I have a huge test and have to add -Xms256M and -Xmx256M when I start
JMeter. I have my own startup scripts. I also added them as jvm parameters
to my ant nightly functional tests as it would run out of memory also.
-Original Message-
From: BAZLEY, Sebastian [mailto:[EMAIL PROTECTED]
I hope your at least going to throttle the requests of those 1 users.
You'll be hard pressed to find a system that requires a test like this.
Measure the scalability of your app first. Then take a true look at how many
people will be using it and then look at HOW they will be using it (where
Ya, but until they get bulk copy/paste working its much less of a pain in
the a$$ than dragging each http request recorded.
:(
-Original Message-
From: Lee Peterson [mailto:[EMAIL PROTECTED]
Sent: Friday, February 06, 2004 10:16 AM
To: JMeter Users List
Subject: RE: Changing controller
Sounds like you need an XSLT processor. Just go to
http://xml.apache.org/xalan-j/index.html
download xalan for java and just dump it in your ant\lib directory.
Hope this helps,
Mike
-Original Message-
From: Vijay [mailto:[EMAIL PROTECTED]
Sent: Thursday, January 22, 2004 7:02 AM
To:
I've distributed my load across multiple computers when running with a lot
of users. BTW, I usually equate about 10 threads to about 100-200 actual
concurrent users. The number of users the number of threads seems to
represent is actually about 10-20X the thread number. This is because the
threads
This is not scientific. I've just used JMeter a lot and tweaked these
numbers. This is about what I've come up with. A factor of 10-20X the thread
amount is roughly equal to the number of actual concurrent users. This
depends on your actual application, user base, etc. but it will usually hold
If you want to monitor the number of actual sessions you can just put a
servlet at the application level that looks at the number of active
sessions. This number comes from my years of experience. No science, just a
number I found that 'feels' right. That's why its 100-200. 100 for an app
that is
I'm not sure it was OOM error but I've been doing this long enough to know
the signs. I didn't want to post the logs here because it annoys me when
people post anything over 30K like large attachments to email lists. I
usually just throw the file on my web site and post the link but I'm at work
The JMeter GUI has nothing to do with a 1 minute reponse from and HTTP
Request unless your running it on a commodore 64. HTTP Requests are usually
VERY small strings sent across a wire. Not much to it. I think you need to
read more about how the whole web protocol works. The bottleneck is on the
I have a very large test plan. Yesterday I noticed that it started saving my
file with 0 bytes! It was almost 6.5 megs. It over wrote what I had so its
gone. Then, it started giving me an error where it can't save it at all. It
did the same thing again today. Luckily I was saving under the names
One thing you can do, if you run in a windows environment, is open files of
type jmx with 'open' under file types with the line;
C:\java\jdk1.4\bin\javaw.exe -jar
c:\java\jakarta-jmeter-1.9.1\bin\ApacheJMeter.jar -t %1
This associates jmx files with JMeter and opens it in JMeter automatically.
I have an application that can have multiple context roots. I need to be
able to change this in one place rather than in every HTTP Request. I tried
putting it in HTTP Request Defaults Path but the proxy recorder only records
urls that match the path exactly. I would like something that will look
I noticed that if you put a http request default in your recording controller then it
will not put the values defined in your default each new http request. This is good. I
switch between servers a lot. The problem is, now the servers are giving me a problem
because it has this error message in
the plan.
Looks like its having trouble with the xml or xsl parser?
any help is greatly appreciate,
thx,
Michael Lee
build.xml task/target
target name=regression_test description=Tests the system functionality.
jmeter
jmeterhome=C:\java\jakarta-jmeter-1.8.1
Actually, what it sounds like you need more is a code profiler such as
Optimize it. That's what we use. It's pretty good. You should look at what
sections of your code take up the most time and focus on those for
performance.
Mike
- Original Message -
From: temidayo ogunsanwo [EMAIL
the impression you weren't
using the provided
scripts).
Regarding the counter problem - the 1.8 counter did have a bug, but not
quite what you are
reporting. Nonetheless, you might want to grab the latest code. Many
problems have been
fixed.
-Mike
On 5 Jan 2000 at 1:01, Michael Lee wrote:
I
There appears to be no counter example in the distribution. I can not, for the life of
me or anyone on the team, get this to work. It starts at 1 and goes to with an
increment of 1 and a name of suffix. I have a user creation screen and I have a userID
field where i set the http parameter
it to run! (The one I compile, not the
distribution jar..which works).
thanks,
Michael Lee
to the correct page, which is wierd. I would think it would
just fail.
ie:
http://some-intranet-server/somepage.html
If I click on this link with the HTTP Proxy Server runing and no VPN it works. With
VPN it doesnt. Im using the VPN that comes with Win XP Pro.
thanks and kick a$$ tool btw,
Michael Lee
.
thanks and kick a$$ tool btw,
Michael Lee
Software Architect
AfterBOT
__
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
http://mailplus.yahoo.com
--
To unsubscribe, e-mail:
mailto:[EMAIL PROTECTED]
For additional
20 matches
Mail list logo