Simpler Navigation for Servers and Operating Systems
Completed: a much simpler Servers and Operating Systems section of the Community. We combined many of the older boards, so you won't have to click through so many levels to get at the information you need. Check the consolidated boards here as many sub-forums are now single boards.
Operating System - Tru64 Unix
cancel
Showing results for 
Search instead for 
Did you mean: 

High values in the output of sar -q commnad(Run Queue).

Ketan Korgaonkar
Occasional Visitor

High values in the output of sar -q commnad(Run Queue).

Hi,

We have Oracle 8.1.7 running on a 2-node Cluster. OS version is Unix Tru64 V5.1B.

When I issue 'sar -q -f sar_dat_190404.dat' command the values for runq-sz and %runocc are very high. e.g. the highest values for runq-sz is 1002.8 and that for %runocc is 97%.

The sampling rate for sar is 96 samples separated by 900 seconds per day.

Is this behaviour fine? Figures like 1002.8 seems to be very high.

Following is the extract of sar -q and sar -u command.

sar -q -f sar_dat_190404.dat

00:00:02 runq-sz %runocc swpq-sz %swpocc
00:15:02 60.0 100
00:30:02 31.6 44
00:45:02 76.7 27
01:00:03 20.5 9
01:15:03 592.8 80
01:30:03 1002.8 97
01:45:03 163.0 38
02:00:03 11.4 6
02:15:03 8.6 6
02:30:03 7.4 5
02:45:03 6.1 4
03:00:03 5.5 4
03:15:03 8.7 6
03:30:03 7.3 5
03:45:03 9.2 6
04:00:03 14.4 7
04:15:03 35.3 20
04:30:03 7.0 5
04:45:03 5.6 4


sar -u -f sar_dat_190404.dat
00:00:02 %usr %sys %wio %idle
00:15:02 34 47 4 15
00:30:02 10 15 39 37
00:45:02 22 15 31 32
01:00:03 9 9 42 40
01:15:03 57 29 0 14
01:30:03 67 31 0 2
01:45:03 35 16 0 49
02:00:03 7 5 0 87
02:15:03 6 5 0 89
02:30:03 5 4 0 90
02:45:03 6 4 0 90
03:00:03 5 4 0 91
03:15:03 5 6 1 87
03:30:03 4 5 0 91
03:45:03 7 5 0 88
04:00:03 8 5 0 88
04:15:03 23 7 0 69
04:30:03 7 4 0 89
04:45:03 5 4 0 91
1 REPLY

Re: High values in the output of sar -q commnad(Run Queue).

Hi Ketan,

It is too difficult to predict completely based on your inputs. However, it is a good idea to take a closer look at the way the oracle is tuned by looking at its statspack.

Also, it would be a good idea to run the following command and view its output in a browser.

sys_check -all > host.html

The above output will give you better indication. One thing about the stats you provided is that your system is coping up with the load and if you think the queue is temporary during a peak time, then take more samples during that time and analyse. Check for any swap space problems or excessive swapping / paging during peak work-loads, if you find the CPU shows idle with still a huge process queue.

Cheers
Baalki