I wrote a short script to do this which takes a CSV format JTL file and
summarizes it in a similar way to the (slow) aggregate report listener.
It uses a local postgres instance to import and summarize the CSV, and it
runs very fast even on huge JTL files. If this fits your needs feel free
to use and share. It's short; I'll paste it below. If anybody uses this
script and enhances it to print 90th percentile, please share again.
Regards,
Ken
On Wed, 29 Jul 2009 05:40:59 -0400, Jantien Sessink <[email protected]> wrote:
The problem is that after running the tests we get a lot of .jtl files
that have to be opened in the Summary Report Listener of the JMeter GUI
one by one, then saved to .csv and then imported in MS Excel for further
analyses. This is very time consuming and we would like to automated
that step. Is it possible to convert the .jtl output to a summary report
in .csv format without using the GUI?
---------------------------jtlsummary.sh------------------------------------
#!/bin/sh -
# summarize results in one or more .jtl (csv) files
#
# by kenstir 2009-07-01
# could be an option
dbname=$LOGNAME
# build sql here
tmpfile=/tmp/jtlsummary.$$.sql
$tmpfile
set -e
cat >>$tmpfile <<EOF
create temp table results (timeStamp bigint,elapsed int,label
text,responseCode text,responseMessage text,threadName text,dataType
text,success boolean,bytes int,Latency int);
EOF
for csvfile in "$@"; do
case "$csvfile" in
/*) ;;
*) csvfile="$PWD/$csvfile";;
esac
cat >>$tmpfile <<EOF
copy results from '$csvfile' csv header;
EOF
done
cat >>$tmpfile <<EOF
--analyze results;
create temp table agg as
select label
,count(*)
,avg(elapsed)
,min(elapsed)
,max(elapsed)
from results
group by label
order by label;
copy agg to stdout csv header;
EOF
psql -q -f $tmpfile
rm $tmpfile
-------------------------------------end---------------------------------------------
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
