Hello, I am making a script to backup all my information on the server and send it to my email. I tried recently, but it didn´t let me because the resulting targz file was bigger than the maximum file size allowed. In order to solve that problem I figured out on making several targz files of a predefined size.
How can I find out or estimate the size of the tar file of that directory before taring it?

Each type of file will compress differently. That is to say that ASCII files will compress very well while JPEG file will not compress at all. So if you know how much space is occupied by each type of file then you can make a formula. For example: (space of ASCII files)/12 + (space of JPEG files) + (space of binary executables)/2.

Or you could use the multi-volume feature of tar which will split the tar file up into parts of a size that you specify. So you could just tell tar that the maximum file size of a partial tar file is 12 MB and it will split the tar output into 12 MB files.

Or you could make one large backup file and use the split utility to split it into segments of a size that you specify. It makes more sense to force tar to split up the archive file, though, so that you don't have to recreate the single archive file in order to restore files from the archive.

Of course it's a little bit suspicious that you want to use email to transport the files to their destination. I hope that you aren't copying things from work. Use your power for good and not for evil.

Not easily, at least not with any accuracy. Gzip compression ratios vary very widely depending on what is to be compressed. English text can be compressed a lot - the compressed size might be 10% of the original. However, mp3 or jpeg files won't get compressed very much, if at all.

Why do you want to know. Why not just make the file, and then use split to chop it into chunks, and then cat the together at the other end?

stress_junkie you beat me to it.

I hate when I'm the one to post second. Oh well. I guess it happens to us all once in a while.

Its even more difficult than this. Attachments will need to be MIME-encoded, so will be inflated by a certain percentage during this process. Be sure to take this into consideration.

Generate the file anyway, then you'll know how big it is and if it's too big you can use split(1) to send it in pieces.

Note the maximum file size, use tar to create multivolume, each volume less than maximum file size. Send.

Thanks to all of you for your information. I didn´t know tar command has a multi volume feature and it would certainly be very useful to me, would you tell me how to use it?
I prefer this rather than using the split command because I don´t have to restore small files to get the original one. I am also intrested in this because I want to make a zip utility in the future just to find out how things work.
I know my question might sound suspiciuous, but I´m a trainee in IT and I have a lot of hard work and information on the server of all my subjects and I have wondered that it is possible that my information could be unavailable or accidentaly deleted. Anyway, due to permision restrictions I can only backup my own information.
Thanks

see tar's man page. Also google for "tar multi volume". you should be able to find something useful

You might want to look at rsync, which will stream your data across, then on subsequent uses, only sends the changes ie a much smaller amt.

if i am not wrong, OP needs to send to his email.

I was wondering whether he NEEDS to use email or just did that way because it seems the obvious/easy way if he hasn't messed much with this level of usage before.
Depends where the target system is...

use the info page instead for tar, you'll find out why...


Something to consider, tar will increase the initial file size. Using the standard block size (512), will produce a 512b header for each file or directory and will round each filesize upto 512. Then it will also want to be a minimum of a multiple of blocks, which if I remember correctly is 20 as default...

For example if you have a single file of 3 bytes:
header: 512
file: 512
= 1024
yet it must be a multiple of 20 blocks, so fill with zeros until 10240 bytes.

result = 10240 bytes

OP should have read the man page by now and know how to go further