You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@jmeter.apache.org by Ivan Rancati <iv...@sharpmind.de> on 2008/03/10 17:55:14 UTC

Testing http dowload of large files

I need to test the performance of a http server with increasingly bigger
files, so I have set up 4 linux hosts that are running jmeter-server 2.3.1
(jdk 1.6, using up to one GB for heap) and one controller running jmeter in
graphical mode (no results tree, only an Aggregate Report).

I don't need to examine the file contents, only the download throughput as
the number of threads and the file size (up to 8GB) increases.
The problem is that, eventually, as I increase the size and/or the number of
threads, I get out of heap space errors on the jmeter-server hosts.
Also, the throughput is slower than if using a plain wget on the same hosts.

This is with both flavours of the http sampler (java and http client).
I have configured the Sample Result Save Configuration to disable saving of
response data and messages.

Is there a setting in the http samplers that tells them to just download the
file as wget -O /dev/null would do, ignoring the contents?

thank you
Ivan
-- 
View this message in context: http://www.nabble.com/Testing-http-dowload-of-large-files-tp15950890p15950890.html
Sent from the JMeter - User mailing list archive at Nabble.com.


---------------------------------------------------------------------
To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jmeter-user-help@jakarta.apache.org


Re: Testing http dowload of large files

Posted by sebb <se...@gmail.com>.
On 12/03/2008, Ivan Rancati <iv...@sharpmind.de> wrote:
>
>
>
>  sebb-2-2 wrote:
>  >
>  >
>  > Unfortunately the data is still saved in the sample result, at least
>  > temporarily.
>  >
>
>
> Thanks for the clarification. I guess a workaround would be using BeanShell
>  with something like
>   return exec("wget -S -O /dev/null http:....");
>

If you are going to use BeanShell you could just use HttpUrlConnection
(or use HttpClient) and read the response yourself.

>  Has anybody tried something similar?
>  I haven't played much with BeanShell and from the documentation on their
>  website I can't seem how to capture the output of an external command

It's Java, so how do you do that in Java?

>  (basically what I need is to see whether the wget failed or not and maybe
>  the throughput, I can get the timing info from jmeter itself)
>
> --
>  View this message in context: http://www.nabble.com/Testing-http-dowload-of-large-files-tp15950890p16002409.html
>
> Sent from the JMeter - User mailing list archive at Nabble.com.
>
>
>  ---------------------------------------------------------------------
>  To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
>  For additional commands, e-mail: jmeter-user-help@jakarta.apache.org
>
>

---------------------------------------------------------------------
To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jmeter-user-help@jakarta.apache.org


Re: Testing http dowload of large files

Posted by Ivan Rancati <iv...@sharpmind.de>.


sebb-2-2 wrote:
> 
> 
> Unfortunately the data is still saved in the sample result, at least
> temporarily.
> 

Thanks for the clarification. I guess a workaround would be using BeanShell
with something like
  return exec("wget -S -O /dev/null http:....");

Has anybody tried something similar?
I haven't played much with BeanShell and from the documentation on their
website I can't seem how to capture the output of an external command
(basically what I need is to see whether the wget failed or not and maybe
the throughput, I can get the timing info from jmeter itself)
-- 
View this message in context: http://www.nabble.com/Testing-http-dowload-of-large-files-tp15950890p16002409.html
Sent from the JMeter - User mailing list archive at Nabble.com.


---------------------------------------------------------------------
To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jmeter-user-help@jakarta.apache.org


Re: Testing http dowload of large files

Posted by sebb <se...@gmail.com>.
On 10/03/2008, Ivan Rancati <iv...@sharpmind.de> wrote:
>
>  I need to test the performance of a http server with increasingly bigger
>  files, so I have set up 4 linux hosts that are running jmeter-server 2.3.1
>  (jdk 1.6, using up to one GB for heap) and one controller running jmeter in
>  graphical mode (no results tree, only an Aggregate Report).
>
>  I don't need to examine the file contents, only the download throughput as
>  the number of threads and the file size (up to 8GB) increases.
>  The problem is that, eventually, as I increase the size and/or the number of
>  threads, I get out of heap space errors on the jmeter-server hosts.
>  Also, the throughput is slower than if using a plain wget on the same hosts.
>
>  This is with both flavours of the http sampler (java and http client).
>  I have configured the Sample Result Save Configuration to disable saving of
>  response data and messages.

Unfortunately the data is still saved in the sample result, at least
temporarily.

>  Is there a setting in the http samplers that tells them to just download the
>  file as wget -O /dev/null would do, ignoring the contents?
>

Not at present.

There's a Bugzilla enhancement for this:

https://issues.apache.org/bugzilla/show_bug.cgi?id=41921


>  thank you
>  Ivan
>
> --
>  View this message in context: http://www.nabble.com/Testing-http-dowload-of-large-files-tp15950890p15950890.html
>  Sent from the JMeter - User mailing list archive at Nabble.com.
>
>
>  ---------------------------------------------------------------------
>  To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
>  For additional commands, e-mail: jmeter-user-help@jakarta.apache.org
>
>

---------------------------------------------------------------------
To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jmeter-user-help@jakarta.apache.org