Platform LSF Administration Guide Version 6.2

What’s New in the Platform LSF Version 6.2
Administering Platform LSF
16
What’s New in the Platform LSF Version 6.2
◆ “Platform LSF”
❖ “Performance and scalability enhancements”
❖ “Platform LSF on Windows enhancements”
❖ “Usability”
◆ “Platform LSF License Scheduler”
❖ “License optimization”
❖ “Policy enhancement”
❖ “Usability”
◆ “Platform LSF Reports”
❖ “Addition of LSF Reports Functionality”
❖ “New and improved license optimization features”
❖ “Improved infrastructure”
❖ “New LSF Analytics cubes”
❖ “New Database ETL flows”
❖ “New Datamart ETL flows”
❖ “New LSF reports”
❖ “New scheduled tasks”
◆ “Platform LSF HPC”
❖ “Interruptible backfill”
❖ “Resource granularity”
❖ “Time-based slot reservation and greedy slot reservation”
❖ “New SGI cpuset integration”
❖ “MPI autodetection”
❖ “New MPI integrations”
❖ “pam scalability enhancements”
Platform LSF
Performance and scalability enhancements
◆ Project scheduling—Improved scheduling to eliminate performance bottlenecks
when many unique project names are used within a Platform LSF cluster.
◆ Bulk job termination—New bkill -b option for terminating a bulk set of jobs
from the system. This eases administration and decreases the time to delete a large
number of jobs.
◆ Improved job event logging—Enhanced performance of job event logging
results in improved job throughput, especially in large clusters running many small
jobs.
◆ Improved I/O performance in the job file info directory—Tune your cluster
to avoid creating a large number of job files stored in the
LSF_SHAREDIR/cluster_name/logdir/info directory by enabling LSF to
create subdirectories in the info directory. Use MAX_INFO_DIRS in
lsb.params to create the subdirectories and enable mbatchd to distribute the job
files evenly throughout the subdirectories.