The command you tried can't work for at least the following reasons (there may be others):There is no need to convert the timestamps to seconds since the Epoch as long as you convert the date into a string that will sort correctly when doing a string comparison (i.e., 20150301 instead of 15-Jan-2015), and awk is certainly capable of converting the date portion of the format in your log file into the format above before comparing the results to the string produced by the command:

Code:

from="$(date +"###<%Y%m%d %H:%M:%S o'clock GMT>")"

.
If we assume that <info> is shorthand for some kind of information being logged in your log files, and if <info> NEVER contains a <newline> character, and if <info> in some of those line contain the string WORD, then the following might do what you want:

Code:

awk -v from=$(date -u -d '-15 min' +'%Y%m%d%H:%M:%S') -v debug="$debug" '
BEGIN {	m["Jan"] = "01"; m["Feb"] = "02"; m["Mar"] = "03"; m["Apr"] = "04"
	m["May"] = "05"; m["Jun"] = "06"; m["Jul"] = "07"; m["Aug"] = "09"
	m["Sep"] = "09"; m["Oct"] = "10"; m["Nov"] = "11"; m["Dec"] = "12"
}
split($1, mdy, /[-<]/) == 4 {
		mod_date = mdy[4] m[mdy[3]] mdy[2] $2
		if(debug) printf("mod_date=%s, from=%s, $1=%s, $2=%s\n",
			mod_date, from, $1, $2) > "debug.out"
}
mod_date >= from
' test.log | grep -m1 -C3 'WORD'

Note that this is untested because the date utility on my system does not have a -d option; AND, you didn't supply any sample data that would produce any output from your sample input.

The way this is coded, if the <info> data in your input files does contain multi-line data, it will copy the entire <info> field to the output if the 1st line meets the date requirements as long as no line in your input data contains exactly three < and or - characters in the first field of a line that is not in the specified format.