We’re pleased to announce the release of Gridware Cluster Scheduler 9.0.9 (based on Open Cluster Scheduler 9.0.9), bringing powerful new capabilities for license management, job debugging, and resource efficiency to HPC environments.
FlexLM Integration: Zero-Configuration License Tracking
The headline feature in 9.0.9 is native FlexLM integration that fundamentally simplifies license management in cluster environments. The scheduler now automatically detects, configures, and tracks FlexLM licenses without any manual configuration.
Setup couldn’t be simpler—just point the scheduler at your FlexLM server, and it handles the rest. No complex manual license configuration in GCS is needed. The scheduler automatically discovers available licenses, tracks their usage, and ensures jobs requesting licensed software only run when licenses are available. This tight integration prevents the all-too-common scenario of jobs failing mid-execution because licenses became unavailable without the manual cluster configuration overhead required before.
Faulty Job Load Sensor: Automated Debugging Support
Debugging failed jobs just got significantly easier with the new faulty job load sensor. When jobs fail, the scheduler automatically copies all relevant trace files and job environment information to a configurable location.
No more hunting through log directories or reconstructing job environments after the fact. Everything you need for post-mortem analysis is automatically preserved and organized, dramatically reducing the time from job failure to root cause identification.
Dynamic Resource Release for Running Jobs
Version 9.0.9 introduces the ability to drop costly resources—like license requests—from already running jobs. This capability addresses a common inefficiency in HPC environments: jobs that request resources for their entire runtime but only actually use them for a portion of execution.
With dynamic resource release, running jobs can voluntarily give up resources they no longer need, making those resources immediately available to other waiting jobs. For expensive resources like commercial software licenses, this translates directly to reduced per-job costs and significantly improved resource utilization across the cluster.
Enhanced Stability and Scheduling Improvements
Beyond the major features, 9.0.9 includes important stability fixes and scheduling enhancements:
Large Group Support: Critical bug fixes ensure stable operation in environments where users belong to thousands of UNIX groups—a common scenario in large enterprise and academic deployments.
Job Simulation Improvements: Enhanced job simulation capabilities provide more accurate predictions of how jobs will be scheduled, helping administrators optimize queue configurations and users understand expected wait times.
Parallel Job Scheduling: Fixes for parallel job scheduling logic, particularly when ignoring worker task requests on master hosts, ensure more reliable distributed job execution.
Download and Upgrade
Gridware Cluster Scheduler 9.0.9 and Open Cluster Scheduler 9.0.9 are available for download today at:
https://hpc-gridware.com/download-ocs-9-0-9
For existing 9.0.x deployments, upgrading to 9.0.9 follows the standard binary replacement approach—stop services, replace binaries, restart services. No configuration changes are required for basic upgrades, though you’ll want to configure the new FlexLM integration and faulty job load sensor to take advantage of these capabilities.
The combination of intelligent license management, automated debugging support, and dynamic resource optimization makes 9.0.9 particularly valuable for environments running commercial applications with expensive licensing models. These features work together to reduce both operational costs and administrative overhead while improving cluster utilization.
