Is it safe to do recursive compression with tar, gzip and pigz?

Question

If I make a .tar.gz via

tar czvf - ./myfiles/ | pigz -9 -p 16 > ./mybackup.tar.gz,

Can I safely unzip an already gzip'd file ./myfiles/an_old_backup.tar.gz within the ./myfiles directory via

gzip -d mybackup.tar.gz
tar -xvf mybackup.tar
cd myfiles
gzip -d an_old_backup.tar.gz
tar -xvf an_old_backup.tar

? And can one do this recursive compression safely ad infinitum?

Answer 1

tar czvf - ./myfiles/ | pigz -9 -p 16 > ./mybackup.tar.gz

is the equivalent of

tar cvf - ./myfiles/ | gzip | pigz -9 -p 16 > ./mybackup.tar.gz

You're not obtaining a tar.gz but a tar.gz.gz, an archive compressed twice, not a compressed archive of compressed files.

That is pointless. Compressed output is not compressible. You won't get any significant space gain by compressing twice. And for extracting, you'd need to decompress twice as well with

gunzip < mybackup.tar.gz | gunzip | tar xf -

Or

gunzip < mybackup.tar.gz | tar xzf -

If you want to use pigz instead of plain gzip for the compression, just do:

tar cvf - ./myfiles/ | pigz -9 -p 16 > ./mybackup.tar.gz

Which you can uncompress with tar zxvf mybackup.tar.gz

Also note that you should never have to uncompress a tar.gz file and store the uncompressed version on disk. The whole point of compressors like gzip/pigz, bzip2/pbzip2, xz/pixz is that they can work on streams, you just insert them in a pipeline.

Answer 2

If your question can be rephrased as "is it OK to have compressed archives within compressed archives?", then the answer is "yes".

This may not be the most convenient (as you note, you will have to run tar several files to get everything unpacked), and applying compression to data that has already been compressed may not yield an additional reduction in size, but it will all work.