jukoff@imram.net.com (Steve Jukoff) (03/20/91)
SYSTEM ACTIVITY REPORT suggests performance degradation
'sar -u' reports:
%idle drops to 0% and %wio increases to near 100%.
Does anyone know what causes this condition and what can be done
to stop it?
System is "normal" for days before entering this state
(and after a re-boot).
Here is a day where it goes from "normal" to 0%idle.
It occurred around 14:40.
The only significant activity is a number of similar "Oracle"
processes. All other 'sar' activity looks normal.
Look at 14:40. Can you shed some light on this?
00:00:01 %usr %sys %sys %wio %idle
local remote
01:00:01 2 4 0 3 90
02:00:00 0 2 0 0 98
03:00:00 0 2 0 0 98
04:00:00 0 2 0 0 98
05:00:00 0 2 0 0 98
06:00:00 0 2 0 0 98
07:00:00 0 2 0 0 98
08:00:01 0 2 0 1 96
08:20:01 2 2 0 5 91
08:40:02 4 2 0 9 84
09:00:02 7 3 0 9 81
09:20:03 5 4 0 16 74
09:40:01 6 4 0 18 71
10:00:03 3 3 0 9 85
10:20:03 4 3 0 17 75
10:40:14 6 4 0 23 66
11:00:05 8 5 0 30 57
11:20:05 7 5 0 26 61
11:40:03 8 6 0 23 63
12:00:08 12 7 0 29 52
12:20:02 6 4 0 24 65
12:40:07 5 4 0 21 70
13:00:06 8 6 0 24 62
13:20:08 12 8 0 19 60
13:40:06 11 8 0 31 49
14:00:05 12 7 0 27 55
14:20:03 11 5 0 26 58
14:40:02 8 5 0 54 33
15:00:02 5 4 0 90 0
15:20:02 9 6 0 85 0
15:40:02 12 10 0 78 0
16:00:05 13 8 0 79 0
16:20:06 9 6 0 85 0
16:40:02 18 17 0 65 0
17:00:02 8 7 0 85 0
17:20:01 4 4 0 92 0
17:40:01 12 28 0 60 0
18:00:01 3 5 0 91 0
19:00:01 1 2 0 96 0
20:00:01 4 4 0 92 0
21:00:02 4 4 0 92 0
22:00:00 2 2 0 96 0
23:00:01 5 5 0 90 0
Average 4 4 0 39 53
root@lazarus:/usr/adm/sa 37% vmstat -s
31640 free swap space
95344 demand zero and demand fill pages
13504 pages on swap
69307 pages in cache
66688 pages on file
240094 protection fault
26651 pages are freed
69 success in swapping out a process
0 fail in swapping in a process
59 success in swapping in a process
206 swapping out a region
177 swapping in a region
5333408 cpu context switches
7633624 system calls
0.2u 1.6s 0:06 29%
root@lazarus:/usr/adm/sa 38% vmstat 10 5
PROCS PAGING SYSTEM CPU
r b w frs dmd sw cch fil pft frp pos pif pis rso rsi sy cs us su id
1 116 0 3164 0 0 0 0 1 0 0 0 0 0 0 119 115 44 18 38
1 116 0 3164 6 0 0 0 0 0 0 0 0 0 0 13 13 28 71 1
1 116 0 3164 0 0 0 0 0 0 0 0 0 0 0 77 70 83 16 1
1 116 0 3164 0 0 0 0 0 0 0 0 0 0 0 21 16 17 82 1
2 115 0 3164 0 0 0 0 0 0 0 0 0 0 0 6 5 5 94 1
0.2u 2.0s 0:52 4%
+----------------------------------------------------------+
| Steven Jukoff DDN: jukoff@net.com |
| Phone: 415-780-5819 uucp: pyramid!unet!jukoff |
+----------------------------------------------------------+les@chinet.chi.il.us (Leslie Mikesell) (03/21/91)
In article <1319@unet.UUCP> jukoff@imram.net.com (Steve Jukoff) writes: >'sar -u' reports: >%idle drops to 0% and %wio increases to near 100%. Do you have anyone running VP/ix? It seems to be normal for it to consume all of the system idle time even when DOS is just waiting for a command (polling the keyboard, I suppose). >Does anyone know what causes this condition and what can be done >to stop it? Try a "ps -ef" and look for processes that have used a lot of time. You may have a program with a bug like missing EOF and continuing to make read() system calls forever. Les Mikesell les@chinet.chi.il.us