UNIX for Dummies Questions & Answers

I want to compress all the files which are three years older ..I have thousands of files...

1) This doesnt work
find ./ -type f -mtime +1176 -print | xargs -n1 -i tar -cvf {}
Errror
tar: Missing filenames

Probably because of -
find ./ -type f -mtime -1 -print returns -
"
./temp.txt"
First line is empty, may be that's the problem.

To experiment, i just tried -
"find ./ | grep "txt" | xargs -n1 -i tar -cvf {}"

"find ./ | grep "txt" |" returns no null characters. Still it's NOT working !!

2)
tar -cvf find ./ -type f -mtime +1176 -print

tar -cvf `find ./ -type f -mtime -1`

none of these working.

I get output as -
"a .// 0K
a . 0K
a .//temp.txt 0K
tar: .//find same as archive file
tar: -type: No such file or directory
tar: f: No such file or directory
tar: -mtime: No such file or directory
tar: -1: No such file or directory
tar: -print: No such file or directory"

3) The only thing working is -
"find ./ -type f -mtime +1176 | gzip -c > temp.gz"

It creates temp.gz but I am not sure -- WHETHER THIS WILL WORK WITH THOUSANDS OF FILES OR NOT?

one more problem is i cant unzip the files -
" gzip -dc temp.gz"
Returns me-
"./temp.gz
./find
./temp
./temp2.txt"

but no unzipping happens ! I am stuck !

Envrionment details -
SunOS 5.8

Code:

find ./ -type f -mtime +1176 -exec gzip {}\;

That won't zip them at the exact same time, but it will filter through the results and zip them up one at a time.

The other point is that tar(1) on its own does not compress it simply concatonates files together which will save wasted space if the files are smaller than the blocksize of the filesystem and would reduce the number of inodes used, but will not compress like gzip(1)ing them would, unless this is an extended version of tar that supports the -z option of course!

The find -type -f mentioned by cbo0485 means that find does not list directories only files.

@ cbo0485.

First of all thanks to you...

I didn't get you actually.. Is your code (quoted ) going to process alll the files' list? Do you mean my code will get only one file compressed? m Bit confused .

The problem is with exec I believe..I got thousands of files.. (Total size 50 Gigs) I think, this code will still get me the same errors. "Too many arguments". I think, xargs is required over here.

The bottom line question is, I got thousands of files, which I can obtain using some find command. Now I want to compress them all.

(Just an added info -All the files under same directory - on the same mound point. None of the file has funny characters like *,&,etc)

@Tony,

Thanks for ellaborating on tar & gzip. It was definately some value addition! I got the point about type -f (no directories).

The problem is the same old thing -

I want to compress list of files at a time. I got thousands of file , so -exec is running short to handle that large no of argument.

I can first copy the name of files into a temp file & use it with gzip, but why to do so if i can do it in one line command ? (if I can do so). The volume is such large that it will take 1/2 hours just to list all the required files !

Please help me to get though this problem.. Plz post the syntax /code as well.

Thanks,
Kedar

If you have a "too many files" problem then break it down so that you put a certain number of files into a tar file, gzip the tar file, delete those files then repeat: I would script it with something like this:

Code:

i=1
DATESTR=`date '+%m%d'`
EXIT="no"
cd <desired_directory> && \
while [ "${EXIT}" = "no" ]; do
  find . -type f -mtime +1176 | head -16000 > /tmp/filelist
  if [ -s /tmp/filelist ]; then
    tar cvf tarfile.${i}.${DATESTR}.tar `cat /tmp/filelist` && \
    gzip tarfile.${i}.${DATESTR}.tar && \
    sed -e 's/^/rm /' /tmp/filelist > /tmp/rmfiles && \
    sh /tmp/rmfiles
  else
    EXIT="yes"
  fi
  i=`expr $i + 1` 
done

I have not been able to test this (not having thousands of files to hand!) so I suggest putting an "ls" in place of the "rm" to start with and you may have to reduce the 16000 to get the tar to work.
I hope you do not have filenames with spaces in!
If this were running on Solaris I could do tar cvf tarfile -I /tmp/listfile and filenames with spaces in would not be a problem...

That's because you are missing the /path/to/tarfile/tarfile.tar argument to the tar command.

Your command needs to be like:

find blah blah blah | xargs blah tar -cvpf /path/to/tarfile.tar {} \;

Also, if you have thousands of files, then try to group them, and tar.gz those groups instead of individual files. Less management headache.

Thanks to you both of you guys : frozentin & TonyFullerMalv. I am feeling bit reluctant (actually lazy) to go for looping & scripting, if we can do it with a single command line...


@frozentin
What is significanse of " \;" in above find command?

Regards,
Kedar