how do I run concurrent background process from shell script? - bash

Question

I tried the following:
#!/bin/bash
while read device; do
name=$(echo "$device" | awk '{ print $1 }')
ip=$(echo "$device" | awk '{ print $2 }')
while read creds; do
community=$(echo "$creds" | awk '{ print $1 }')
version=$(echo "$creds" | awk '{ print $2 }')
mkdir -p walks/$name;
`echo -e "snmpwalk -v$version -c \x27$community\x27 $ip system > walks/$name/$community-$version.txt
done < <(##MySQL query that returns tuples in form: (snmp_ro,(1,2c,3))##")
done < <(cat devices.txt)
exit 0
This is meant to go through and find the snmp string and version of each device.
devices.txt is a list of devices in form: hostname ip
It doesn't create the file: walks/$name/$community-$version.txt, and it only seems to run through the walks 1 at a time, something I don't want.

Answer 1

Use & to put the contents you want backgrounded in, well, the background.
pids=( )
while read -r -u 3 name ip _; do
while read -r -u 4 community version _; do
mkdir -p "walks/$name"
snmpwalk -v"$version" -c "$community" "$ip" system \
</dev/null >"walks/$name/$community-$version.txt" & pids+=( "$!" )
done 4< <(: get data for "$name" and "$ip")
done 3<devices.txt
wait "${pids[#]}"
Other items of note:
read can already split fields into their own variables; using awk for this is silly.
The _ in read -r foo bar _ ensures that if more than two columns exist in the input file, the third column and onward are discarded (actually, put into a variable named _, but this is considered discard by convention) rather than appended to bar.
Make a habit of quoting expansions unless you have a specific and compelling reason to do otherwise; otherwise, you get string-splitting and glob expansion of string contents.
This example puts each input stream on its own file descriptor, and redirects each read to its own FD. This prevents any other content within your loop from consuming stdin.