The Bash code
I wrote this small code to measure time and peak in memory usage of a process.
# Get arguments
MaxMemory="$1"
MaxTime="$2"
Command="$3"
for (( i=4 ; i<="$#"; i++)); do
Command="${Command} ${!i}"
done
echo -e "MaxMemory = ${MaxMemory}\nMaxTime = ${MaxTime}\nCommand = ${Command}"
#### run the given Command in the background
${Command} &
#### Get pid
pid=$!
echo "pid = ${pid}"
#### Monitor resources
MemoryPeak=0
timeBefore=$(date +"%s")
while true;do
# Get memory
mem=$(ps -o rss,pid | grep ${pid} | awk '{print $1}')
# Break if the process has stopped running
if [[ ${mem} == "" ]]; then
#echo "process has stopped"
break
fi
# Set the MemoryPeak of memory
if [ "${mem}" -gt "${MemoryPeak}" ]; then
MemoryPeak=$mem
fi
# If it consumed too much memory, then kill
if [ "${MemoryPeak}" -gt "${MaxMemory}" ];then
#echo "process consumed too much memory"
kill ${pid}
break
fi
# If it consumed too much CPU time, then kill
timeAfter=$(date +"%s")
timeUsage=$((timeAfter - timeBefore))
if [ "${timeUsage}" -gt "${MaxTime}" ];then
#echo "process consumed too much time"
kill ${pid}
break
fi
# sleep
sleep 0.1
done
timeAfter=$(date +"%s")
timeUsage=$((timeAfter - timeBefore))
echo "MEM ${MemoryPeak} TIME ${timeUsage}"
The command called memAndTime works as in
memAndTime ${MaxMemory} ${MaxTime} ./myProcess arg1 arg2 arg3
Question
I am not here to discuss reasons why I am not using timeout, gtimeout, memusg or ulimit.
The code sometimes send the following series of error messages
264: integer expression expected
/usr/bin/memAndTime: line 36: [: 473268
644
912
372: integer expression expected
/usr/bin/memAndTime: line 36: [: 705108
652: integer expression expected
/usr/bin/memAndTime: line 36: [: 842760
384: integer expression expected
The line 36 is
if [ "${mem}" -gt "${MemoryPeak}" ]; then
it feels like the object mem built in mem=$(ps -o rss,pid | grep ${pid} | awk '{print $1}') can sometimes start with a column.
I fail to reproduce the error message outside the usage of this command. Also, the error message is rare enough to make it difficult to reproduce. I fail to understand what is causing it. Can you help finding out the source of this error message and how to fix it?
Aucun commentaire:
Enregistrer un commentaire