Shell Programming and Scripting

Hi,

I need some advise on whether there is a better way of doing what I am currently planning to do. Perhaps I should be using arrays instead of re-directing output to files?

I need to use a tool/program named ADRCI provided by Oracle to remove trace files that it generates. Honestly it is not the best one to use but unfortunately the nature of the job calls for "WE MUST USE IT".

Anyway, the tool has issues where even if I say purge files that are x-number of days old, it still was not able to delete some of the very old files. I am guessing that it may be checking something internally in the database and hence not recognizing these very old files.

In this scenario, am ending up with files that are old and has to be deleted but not getting deleted.

At the moment, what I am looking at doing is as below:

Code:

01 - Run ls -1tr and get a file listing
02 - Run Oracle's ADRCI which will remove some files
03 - Run another ls -1tr
04 - Remove all files that are listed before the first file removed by ADRCI, retain the ones from the last file removed by ADRCI

Not sure if I stated it clearly. To illustrate what am wanting to do.

For example, if step 01 gives a list as below that I placed in x1.txt

Code:

 
 
x1.txt:
 
file1
file2
file3
file4
file5
file6
file7
file8
file9

- Then Oracle's ADRCI run and after I run I do another ls -1tr and redirect it to a file named x2.txt that has the following list:

Code:

 
 
akl1db922(oracle)[]/nas_mnt/common/dba_utils/scripts/tmp$: diff x1.txt x2.txt
4,6d3
< file4
< file5
< file6

 
x2.txt:

file1
file2
file3
file7
file8
file9

- This means, Oracle ADRCI remove the files named file4, file5, file6. And that also suggest that all files before file4 are old and not recognized by Oracle ADRCI anymore. Hence they can be deleted

- What am thinking is to use the diff command and then cat/awk/head/sed, am confused to get the list of files that I want to remove

Using diff for these two files, I get the result below:

Code:

 
 
$: diff x1.txt x2.txt
4,6d3
< file4
< file5
< file6

- Looking at the result, I only need the first line, 4,6d3, and then the first number which is 4 and subtract one from it. I then use cat/awk/head/sed x1.txt then run the rm on these files

- Using head, I get the list below which are the files that I want to remove

Code:

 
 
$: head -3 x1.txt
file1
file2
file3

Please advise if there is a better way of doing what I am intending to do. Should I use awk/sed instead of diff'ing and head'ing?

Thanks in advance.

The following seems to do what you want:

Code:

awk 'FNR==NR{f[FNR]=$0;next}
{       if(f[FNR] != $0) {
                end = FNR
                exit
        }
}
END{    if(end > 1) {
                printf("echo rm ")
                for(i = 1;i < end; i++)
                        printf(" '\''%s'\''", f[i]);
                printf("\n");
        } else  printf("echo \"ADRCI did not remove any files.\"\n")
}' x1.txt x2.txt | sh

Try it and if it looks like it identifies the correct set of files, change the first printf call in the END section from:

Code:

printf("echo rm ")

to:

Code:

printf("rm ")