vmstat prob

[mohmin]

my machine running WebSphere MQ V5.3 is
reporting high vmstat values of kthr/r when cpu/id is low.

 

[kHz]

kthr/r is the run queue - the number of processes running or waiting to run. The value can be (NUM_CPU_PROCS * 4). What is high for you and how many cpu's do you have? If you have a process that is cpu-bound it will drive up the run queue because the process is always running or waiting.

 

[mohmin]

here is peace of it to look at.
r b
611 0 307613 651043 0 0 0 0 0 0 0 0 0
id wa
27 55 18 0
19:43:36
r b
349 0 307367 650692 0 0 0 0 0 0 0 0 0
id wa
45 18 37 0
01:43:39
179 0 307685 650124 0 0 0 0 0 0 1147 177656 18809 44 12 40
5 04:05:23
132 0 307863 650653 0 0 0 0 0 0 1300 221779 21898 2 9 90
0 21:00:53
128 0 307422 650222 0 0 0 0 0 0 1297 199368 21627 1 10 89
0 05:37:43

 

[RMGBELGIUM]

if you have about 30 procs or so, 611 could be a normal value?As long as the "b" value isn't constantly non-zero you don't have a cpu issue."b" is the number of threads that have to wait on a cpu to get some cputime.

You have a runqueue of 611 , but all threads are being served.It still are enormously hight values.Perhaps an AIX 5.3 bug?

regards,

R.