From: andreas.koenig@anima.de (Andreas J. Koenig) Subject: Re: CPAN is getting too big To: Elaine -HFB- Ashton Cc: packrats@history.perl.org Date: 04 Feb 2000 17:03:59 +0100 >>>>> On Mon, 10 Jan 2000 12:37:14 -0600, Elaine -HFB- Ashton said: > Andreas J. Koenig [andreas.koenig@anima.de] quoth: > *> > Ah, gotcha. How much disk space would it need at the outset? I could put > *> > *>Less than 2 GB. > I can handle that. :) Turns out to be much less: ~ftp/pub/backpan% du -sk * 1126602 authors 12 lost+found 1506 modules 235437 ports 2724 src The bad news: ports/ and src/ are only random not systematic collections. The obvious: modules/ is bogus in that context. It is a symlink tree that tries to help point&click, but has no meaning for an archeological collection. The grain of salt: missings. The backup system made a few hickups over the time and a few files may be missing if they had a short life on CPAN. The critical time zones are June 1999, December 1999, January 2000. I have deleted all symlinks and all CHECKSUMS files in the authors/ tree as well as the authors/0* files. I believe the really interesting part is only authors/id/ directory. I'm now running the following cronjob to keep the tree up to date: rsync --exclude CHECKSUMS -vrptgx /home/ftp/pub/PAUSE/authors/id/ /home/ftp/pub/backpan/authors/id/ Note the absense of -l, I believe symlinks are no good in there. I do not have access to a good backup system anymore, so I'd be glad if you would not rely on me and would take over the further maintainance. The whole thing is available at ftp://pause.perl.org/pub/backpan rsync: pause.perl.org/backpan Let me know if I can help sorting things out, Hope you like it, -- andreas