File compression to a different dir dynamically

[HP System Handle Owner]

Can anyone tell me if there is a way of dynamically compressing files to another filesystem using either compress or gzip WITHOUT having to cp the uncompressed file to the target dir first.

i.e

I start with
/mountpoint1/file1.dbf

I want to end up with
/mountpoint1/file1.dbf
and
/mountpoint2/file1.dbf.Z

It is for performing a disk backup, but I do not want to copy the file to mountpoint2 and then compress it. I would like the compression program to read /mountpoint1/file1.dbf as a source and create /mountpoint2/file1.dbf.Z as an archive in it's compressed size leaving /mountpoint1/file1.dbf intact.

Thanks in advance

Russ

We are running 10.20

[John Carr_2]

cd /newfile_system

compress /explicit_path/filename

this should put the compressed file into the directory you are located in when you run the command , I cant test this at moment but you could with some small file :-) John.

[Robert-Jan Goossens]

Hi

Take a look at this thread, and the answer from Kenneth Platz.

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=3212

Regards,
Robert-Jan

[Pete Randall · Accepted solution]

Russ,

From looking at the man page, it appears that gzip will do what you require:

gzip -c /mountpoint1/file1.dbf > /mountpoint2/file1.dbf.Z

I would suggest trying a little test scenario first.


Pete

[Mark Grant]

Using the "-c" switch with "gunzip" causes output to go to the standard output so you can do "gzip -c myfile > /somewhere/else/completely/newfile.z"

[John Palmer]

gzip -c /mountpoint1/file1.dbf > /mountpoint2/file1.dbf.gz

As you are talking about database files, be aware that the 'standard' gzip doesn't handle files > 2Gb though.

Regards,
John

[Jean-Louis Phelix]

Hi,

I confirm what Pete wrote ... You can even read from standard input to be sure that compress will not open your file directly :

cat /mountpoint1/file1.dbf | compress > /mountpoint2/file1.dbf.Z


Regards

[John Carr_2]

Hi

you can use -c option with compress command as well as gzip :-) John.

[Massimo Bianchi]

why not

gzip -9 -c /ABSOLUTE/filename > /ABSOLUTE/BACKUP/filename.gz

[HP System Handle Owner]

We do have source files over 2Gb. Does this mean using compress ?

I guess it discounts gzip. We only have standard gzip

Cheers

Russ

[Pete Randall]

Russ,

Get the gnu version:

http://hpux.cs.utah.edu/hppd/hpux/Gnu/gzip-1.3.5/


Pete

[Mark Grant]

Actually, I believe that modern gzip >= version 1.3 does support files larger than 2GB.

[Massimo Bianchi]

With standard unix utilities you can do the following:


mkdir /tmp/pipe p

dd if=/tmp/pipe of=/ABSOLUTE/BACKUP/filename.gz

gzip -9 -c /ABSOLUTE/filename > /tmp/pipe


Massimo

[Mark Grant]

Did you perhaps mean "mknod" and not "mkdir" there Massimo :)

[Massimo Bianchi]

geee... looks like i must stop writing while sleeping :)

Thanks for the correction!

Massimo

[Steve Lewis]

If you cat the file and pipe it to gzip or compress, then redirect the standard output to the destination, you avoid large file problems.

cat dbfile.dbf | gzip > /backup_dir/dbfile.dbf.gz

[ I think they really meant mkfifo, which can be used in conjunction with gzip/compress and tar to copy compressed versions of big files from a remote system. If you want to be really flash, you can use a named pipe as an archive device file for your database. ]

[Mark Grant]

Actually Steve,

Massimo's mkdir/mknod tempory mind block is creating a named pipe. "mknod" is the way to create such niceties :)

[Patrick Wallek]

The following will also work:

# gzip < /dir/inputfile > /dir/outputfile.gz

[Steve Lewis]

... quickly looks at mknod man page... so it is! Well I never knew that. Thank you for pointing that out.
/usr/bin/mkfifo appears to be a specific subset of /usr/sbin/mknod that anybody can run, not just the super-user.

[Steve Lewis]

Actually i screwed up again, it lets anyone mknod filename p , 0 points today again I think.