Sponsored Content
Operating Systems AIX How to measure waiting time in run queue? Post 302803019 by GiiGii on Sunday 5th of May 2013 05:45:03 PM
Old 05-05-2013
Quote:
Originally Posted by MichaelFelt
This would depend on what you are trying to collect.

One, classic approach, might be to use sar perhaps.

A different approach would be to use AIX Advanced Accounting.

If you are looking for a deep approach, and are looking at a specific process (rather than all processes) of a know PID and/or TID you could use various trace based tools (e.g., tprof, curt, etc) and/or trace and trcrpt. Rather than PID this could also be command name - there are too many options to name them all.

And, another option could be using probevue mechanism.

Hope this helps!

Hello,

Thank you for your answers. I should have specify my working environment :

Business app : SAP R3 on Oracle Cluster-ware
DB : Oracle 11g + RAC
FS : GPFS
OS : AIX 6.1
Servers : IBM psystem (Power7 CPU)

We are investigating divergence between real CPU consumption versus 'sap + rac oracle + gpfs' benchmarks in order to challenge our advanced support from SAP and Oracle.

By the way, I am not a AIX sysadmin neither I am an AIX expert. And unfortunately, I do not have hands over the system. To do something on the system, I have to ask our contractor in charge of. However, I understand thing well ... I think Smilie

About your propositions :

First thing first, I want just graph this waiting time over the time no matter the process. I should read more carefully the SAR doc. I didn't know about the AIX advanced accouting system. Hence, I printed the doc and now I have to read it Smilie

If we find something particuliar, I may go for a deep analysis but not with a real time trace I that is possible. We ran some CURT (3 time 1 minutes) and heavily impacted our 6000 users :x

Thank you again for your time. I will be certainly back with others questions and I hope results.

Regards
 

9 More Discussions You Might Find Interesting

1. HP-UX

Run Queue Thresholds

Gurus, Having a GS1280 box with OSF1 v5.1 installed (16 processors), the run queue value from the vmstat command reports a very high value (about 25 to 30). Does this reflect a CPU bound system. Note that the average CPU utilization is about 60 % which means that the CPU is not that loaded. Can... (0 Replies)
Discussion started by: Negm
0 Replies

2. AIX

Waiting time of Process

Hi all, I am trying to find out the process wait time on Unix(AIX/SOLARIS) machine( only sh/ksh/csh): Like EXAMPLE 1 : $ vmstat 2 System configuration: lcpu=16 mem=32000MB kthr memory page faults cpu ----- -----------... (4 Replies)
Discussion started by: chandrakala.sg
4 Replies

3. UNIX for Dummies Questions & Answers

Display just CPU run queue number (Nothing Else)

Im using the vmstat command to display the CPU run queue, but i want to put that into a program so is there a way to just display the number under the r? Thanks, (1 Reply)
Discussion started by: RAFC_99
1 Replies

4. UNIX for Advanced & Expert Users

How to measure g++ performance?

I am working on an application with some rather interesting build performance issues. If we build on Solaris/Linux x86/AMD64 the build is rather fast, but it takes more than five times as long on our Solaris Sparc servers (single-threaded builds on the workstations, but multi-threaded on the... (5 Replies)
Discussion started by: Elric of Grans
5 Replies

5. Shell Programming and Scripting

Run from waiting mode

I have no idea why this job has been put into waiting state by server. Can anyone help me to run this job from waiting state. ps -elf | grep 'usr_script' 4 S usr_script 3929 3926 0 77 0 - 2976 wait Oct21 ? 00:00:00 /bin/ksh /application_folder/script/report 0 S usr_script ... (2 Replies)
Discussion started by: zooby
2 Replies

6. Linux

How to Calculate Disk Queue length and Disk Busy Time

Hi, Am doing an enhancements related to monitoring a Linux disk I/O statistics. The /proc/diskstats file is used to get the each disk I/O statistics. But, It returns the raw value. How to calculate the Disk Queue Length and Disk Busy time from the raw values. Guide me. (1 Reply)
Discussion started by: maruthu
1 Replies

7. Shell Programming and Scripting

waiting on jobs in bash, allowing limited parallel jobs at one time, and then for all to finish

Hello, I am running GNU bash, version 3.2.39(1)-release (x86_64-pc-linux-gnu). I have a specific question pertaining to waiting on jobs run in sub-shells, based on the max number of parallel processes I want to allow, and then wait... (1 Reply)
Discussion started by: srao
1 Replies

8. UNIX for Advanced & Expert Users

idle% cpu and run queue

Hi Everybody, Can anybody explain how CPU idle% is about 50%, but runq-sz more than 1? sar from Solaris 10: 00:00:05 %usr %sys %wio %idle 17:00:08 27 12 0 61 17:20:05 40 15 0 45 17:40:05 27 12 0 61 18:00:05 23... (2 Replies)
Discussion started by: sant
2 Replies

9. Solaris

How to measure IOPS?

Hi I have a system running solaris 10, and I intend to use a NetApp as its storage system. The application requires a throughput between the server and the storage 7000 disk IOPS (random IO sustained throughput with response time of 20 mili second and 16k block size). How to make sure that I... (6 Replies)
Discussion started by: fretagi
6 Replies
trace(1)						      General Commands Manual							  trace(1)

Name
       trace - trace system calls of programs

Syntax
       trace [options] cmd args...

Description
       The  command  with  no flag arguments traces for the given cmd and args all system calls made and prints a time stamp, the PID, call and/or
       return values and arguments and puts its output in the file trace.dump.

Options
       -f filename
	       Puts dump in file filename.

       -z      Echos arguments only.

       Only one of the following option arguments can be specified at one time.

       -c#     Traces given PIDs and their children.  Up to sixteen PIDs can be specified.

       -g#     Traces given groups only.  Up to sixteen Group IDs can be specified.

       -p#     Traces given PIDs only.	Up to sixteen PIDs can be specified.

       -s#     Traces given system calls only.	Up to sixteen PIDs can be specified.

       -u#     Traces given UIDs only.	Up to sixteen PIDs can be specified.

Examples
       trace -f ls.dump ls -l /dev >ls.out
       runs the cmd ls -l /dev and puts the trace in ls.dump and output in ls.out.
       trace -f csh.trace -p $$ &
       will trace your login shell in the background. To stop the trace just send it a termination signal (that is, kill -TERM trace_pid).

Restrictions
       Due to security, no one, not even the super-user can trace anyone else's programs. This sort of negates some of the usefulness  of  the	-g
       and -u flags.

       The program cannot be traced.

       Only 16 numbers can be given to the -c, -p, -g, -u, and -s flags.

       The kernel configuration file must contain the following:
       options	       SYS_TRACE
       pseudo-device   sys_trace

       In addition, the superuser must use the following command sequence to create the device:
       cd /dev
       MAKEDEV trace
       If both lines are not in the configuration file or if the device is not made, the message "Cannot open /dev/trace" appears.

Files
       /dev/trace     read only character special device for reading syscall data.

       trace.dump     default file for the system call trace data.

See Also
       open(2), close(2), ioctl(2), select(2), read(2), trace(5)

																	  trace(1)
All times are GMT -4. The time now is 12:54 PM.
Unix & Linux Forums Content Copyright 1993-2022. All Rights Reserved.
Privacy Policy