I have 40000+ files in one directory and I need to FTP them to another machine. I cannot even get an ls of them unless I use file globbing

i.e ls file[12]*

If I do an ls or an mget is shows 0 files. I can transfer the files like this by globbing using mget file[12]* but the file names are all very similar and have thousands of numbers associated with them, and so this method takes forever. The machine is a production server and I do not want to recompile the kernel to have more command-line argument memory. Is there a ftp program that can handle this many files or a script that I could use to ftp these files?

i dont know about that quantity of files but you can use ftp's mget (or mput)

if you are ftp'ing TO the server to download the files from
then once in the directory with all the files, use mget to download them all without interaction (the ftp -i turns off prompting when mget/mput multiple files so that it will run down the list without asking you to confirm that you want to get each file

again i have never tried to download that many files from a single ftp directory before so could not tell you that it will work in your particular situation, but worth a shot none-the-less

hope this helps...

Hi gvec,

The problem is that there is so many files that mget cannot cope and hangs or doesn't tranfer any files. From what I have read this is to do with the command-line memory that is compiled into the kernel.

I have a work around solution that uses perl and the Net::FTP package. Perl doesn't have the memory constraint like the command line does so file globbing is no problem.

I think this is the best solution as ftp programs use local commands such as 'ls' which don't work due to the number of files.

The best method is to tar the files and transfer the data in a single stream. if you do not have enough free space you will probably have to use a file system to copy the files. the ftp protocol is not very good at transfering many files usually bailing out at a few thousand.
Though you might also have more luck with scp, but if it was me i'd just mount a nfs or smb directory and perform a simple copy.

Tim1235:
You said "I have a work around solution that uses perl and the Net::FTP package. Perl doesn't have the memory constraint like the command line does so file globbing is no problem.

How do you do it? Net:FTP doesn't implement "mget", but only "get".

I am writing a script where I want to mget all files starting with a specific id, like:
"mget 310*"

Any ideas?

You can use ''glob"

glob uses pattern matching operators like ? and * as well as multiple searches i.e glob "pattern1 patern2 pattern3"

Then you can loop through this array of files and use get or put.