From: Yard, John (jyard@AIS.UCLA.EDU)
Date: Wed Nov 19 2003 - 21:35:48 EST
I have an H80 OLTP database system (Sybase)
which runs as basically 3 subtsks. Even during a
period in which the cpu utilization is ~50% for 4 cpus,
I recently see long periods where the run queue utilization is
a solid 100%:
02:00:00 runq-sz %runocc swpq-sz %swpocc
11:05:00 2.1 100 1.0 100
11:10:00 1.9 100 1.0 100
11:15:00 2.1 100 1.0 100
11:20:00 2.1 100 1.0 100
11:25:00 2.1 100 1.0 100
11:30:00 2.1 100 1.0 100
11:35:00 2.2 100 1.0 100
11:40:00 2.1 100 1.0 100
11:45:00 2.0 100 1.0 100
11:50:00 2.3 100 1.0 100
11:55:00 2.1 100 1.0 100
12:00:00 2.2 100 1.0 100
12:05:00 2.3 100 1.0 100
12:10:00 1.9 100 1.0 100
12:15:00 2.0 100 1.0 100
cpu :
02:00:00 %usr %sys %wio %idle
11:00:00 26 25 0 48
11:05:00 25 28 0 47
11:10:00 22 27 0 51
11:15:00 27 27 0 46
11:20:00 26 25 0 49
11:25:00 26 28 0 46
11:30:00 27 28 0 45
11:35:00 27 29 0 44
11:40:00 22 31 0 47
11:45:00 27 22 0 51
11:50:00 27 29 0 43
11:55:00 27 22 0 51
12:00:00 27 25 0 48
12:05:00 32 24 0 44
12:10:00 27 20 0 52
12:15:00 12 32 0 55
What could cause such a solid 100%
run queue occupied ?
JYard
UCLA
This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 22:17:22 EDT