Debug bad formatted performance data

Hello,

ok that’s already the workflow I had to identify bad datas, I’ve hope there were a way with maybe a DB query to identify services in trouble more easily.

I’ll developp a logfile template or script in order to generate a report as expected so.

Thks for your time @ponchoh 

Hello @ponchoh 

sorry for delay, I finnaly write a short script to proceed:

#!/bin/bash
#
# Script to detect bad formatted performance data sent to Centreon
#
broker_log="/var/log/centreon-broker/central-broker-master.log"
API_login=yyyy
API_pwd=xxxx
CMD_CENTREON="/usr/share/centreon/bin/centreon -u ${API_login} -p ${API_pwd} -o SERVICE -a SHOW"
regex=".*invalid perfdata format in service \((.*):(.*)\):.*"

# temp files
tmp_result="/tmp/perf_tmp"
sort_result="/tmp/perf_sort"
perf_result="/tmp/perf_result"

# Cleaning
rm -f /tmp/perf_*

# populate an array with all entris found in broker log
while IFS= read -r line; do
        my_array+=( "$line" )
done < <(grep "invalid perfdata format" $broker_log | sed -E "s/$regex/\1;\2/")

# echo all entries one per line
for i in ${my_array[@]}
do
  echo $i >> $tmp_result
done

# remove duplicates
cat $tmp_result | sort -u > $sort_result

# search services infos with clapi
for line in $(cat $sort_result)
do
  host=$(echo $line | awk -F';' '{print $1}')
  service=$(echo $line | awk -F';' '{print $2}')
  `${CMD_CENTREON} | grep "$host;.*;$service;" >> $perf_result`
done

if [ -a $perf_result ]
then
  cat $perf_result | /usr/bin/mailx -s "Centreon bad Perf Data format" <email>

fi

As you see not a breakthrough stuff, but at least it provide an easier and clearer output with host and service name rather than just ids.