IBM Spectrum LSF performance tuning

Tune the performance and scalability of your LSF cluster.

Tune your IBM Spectrum LSF cluster
Tune LIM policies and parameters to improve performance. Use badmin mbdrestart -p to enable parallel restart to Improve mbatchd response time after mbatchd restart. On UNIX platforms that support thread programming, change default mbatchd behavior to use multithreading and increase performance of bjobs query requests. Configure Hard CPU affinity to specify the management host CPUs on which mbatchd child query processes can run. Use the JOB_SWITCH2_EVENT parameter in lsb.params to log mbatchd performance metrics. Enable DISPATCH_BY_QUEUE to increase queue responsiveness.
Achieve performance and scalability
Tune LSF for large clusters and monitor performance metrics in real time. Optimize performance in large sites by tuning queries, scheduling, and event logging.

Parent topic: Administer IBM Spectrum LSF