On Jan 18, 2008 7:46 AM, Rafael Pires <[EMAIL PROTECTED]> wrote: > Hi, > I've been requested to investigate a performance issue with a sybase instance > bound to a processor set of 3 processors (server contains 16 in total). > After some investigation, I could not find any bottleneck with prstat, vmstat > or sar output reports. However, mpstat reports from time to time a very busy > cpu in the processor set and I am unsure if this is expected or would be a > bottleneck. > Could someone pls help me? The mpstat output is listed below. > Thanks in advance, > Rafael > > [EMAIL PROTECTED]> mpstat -P 2 1 > CPU minf mjf xcal intr ithr csw icsw migr smtx srw syscl usr sys wt idl > 3 0 0 25 11 1 0 16 0 23 0 150742 85 15 0 0 > 4 0 6 8 28 1 2291 22 139 13 97 4655 2 7 0 91 > 5 0 11 7 35 1 2360 28 137 24 107 14475 7 9 0 84 > CPU minf mjf xcal intr ithr csw icsw migr smtx srw syscl usr sys wt idl > 3 0 0 31 12 2 0 15 0 19 0 177660 84 16 0 0 > 4 0 10 13 44 2 2376 43 163 26 133 15782 6 9 0 85 > 5 0 14 13 40 2 2570 38 181 11 135 3647 3 8 0 89 > CPU minf mjf xcal intr ithr csw icsw migr smtx srw syscl usr sys wt idl > 3 0 13 6 36 1 2218 23 191 14 91 18094 8 10 0 83 > 4 0 10 7 26 1 2210 25 132 10 95 17741 12 9 0 80 > 5 0 3 23 9 1 223 14 32 11 12 172230 73 17 0 10 > CPU minf mjf xcal intr ithr csw icsw migr smtx srw syscl usr sys wt idl > 3 0 10 15 25 2 1226 27 86 8 54 81625 44 11 0 45 > 4 0 2 29 22 1 1015 24 133 9 52 102544 45 13 0 42 > 5 0 11 6 41 1 2259 39 188 15 105 8197 5 8 0 87 > CPU minf mjf xcal intr ithr csw icsw migr smtx srw syscl usr sys wt idl > 3 0 0 28 10 1 0 15 0 25 0 181293 82 18 0 0 > 4 0 14 8 43 2 2598 38 163 29 127 9702 5 9 0 86 > 5 0 6 7 32 2 2302 29 152 22 119 12660 5 9 0 86
You might want to fire up DTrace to see what is causing the spike in system calls when the CPU is pegged at 100%. - Ryan -- UNIX Administrator http://prefetch.net _______________________________________________ perf-discuss mailing list perf-discuss@opensolaris.org
