jstats2 - Parse Jmeter output (csv), compute statistics about the test execution and response times.

jstats2

#!/bin/bash
# Parse jmeter output and get various stats
# Hat-tip to 'stig', https://git.io/vFmeI
set -e     # Exit immediately on error
# set -x   # Debug
usage="jstats2 <jmeter-csv-output-file>"
[ $1 ] || { echo "$usage" ; exit 1 ; }
JFILE=$1

# Show timestamp info
UT1=$(sed -n '2p' $JFILE | cut -c -10)
UT2=$(tail -1 $JFILE | cut -c -10)
DUR=$(( $UT2 - $UT1 ))
echo -e "\n\033[1m --test run details--\033[0m"
echo " started  $(date --date="@$UT1" +"%y%m%d-%H%M%S")"
echo " stopped  $(date --date="@$UT2" +"%y%m%d-%H%M%S")"
echo "duration  $DUR sec"

# Create temp file of elapsed times (column 2 of the output csv)
WORK=$(mktemp /tmp/$(basename $0).XXXXX) || exit 1
cut -d, -f2 $JFILE | grep . | grep -v "elapsed" | sort -n > $WORK
echo "tmp-file  $WORK"

N=$(wc -l < $WORK)
echo " count n  $N"
echo "avg-load  $(( $N / $DUR )) tps"

# fail count
FC=$(awk -F, '$8 ~ /false/{print}' $JFILE | wc -l)
FP=$(echo "scale=6; 100 * $FC / $N" | bc | awk '{printf "%3.3f", $0}')
echo "   fails  $FC ($FP %)"

# success count
SC=$(( $N - $FC ))
SP=$(echo "scale=6; 100 * $SC / $N" | bc | awk '{printf "%3.3f", $0}')
echo " success  $SC ($SP %)"

echo -e "\n\033[1m --response stats (ms)--\033[0m"
P50=$(dc -e "$N 2 / p")
P90=$(dc -e "$N 9 * 10 / p")
P95=$(dc -e "$N 95 * 100 / p")
P99=$(dc -e "$N 99 * 100 / p")
echo "  median  $(awk "FNR==$P50" $WORK)"
echo "     p90  $(awk "FNR==$P90" $WORK)"
echo "     p95  $(awk "FNR==$P95" $WORK)"
echo "     p99  $(awk "FNR==$P99" $WORK)"
echo -e "  mean \u03BC  $(awk '{sum+=$1}END{if (NR>0) print sum/NR}' $WORK)"
echo -e "stddev \u03C3  $(awk '{x+=$0;y+=$0^2}END{print sqrt(y/NR-(x/NR)^2)}' $WORK)"
echo "     min  $(awk "FNR==1" $WORK)"
echo "     max  $(awk "END{print}" $WORK)"

Example output

$ jstats2 OUTPUTFILE-2017-10-31-114547.csv

 --test run details--
 started  171031-114548
 stopped  171031-115703
duration  675 sec
tmp-file  /tmp/jstats2.tQinR
 count n  316607
avg-load  469 tps
   fails  752 (0.238 %)
 success  315855 (99.762 %)

 --response stats (ms)--
  median  30
     p90  32
     p95  34
     p99  93
  mean μ  32.5023
stddev σ  19.9347
     min  27
     max  3201

Next step: graphical analysis with gnuplot

A more generic version that computes stats for any data on stdin: istats