Bash sizeout script - linux

Question

I like very much the style, how bash handle the shells.
I am looking for the native solution to cover a bash command for testing the size of a result file and exit in the case of that became too big in size.
I am thinking about a command like
sizeout $fileName $maxSize otherBashCommand
It would be usefull to use it in a backup script like:
sizeout $fileName $maxSize timeout 600s ionice nice sudo rear mkbackup
To make it one step more complicated, i would call it over ssh:
ssh $remoteuser#$remoteServer sizeout $fileName $maxSize timeout 600s ionice nice sudo rear mkbackup
What kind of design pattern should i use for this ?
Solution
I have modified Socowi's code a little
#! /bin/bash
# shell script to stop encapsulated script in the case of
# checked file reaching file size limit
# usage
# sizeout.sh filename filesize[Bytes] encapsulated_command arguments
fileName=$1 # file we are checking
maxSize=$2 # max. file size (in bytes) to stop the pid
shift 2
echo "fileName: $fileName"
echo "maxSize: $maxSize"
function limitReached() {
if [[ ! -f $fileName ]]; then
return 1 # file doesn't exist, return with false
fi
actSize=$(stat --format %s $fileName)
if [[ $actSize -lt $maxSize ]]; then
return 1 # filesize under maxsize, return with false
fi
return 0
}
# run command as a background job
$# &
pid=$!
# monitor file size while job is running
while kill -0 $pid; do
limitReached && kill $pid
sleep 1
done 2> /dev/null
wait $pid # return with the exit code of the $pid
I added wait $pid to the end, that returns with the exit code of the background process instead of it's on exit code.

Answer 1

Monitor the File Size Every n Time Units
I don't know whether there is a design pattern for your problem, but you could write the sizeout script as follows:
#! /bin/bash
filename="$1"
maxsize="$2" # max. file size (in bytes)
shift 2
limitReached() {
[[ -e "$filename" ]] &&
(( "$(stat --printf="%s" "$filename")" >= maxsize ))
}
limitReached && exit 0
# run command as a background job
"$#" &
pid="$!"
# monitor file size while job is running
while kill -0 "$pid"; do
limitReached && kill "$pid"
sleep 0.2
done 2> /dev/null
This script checks the file size every 200ms and kills your command if the file size exceeds the maximum. Since we only check every 200ms, the file may end up with (yourWriteSpeed Bytes/s × 0.2s) more than the specified maximum size.
The following points can be improved:
Validate parameters.
Set a trap to kill the background job in every case, for instance when pressing Ctrl+C.
Monitor File Changes
The script from above is not very efficient, since we check the file size every 200ms, even if the file does not change at all. inotifywait allows you to wait until the file changes. See this answer for more information.
A Word on SSH
You just need to copy the sizeout script over to your remote server, then you can use it like on your local machine:
ssh $remoteuser#$remoteServer path/to/sizeout filename maxSize ... mkbackup