can continue to download the query results in .csv format, which is lighter and more efficient.
  • To enhance the stability during large export jobs, Pulse introduced a few system limitations for download job retention, job concurrency, lock lifetime, and compression level. These settings help reduce memory and CPU pressure during large exports and improve recovery for stuck jobs. For details, see Download Records.

• Download up to 10,000 Records: This release increases the maximum number of records that you can download from 100 to 10,000 on the HBase tables UI on Pulse. Previously, this higher limit was available only for Tez, Impala, Spark, YARN, etc. dashboards

• The Spark Jobs Page Updates: The following new columns have been added to the Spark Jobs page in the Pulse UI (for details, see Spark Jobs):

  • Total Jobs and Failed Jobs
  • Maximum Driver Memory and Maximum Executor Memory

• Tez Queries-Database Name Filter Added: The Database Name column has been added to the Tez Queries page, enabling you to filter and analyze Hive queries based on specific database names.

Resolved Issues

• Fixed an issue where the Hydra instance was being installed on all nodes during parcel-based CDP deployments, even when agents were already present. Pulse now verifies installation by checking for the /opt/pulse directory before proceeding, preventing unnecessary reinstallation.
• Fixed an issue where Pinot alert metrics were not loading during alert creation and required multiple alert container restarts to become available.
• Fixed an issue where the Pulse HBase UI displayed incorrect region-level write metrics, showing 0 or 1 values even though table-level write metrics were accurate.
• Fixed an issue where some Impala queries were not shown in the Pulse UI despite being present in Impala profile logs.
• Fixed an issue in the Pulse HBase Dashplots UI where region-level write metrics were incorrectly displayed as 0 or 1, while table-level metrics appeared accurate. Region-level write metrics are now correctly reported, ensuring consistency and accurate visibility across both table and region views.
• Fixed an issue where MapReduce metrics were not displayed in the Pulse YARN Application Explorer UI page after job completion.
• Fixed a file descriptor leak in the ad-director service that caused the number of open files to grow over time.