Re: Grinder data files only populated at end of run

On 19/09/13 14:09, Annorax wrote:
> I am relying on the metrics provided in the Grinder data file for analysis.
> I notice that these files are only populated at the end of a process's
> execution. I understand that this is probably working as designed, but a
> problem occurs if a run stops prematurely - all of the data is lost. 
>
> Is there any option to enable these data files to be updated periodically
> during a test? I understand this might slow down execution but if a 3 day
> run stops after 2 days unexpectedly, I would rather have some data than
> none.
>

The data files are written continuously, but are buffered in 64K chunks.

You can alter the buffer size by changing the logback-worker.xml
configuration of the data-file encoder (see
http://grinder.sourceforge.net/g3/logging.html#Changing+the+Logback+configuration)
to replace

    <encoder class="net.grinder.util.logback.BufferedEchoMessageEncoder"/>

with

    <encoder class="net.grinder.util.logback.BufferedEchoMessageEncoder">
      <bufferSize>100</bufferSize>
    </encoder>

for example.

If you don't want to lose any data line if the process crashs, you will
want to flush after each line instead. To do this, use a PatternEncoder
instead:

    <encoder>
        <pattern>%msg%n</pattern>
    </encoder>

- Phil

------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from 
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60134791&iu=/4140/ostg.clktrk

_______________________________________________
grinder-use mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/grinder-use