Reporting with Excel

[James Hill <ev...@gmail.com>]

Hi all.

I've been struggling a little getting reports to output in a friendly
format. I've managed to use the XSL script found here
http://wiki.apache.org/jakarta-jmeter/LogAnalysis to get jtl files into a
readable format. Well, a sort of readable format. I get the column headers
displaying but the actual column data is strangely invisible. It can be
viewed via View Source and, when viewed in IE7, imported into Excel with no
stylesheet applied.

I can use the formula =(x/1000+(t*3600)+((365*70+17)*86400))/86400 where x =
the cell and t = the GMT offset to convert the UNIX timestamp to a epoch
time value. No problem there.

My problem comes when I'm importing large files into Excel. With a
relatively short run I'm already at 31000 rows. Excel 2003 has an upper
limit of ~65000 rows. My question is how do I deal with results from long
running tests? I'm not looking to graph every single response from those
larger runs but the timing data allows me to get 90th percentile numbers as
well as graph the results.

I've tried using the script jtlmin.sh but it only returns results like the
below, even if I reduce the time of the slice to say 5 seconds:

unixtime     date     time     thruput(tpm)     response(ms)
0    1970.Jan.01    10:00    61093    0

Not sure how I can have a response time of 0 even if the file imported into
excel has a 90th percentile of 24ms.

Any suggestions or pointers to help?

Regards,
James.

Re: Reporting with Excel

[sebb <se...@gmail.com>]

On 14/12/2009, James Hill <ev...@gmail.com> wrote:
> Hi all.
>
>  I've been struggling a little getting reports to output in a friendly
>  format. I've managed to use the XSL script found here
>  http://wiki.apache.org/jakarta-jmeter/LogAnalysis to get jtl files into a
>  readable format. Well, a sort of readable format. I get the column headers
>  displaying but the actual column data is strangely invisible. It can be
>  viewed via View Source and, when viewed in IE7, imported into Excel with no
>  stylesheet applied.
>
>  I can use the formula =(x/1000+(t*3600)+((365*70+17)*86400))/86400 where x =
>  the cell and t = the GMT offset to convert the UNIX timestamp to a epoch
>  time value. No problem there.
>
>  My problem comes when I'm importing large files into Excel. With a
>  relatively short run I'm already at 31000 rows. Excel 2003 has an upper
>  limit of ~65000 rows. My question is how do I deal with results from long
>  running tests? I'm not looking to graph every single response from those
>  larger runs but the timing data allows me to get 90th percentile numbers as
>  well as graph the results.
>
>  I've tried using the script jtlmin.sh but it only returns results like the
>  below, even if I reduce the time of the slice to say 5 seconds:
>
>  unixtime     date     time     thruput(tpm)     response(ms)
>  0    1970.Jan.01    10:00    61093    0
>
>  Not sure how I can have a response time of 0 even if the file imported into
>  excel has a 90th percentile of 24ms.
>
>  Any suggestions or pointers to help?

You can read JTL files into JMeter Listeners; both Summary Report and
Aggregate Report allow the calculate results to be exported, and for
other Listeners you may be able to copy/paste.

>  Regards,
>
> James.
>

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