Sponsored Content
Operating Systems Linux Doubt about programs for testing Linux performance Post 302328900 by MarkSeger on Thursday 25th of June 2009 12:00:25 PM
Old 06-25-2009
I think one of the biggest mistakes people make when testing is to time the test from start to finish and not pay attention to anything in between. For example, I always use netperf to measure network performance, or dt to measure disk as well as others, BUT I always run collectl in another window to see what's happening to my cpu, disk, network, memory and other subsystems while the test is in process. If you get an bad end-to-end number, and the intermediate numbers are very erratic it just may be a system or network switch is misconfigured. Looking at the elapsed time or average load will never give you a true picture.

In fact, if you run collectl with a monitoring interval of 0.1 seconds or even less which doing disk tests you can actually watch the cache fill as the tests run faster at first.

-mark
 

7 More Discussions You Might Find Interesting

1. AIX

Performance testing on AIX

I'm doing performance testing for one application which works on AIX. But I don't know which performance parameters of memory need to be collected. Now, I just know very few: 1. page in 2. page out 3. fre They are all collected by "vmstat" command. I want to know, except for above... (2 Replies)
Discussion started by: adasong
2 Replies

2. Web Development

Suggested tool / approach for performance testing

What is a good approach for a performance testing tool suite for web applications? I am specifically interested in tools that execute a certain set of tasks well as opposed to tuning high traffic sites. In other words, a profiler would be a good idea to have, although I understand these tools are... (4 Replies)
Discussion started by: figaro
4 Replies

3. IP Networking

using multiple iperf instances for performance testing

Hello all! I have the need to performance-test a MPLS switch, I was thinking of using iperf to accomplish the task. I had in mind using a linux box with a Gigabit interface connected to a L2 switch on a 802.1Q trunk. In the interface I would create 20 VLANs with 20 different IP subnets. ... (0 Replies)
Discussion started by: ppucci
0 Replies

4. Emergency UNIX and Linux Support

using multiple iperf instances for performance testing

Hello all! I have the need to performance-test a MPLS switch, I was thinking of using iperf to accomplish the task. I had in mind using a linux box with a Gigabit interface connected to a L2 switch on a 802.1Q trunk. In the interface I would create 20 VLANs with 20 different IP subnets. ... (2 Replies)
Discussion started by: ppucci
2 Replies

5. UNIX for Dummies Questions & Answers

Unix Performance testing - training content

Hi Everyone, My company is involved in performing performance testing and now they want to perform couple of training related with executing those tests on the servers based on Unix sytems. And I have to provide them draft of the content for those trainings. I think this kind of training... (2 Replies)
Discussion started by: Bartuss
2 Replies

6. Shell Programming and Scripting

Help with extract application logs through shell script in performance testing

Hi Experts, I am new to shell.How to extract logs (Web,APP,Database) using shell in performance testing? Need for webserver logs,app server logs and d/b logs code. Thanks in advance Sree (3 Replies)
Discussion started by: sree vasu
3 Replies

7. AIX

AIX Memory and Performance Doubt

Hi All, I have 5 Servers ( 3 DataStages Server and 2 Database Servers running HACMP and DPF). My question is 1 of the Main Core DataStage Server has few unsolved issues that I will post the question as following 1. Why most of time the File Cache in the memory seems constantly... (3 Replies)
Discussion started by: ckwan
3 Replies
GENPLOTFILES(1) 						      colplot							   GENPLOTFILES(1)

NAME
genplotfiles - utility to help automate creation of collectl plot file from raw ones SYNOPSIS
/opt/hp/colplot/genplotfiles.pl -outdir [-switches] DESCRIPTIONS
This utility is not intended to meet all plot file creation needs, but quite frankly should meet most. Features not provided with this utility are best realized by running collectl directly in playback mode (see collectl -p). It is therefore important to understand what this utility will and will not do for you before building an automation strategy around it. The algorithm used is fairly simple - genplotfiles examines each collectl raw file to see it it falls within the data ranges (if specified) and contains a matching hostname (if specified). If a plot file already exists, the timestamps are compared and the raw file skipped if its newer than the plotfile. Any raw files that pass all these tests are then passed to collectl with their timestamp wildcarded. The result is all raw files for that host/date will be processed. This means if you copy over collectl raw files multiple times over the course of the day using rsync, you can immediately genererate up-to-date plots for them using genplotfiles. Furthermore, you won`t even need any date parameters if the older plot files are alreay up to date. For an example of what a cron job that aggregates multiple collectl logs onto a single system for plotting purposes, see crontab.sh in the same directory as this script. SWITCHES
All switches are optional with the exception of -outdir: -colbin path-to-collectl This points to collectl and is only necessary if /usr/sbin is not in your path or you`ve chosen to install collectl in a non-stan- dard directory. -debug number Primarily for development, will provide intermediate processing information based on its value. See souce for details. -force Always generate plot files even in there is an existing one that is newer than the latest raw file. This will be necessary if you want to rerun this script and just change the value of -subsys. -help Print a summary of available switches. -host string Only process files that contain this string in their hostname portion of the filename. -indir directory Directory where to look for collectl logs. The default is /var/log/collectl and is therefore only necessary if your logs are in a non-standard location. -outdir directory Directory into which to write the plottable files. This field is required. -fromdate date Ignore files with a datestamp in their name less that this value. The default is 20050101 and so defaults to all files. The value of `yesterday` is also recognized to simplify scripts that call this. -thrudate date Ignore files with a datestamp in their name with a datestamp greater than this value. The default is the current day and one can also specify a value of `yesterday`. -subsys string Instead of using the default subsystems for which data was collected, generate output files based on this set. There are basically two reasons for doing this. The first is that you may want to reduce the size of the plot files, though with compression there is rarely much point in this. The second is if you`ve asked collectl to collect summary data but want to plot detail data. See the colplot FAQ for more information. AUTHOR
This program was written by Mark Seger (mjseger@gmail.com). Copyright 2005 Hewlett-Packard Development Company, L.P. SEE ALSO
LOCAL
OCTOBER 2005 GENPLOTFILES(1)
All times are GMT -4. The time now is 05:19 PM.
Unix & Linux Forums Content Copyright 1993-2022. All Rights Reserved.
Privacy Policy