joining a critsit in a large bank in Athens – WebSphere Portal 5.1, WCM, TAM ,TDS, TDI, ESB, db2, etc. Massive performance issues caused by inadequately tuning the environment, lack of system resources – basically returned to default values following tuning guides. Primary intention was to stabilize environment, as well as providing mid- to long term recommendations about how to manage the infrastructure and systems / applications.

Athens is a beast with traffic collapsing – would go mad living here. Akropolis area (Plaka) very touristic (what did I expect), but nice to switch off from a busy day (apart from all the little gangsters trying to get hold of your money).

edit /etc/apache2/mod_log_config.conf and add a new logformat (plots) – (“mod_log_config.conf” is referenced by “httpd.conf”)

LogFormat “%h %l %u %t \”%r\” %>s %b \ \”%{Referer}i\” \”%{User-Agent}i\”" combined
LogFormat “%h %{%d.%m.%Y:%H:%M:%S}t %D %U” plots

The characteristics of the “%” directives are:

%…h Remote host
%…{format}t time & date
%…D time taken to serve the request, in microseconds.
%…U URL path requested

Add another “customlog” directive to your “virtualhost” section in “httpd.conf”:

DocumentRoot /home/h/hensler.net/public_html/bernhard/
ServerName bernhard.hensler.net
IndexOptions
DirectoryIndex index.htm index.html index.shtml start.htm start.html start.shtm index.php

CustomLog “/usr/local/visas/logfiles/hensler.net/%Y/%m/%d/access_log” vhost_combined
CustomLog “/usr/local/visas/logfiles/hensler.net/bernhard.access_log” plots

Concatenate logs from all virtual hosts e.g.: cat hensler.access_log niko.access_log bernhard.access_log max.access_log > plot_log (sample line: 66.249.111.111 30.08.2009:14:15:17 4372853 /blog/) and start gnuplot from the command line:

$ gnuplot

reset
set terminal png small color
set output “2dplot.png”
set title “average response time”

set style data points
set pointsize 1
set grid

set xlabel “time”
set timefmt “%d.%m.%Y:%H:%M:%S”
set format x “%H:%M\n%d/%b”
set xdata time
set xrange [ "30.08.2009:00:00" : "30.08.2009:23:59" ]

set ylabel “response time”
set yrange [ 0 : 10000 ]

plot “/usr/local/visas/logfiles/hensler.net/plot_log” using 2:3 title “2d”

gnuplot 3d

Then read this excellent article about “A New Visualization for Web Server Logs” and create a perl script:

#
# prepare-for-gnuplot.pl: convert access log files to gnuplot input
# Raju Varghese. 2007-02-03

use strict;

my $tempFilename = “./tmp/temp.dat”;
my $ipListFilename = “./tmp/iplist.dat”;
my $urlListFilename = “./tmp/urllist.dat”;

my (%ipList, %urlList);

sub ip2int {
my ($ip) = @_;
my @ipOctet = split (/\./, $ip);
my $n = 0;
foreach (@ipOctet) {
$n = $n*256 + $_;
}
return $n;
}

# prepare temp file to store log lines temporarily
open (TEMP, “>$tempFilename”);

# reads log lines from stdin or files specified on command line

while (<>) {
chomp;
my ($ip, $time, $D, $url, $sc) = split;
$time =~ s/\[//;
next if ($url =~ /(gif|jpg|png|js|css)$/);
print TEMP "$ip $time $D $url $sc\n";
$ipList{$ip}++;
$urlList{$url}++;
}

# process IP addresses

my @sortedIpList = sort {ip2int($a) <=> ip2int($b)} keys %ipList;
my $n = 0;
open (IPLIST, ">$ipListFilename");
foreach (@sortedIpList) {
++$n;
print IPLIST "$n $ipList{$_} $_\n";
$ipList{$_} = $n;
}
close (IPLIST);

# process URLs

my @sortedUrlList = sort {$urlList {$b} <=> $urlList {$a}} keys %urlList;
$n = 0;
open (URLLIST, ">$urlListFilename");
foreach (@sortedUrlList) {
++$n;
print URLLIST "$n $urlList{$_} $_\n";
$urlList{$_} = $n;
}
close (URLLIST);

close (TEMP); open (TEMP, $tempFilename);
while () {
chomp;
my ($ip, $time, $D, $url, $sc) = split;
print "$time $ipList{$ip} $urlList{$url} $sc\n";
}
close (TEMP);

Run this perl script and redirect output to a file from the command line:

$ perl gnuplot.pl "/usr/local/visas/logfiles/hensler.net/bernhard.access_log" > gnuplot.input

The fields in gnuplot.input, the output file of the Perl script, are date/time, ip rank, url rank.

Run gnuplot from the command line: $ gnuplot and the following commands:

reset
set terminal png small color
set output "3dplot.png"
set style data dots
set xdata time
set timefmt "%d.%m.%Y:%H:%M:%S"
set zlabel "Content"
set ylabel "IP address"
splot "gnuplot.input" using 1:2:3 title "3d"

Image taken from oreillynet, my website is not producing sufficient data …

• X, the time axis–a full day from midnight to midnight of November 16.
• Y, the requester’s IP address, with the conventional dotted decimal format sorted and given an ordinal number between 1 and 120,000, representing the number of clients that accessed the web server.
• Z, the URL (or content) sorted by popularity. Of the approximately 60,000 distinct pages on the site, the most popular URLs are near the zero point of the Z-axis and the least popular ones at the top.

http://www.ibm.com/developerworks/linux/library/lgnuplot

http://www.oreillynet.com/pub/a/sysadmin/2007/02/02/3d-logfile-visualization.html?page=1

http://phasorburn.com/index.php/archive/excel-0-gnuplot-1

A final step will cover loadrunner tools like openSTA and jmeter.

See also Part I of this tutorial.

[SinglePic not found]

There are many tools for verbose GC log analysis, these tools can abstract useful information, and illustrate the trend of JVM heap size usage over time as the following. [SinglePic not found]

IBM Pattern Modeling and Analysis Tool for Java Garbage Collector (PMAT) parses the IBM verbose GC trace and provides a comprehensive analysis of the Java heap usage. It then recommends key configurations by first executing a diagnosis engine and then employing a pattern modeling algorithm in order to make recommendations to optimize the Java heap usage for a given JVM cycle.

source for this article
PMAT
JDK Diagnosis

see also here

WASX7209I: Connected to process “WebSphere_Portal” on node wcl using SOAP connector; the type of process is: unManagedProcess; WASX7029I: For help, enter: “$Help help”

wsadmin> set jvm [$AdminControl completeObjectName type=JVM,process=WebSphere_Portal,*]

WebSphere:name=JVM,process=WebSphere_Portal,
platform=dynamicproxy,node=wcl,j2eeType=JVM,
J2EEServer=WebSphere_Portal,version=6.0.2.17,
type=JVM,mbeanIdentifier=JVM,cell=wcl

wsadmin>$AdminControl invoke $jvm generateHeapDump
wsadmin>$AdminControl invoke $jvm dumpThreads

IBM’s support assistant (ISA) is an Eclipse based tool with quite a number of plugins to analyze heapdumps – a good developerworks article can be found here (part I) and here (part II).

To analyze WebSphere Portal performance issues, although this document is applicable to isolate performance problems around Java based application servers in general , see here.

see also here

joining a critSit in Doha – Sametime 802 Domino cluster, WebSphere based Enterprise Meeting Service (EMS) cluster in one local zone on top of SUN Solaris. Presenter & viewer out of sync in a web conference, frame size de-, reassembling going on, IOCP layer ?

This post is password protected. To view it please enter your password below:

Password:

This post is password protected. To view it please enter your password below:

Password:

Auf alle Fälle wird jede Minute kostbar genutzt, nach 3 Tagen mit dieser Mannschaft brauche ich 1 Woche Erholung.

WiKi Eclipse

codeproject Hannover

Jeff Eisen Blog

Hannover brings the most significant changes to the Notes client since probably Notes 4.0, when we introduced LotusScript.

The technical key to the changes is the tight integration with Eclipse. I highlight technical, because there is also an enormous emphasis in Hannover on user interface and productivity improvements and innovations such as Activity-based collaboration as well as the productivity editors. The UI I’ll leave to Mary Beth Raven to discuss in her blog. Other stuff, I’ll get to in good time.

For those uninitiated in Eclipse, here’s a quick description of Eclipse and why it matters.

Eclipse is a cross-platform open source Java project with heavy IBM involvement. It began as a platform for building IDEs (compilers, editors, debuggers, etc.) It uses a contribution model, which means that if the platform lacks functionality or behavior, you can add it yourself by writing some Java code, packaging it in something called a plug-in, and installing your plug-in. Example: If Eclipse could compile and edit Java, C++, and Pascal, but you really needed COBOL, you could just write a COBOL plug-in (in Java), install it, and there would suddenly be COBOL related menus, actions, behaviors, etc., throughout the user experience. Eclipse 3.0 introduced RCP – the Rich Client Platform, allowing Eclipse to be used for something other than an IDE by allowing one to get rid of the geeky IDE stuff (such as compilers) and by allowing one better control over the UI in order to make it more polished and customizable (such as for branding).

IBM is building a common platform titled Lotus Expeditor (previously named WED), that packages Eclipse RCP with some extra services (around security, synchronization, data, deployment, and so forth) to be used across the Software Group product set. Hannover is simply a consumer of this Lotus Expeditor common platform.

Note that when I said above that a plug-in is written in Java, that does not mean it must be exclusively written in Java. With Java mechanisms such as JNI — allowing Java code to call out to C++ or other language code — as well as Service Oriented Architecture techniques such as web services (OK, I’ve mentioned SOA — now I’m buzzword compliant), a plug-in can be used to bridge between the Java world and a “legacy” non-Java codebase. This is what Hannover does — integrate Java code packaged in plug-ins with the mostly pre-existing and mostly C/C++ Notes codebase. This existing codebase provides the bug for bug compatibility with older versions of Notes, allowing Notes applications written way back in V1 days to still run in Hannover.

So, why is this Eclipse platform interesting to the Notes/Domino customer base and particularly to the application development community?

* Eclipse’s contribution model allows modification of the Notes user experience and general behavior of Notes in ways never before possible. You can go beyond just affecting your application “rectangle”, making more global cross-application extensions.
* Eclipse allows development for Notes by those without traditional Notes skills, enabling the expansive Eclipse development community to participate in Notes development.
* The common platform used throughout IBM (Lotus Expeditor atop Eclipse) allows you to write a plug-in contributing behavior for one product (such as the newest version of Sametime) and have it work in another product (such as Hannover).
* By standing on the shoulders of others (developing Eclipse and Java as well as extensions), we are able to build a cross-platform product (Windows, Linux, and MAC) much more quickly than if we had to do it all ourselves, greatly improving our time to market and your opportunity to get new value. Eclipse enables us to leverage the skills, creativity, and sheer manpower of the dedicated Eclipse open source community.

While the focus of Hannover is clearly the Eclipse-based client, there will still be a “traditional” technology non-Eclipse version of the Notes client delivered as part of the Hannover release. We have admittedly been not too succinct around the nomenclature of how to refer to the Eclipse vs. the non-Eclipse Notes instantiation. In general, unless I make an explicit distinction, I will be using the term Hannover rather loosely, to refer to the Eclipse-based version of the client.

Finally, a word about the server. Like previous releases of Notes, the basic server instrastructure of Hannover is a Domino server — all you need for the vast majority of the functionality including mail, calendaring, contact management, Notes applications, and composite applications is a Domino server. Some functionality, such as activity based collaboration require more than just plain Domino, but that’s optional.