[EMAIL PROTECTED] said: > 1) So, my question is why is cpu% so high in cpu#6 only? Though the problem > is solved with jumbo frames for 2 machines, if we increase number of nodes > this scalability problem will be seen with 3,4,5 ...machines (since cpu > utilization is very high with current state).
I think you don't have all the necessary tunables set. Here's what works on our T2000, Solaris-10U4 plus 10U5 patches: * Network throughput tuning (for 4x e1000g NIC's, 8 cores). set ddi_msix_alloc_limit=0x8 set ip:ip_soft_rings_cnt=0x8 set ip:ip_squeue_fanout=0x1 * the below assumes psrset's for interrupts are in effect set ip:ip_squeue_bind=0x0 We have not changed the settings for ip:tcp_squeue_wput nor ip:ip_squeue_wait here. I don't think you would want the ip:ip_squeue_bind set to 0, unless you use "psradm -i", as in: http://www.sun.com/bigadmin/features/articles/jboss_sf_t2000.jsp#3.3 Regards, Marion _______________________________________________ perf-discuss mailing list perf-discuss@opensolaris.org
