Hi All, I have a requirement something like this..there are set of files with name sample_plant_date..I need to sort in acesnding order based on plant and date filed and then retain only the latest file in the directory for a particular while delete the rest.

First, I am assuming you want a Unix script: #!/bin/ksh # cd <to your directory> # skip the first filename, remove the rest fread=0 ls -1t sample_plant*|while read fname do if [[ fread -eq 0 ]] then fread=1 continue fi rm "$fname" done

Thankyou .But the problem here is , I have thousands of file for different plants and timestamp, I need to retain the latest file for all the plant code. sample_1006_045600.txt sample_1009_045500.txt sample_1011_045400.txt sample_1006_052500.txt sample_1011_025600.txt ie.. my resultant files should be sample_1009_045500.txt sample_1011_045400.txt sample_1006_052500.txt Again here there are only few files but in real scenario , I will be having thousands of files for different plants. Regards, Kaveri

What you are asking for is rather difficult. Can I assume that the plant id is always columns 8 to 11? sample_1009_045500.txt

to make sure i understand, you only need the latest file for EACH plant, not the latest file in the directory?

#!/usr/bin/ksh for var in $(ls *.txt | sort -t_ -k1,2|awk -F_ '{print $1"_"$2}'|sort -u);do keepfile=$(ls|grep $var|tail -1) cp $keepfile $keepfile.keep done rm *.txt for var in $(ls *.keep);do mv $var $(echo $var|awk -F. '{print $1"."$2}') done The above will delete the non-oldest files immediately, I am not sure what you meant by "retain only the latest file in the directory for a particular while". Since this is deleting files, when you test, make sure you create a test directory, dont use live data.

Excellent code .Thankyou :).Please me know if I can rate you somewhere as I am new to UNIX forum.