First consider the following script stored in a file named tester:

Code:

#!/bin/bash
IAm=${0##*/}

var1="""
Custom JAX-RS
$IAm
Custom Shared
$(who am I)
Web 2.0
"""

var2="
Custom JAX-RS
$IAm
Custom Shared
$(who am I)
Web 2.0
"

var3='
Custom JAX-RS
$IAm
Custom Shared
$(who am I)
Web 2.0
'

printf 'var1begin%svar1end\n\nvar2begin%svar2end\n\nvar3begin%sbar3end\n\n' \
    "$var1" "$var2" "$var3"

if [ "$var1" = "$var2" ]
then	echo '"$var1" is the same as "$var2".'
else	echo '"$var1" is not the same as "$var2".'
fi

if [ "$var1" = "$var3" ]
then	echo '"$var1" is the same as "$var3".'
else	echo '"$var1" is not the same as "$var3".'
fi

if [ "$var2" = "$var3" ]
then	echo '"$var2" is the same as "$var3".'
else	echo '"$var2" is not the same as "$var3".'
fi

From your description of how """string""" works, the variable expansion $IAm and the command substitution $(who am i) should not occur in the assignments to var1 and var3, but should occur in the assignment to var2. But, as RudiC said and according to the standards and the bash man page, the above construct is a concatenation of an empty string (""), the evaluation of "string" (including parameter expansions, command substitutions, and arithmetic expansions), and another empty string (""). In other words, """string""" produces exactly the same results as

Code:

"string"

no matter how many <newline> characters are contained in string (as long as there are no unquoted double-quote characters in string).

On my system, running the above code produces the output:

Code:

var1begin
Custom JAX-RS
tester
Custom Shared
dwc      ttys007  Aug 17 13:55 
Web 2.0
var1end

var2begin
Custom JAX-RS
tester
Custom Shared
dwc      ttys007  Aug 17 13:55 
Web 2.0
var2end

var3begin
Custom JAX-RS
$IAm
Custom Shared
$(who am I)
Web 2.0
bar3end

"$var1" is the same as "$var2".
"$var1" is not the same as "$var3".
"$var2" is not the same as "$var3".

Note also that all of the assignments in the above script produce an empty line as the first line in each of the assignments to var1, var2, and var3. If you don't want to include an empty line and don't want variable expansions, command substitutions, or arithmetic expansions to be performed in a variable intended to contain three lines, you should use single quotes like this:

Code:

var='Line1 $var
Line2 $(command)
Line3 $((2 * 3))
'

Moving on. As RudiC said, this question has been asked many times before, except that your input is in shell variables instead of in files. (Although you have not given any reason that makes sense to me as to why you are hard-coding the values of these variables into strings instead of the storing the output produced on the various servers and your expected configuration values in files instead.) The diff utility would not be a way to compare two files unless all lines that you are comparing are guaranteed to always appear in the same order if they are present. But, you might want to consider:

Code:

grep -Fx -f file1 file2
grep -Fxv -f file1 file2
grep -Fxv -f file2 file1

But, grep won't work the way you need it to when you have empty lines in one or both of your files (and you have an empty line as the 1st line in both of your variables).

But what you want can be done with one invocation of awk:

Code:

#!/bin/bash
# configuration from server
var1='
Custom JAX-RS
Custom Shared
Web 2.0
'

# required configuration
var2='
Web 2.0
Custom Shared
IBM WebSphere JAX-RS
'

# matching ones between var1 and var2
printf '%s,\n%s' "$var1" "$var2" | awk -v sq="'" '
BEGIN {	printf("valid=%s", sq)
}
n {	v2[$0]
	if($0 in v1) {
		printf("%s%s", sep, $0)
		sep = ","
	}
	next
}
/^,$/ {	n = 1
	next
}
{	v1[$0]
}
END {	printf("%s\ninvalid=%s", sq, sq)
	sep = ""
	for(i in v1)
		if(!(i in v2)) {
			printf("%s%s", sep, i)
			sep = ","
		}
	sep = ""
	printf("%s\nmissing=%s", sq, sq)
	for(i in v2)
		if(!(i in v1)) {
			printf("%s%s", sep, i)
			sep = ","
		}
	print sq
}'

which produces the output:

Code:

valid=',Web 2.0,Custom Shared'
invalid='Custom JAX-RS'
missing='IBM WebSphere JAX-RS'

But the order in which input lines appear in the output may vary with different versions of awk. (And, note that the leading comma in the valid= line indicates that there is an empty line in both $var1 and $var2.)