On 03/02/2009, Axel Knauf <[email protected]> wrote:
> Hi.
>
>  On Tue, Feb 3, 2009 at 13:13, sebb <[email protected]> wrote:
>  >> [ Firewall issues with random RMI ports ]
>
> > Seems to me that the simplest solution would be to abandon trying to
>  > use client-server mode and just run the tests as non-GUI JMeter
>
> > instances on each of the servers. [..]
>
>  Thanks for the suggestion. But in case I run the test cases on each
>  load generator host as a separate process without having them
>  synchronized by a common controller instance, how do I get the
>  aggregated load statistics and results?

This has been covered in the past; there may be some useful info in
the mail archives.

>  When running several load tests from the diverse hosts each one would
>  log its own result data which might be really hard to put together
>  afterwards. I have seen that there are several different approaches
>  for analyzing the generated logs (e. g. by xslt transformation or perl
>  analyses[0]), but neither of them aggregates several _separate_ log
>  files (or am I missing something?).

If you use CSV output (which is cheaper), then it is trivial to merge
the files - the entries don't have to be in timestamp order. You just
need to remove all but one header.

In any case, you need to ensure that each host collects the same
information (i.e. the same columns).

>  Any suggestions for this one? Since the environment we are trying to
>  test is decently powered (near-productive performance), only using a
>  single load generator will surely not be enough.

Quite possibly, though a single JMeter can generate a substantial load
if the test plan does not have much in it except samples.

If you do need lots of JMeter instances, then using client-server mode
is likely to put a big strain on the client system and its network
connection.

>  Cheers,
>  Axel
>
>
>
>  --
>  [0] http://wiki.apache.org/jakarta-jmeter/LogAnalysis
>
>
>  ---------------------------------------------------------------------
>  To unsubscribe, e-mail: [email protected]
>  For additional commands, e-mail: [email protected]
>
>

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to