I don't think RudiC's suggestion is going to work since you want separate counts of the number of lines containing each of the three strings, not a total of number of lines that contain one or more of the strings.

Since you have specified that the name of the log file to be processed is the empty string (not the name of a file), jim mcnamara's script won't work either. And, if you did provide the name of your log file, getting the last 200000 lines three times (while additions are being added to the log file by other processes) means that it will run slow and give you counts for the three strings from different sets of lines in your file. It is also dangerous to expand a variable without double quoting it if there is ANY possibility that it could contain whitespace characters, characters that are special in pathname expansions, or (as in this case) an empty string.

I would be tempted to try something more like:

Code:

#!/bin/bash

foldername=$(date +%Y%m%d)
LOG_NAME="/path/to/your/logfile"
mkdir -p  /root/kpi/"$foldername"

tail -200000 "$LOG_NAME" | awk -v ts=$(date +%Y/%m/%d/%H/%M) '
	/ACCESS_REQUEST/{ a++ }
	/ACCESS_ACCEPT/	{ b++ }
	/ACCESS_REJECT/	{ c++ }
	END 		{ printf("%d,%d,%d,%s\n", a, b, c, ts) }
' >> /root/kpi/"$foldername/kpi_$foldername.csv"

This only reads the log file once and gathers individual counts for each string, but you still have to fill in the real name of the log file this script is supposed to process.

And, as always, if you want to try this on a Solaris/SunOS system, change awk to /usr/xpg4/bin/awk or nawk.