Re: dump memory footprint

[christos%astron.com@localhost (Christos Zoulas)]

In article <20130321135300.GC12078%homeworld.netbsd.org@localhost>,
Emmanuel Dreyfus  <manu%netbsd.org@localhost> wrote:
>Hi
>
>I have a problem with dump memeory footprint. I would like to dump
>a filesystem containing 34 kB, and dump will not do it unless I setup
>around 10 MB of swap:
>
># ls -loa /flash/                                                              
>total 251561
>drwxr-xr-x   3 root  wheel  -            512 Mar 21 13:46 .
>drwxr-xr-x  18 root  wheel  -            512 Mar 21 13:11 ..
>-r--r--r--   1 root  wheel  nodump     75220 Mar 20 14:39 boot
>drwxr-xr-x   6 root  wheel  -            512 Mar 21 13:10 etc
>-rw-------   1 root  wheel  nodump 117514598 Mar 21 13:28 netbsd
>-rw-r--r--   1 root  wheel  nodump  11118529 Mar 21 13:27 netbsd.old.gz
># du -ks /flash/etc/                                                           
>34      /flash/etc/
># dump -h0 -0uf /dev/null /flash/ 
>  DUMP: Ignoring u flag for subdir dump
>  DUMP: Dumping sub files/directories from /flash
>  DUMP: Dumping file/directory /flash/
>  DUMP: Found /dev/rwd0a on /flash in /etc/fstab
>  DUMP: Date of this level 0 dump: Thu Mar 21 13:47:38 2013
>  DUMP: Date of last level 0 dump: the epoch
>  DUMP: Dumping a subset of /dev/rwd0a (a subset of /flash) to /dev/null
>  DUMP: Label: none
>UVM: pid 874 (dump), uid 0 killed: out of swap
>Killed                                        
>
>It seems rather unreasonable to allocate more than 10 MB just to dump 34 ko;
>Are there any relevant options to reduce dump memory footprint?

Dump needs to read the whole filesystem and build data structures for it;
it has not been optimized for space. Tar is not good enough?

christos