Awk script to analyse apache log to find bottleneck

awk-analyse-apachelog.sh

#!/bin/bash

#command line to analyse apache log to get most consuming ressources

## Configure apache log

### Apache log format must be changed
#- %D is added to get service time in microsecondes
#- separator | is used in apache log to facilitate the analysis (awk -F"|")

### Change configuration in your vhost or apache configuration
#	LogFormat "%h|%{X-Forwarded-For}i|%t|%v|%p|%r|%>s|%b|%D" customlogtime
#	CustomLog /var/log/apache2/access.log customlogtime

# adjustement needed to match your business case
# change the 26 value (length of the url)
# change the sed value (don't take into account queryr parameters that change too much)
# change the grep value (to include or exclude patterns)


echo -e "          Page         TotalServiceTime         NbCalls      AverageServicetime"

cat /var/log/apache2/access.log | grep -v 'images' | sed 's/POST/POS/g' |sed 's/DELETE/DEL/g' | sed 's/\/v1\//\/v1***                           /g'| awk -F'|' '{ key=substr($6,6,26); tot[key]+=$9; cpt[key]++; } END { for (key in tot) printf("%20s %14d %14d %14d\n", key, tot[key], cpt[key], tot[key]/cpt[key]); }' | sort -k2n | tail -50