Monday September 05, 2005 Weblog
Monday September 05, 2005 my experiments with reiserfs - part1
As I don't have a spare disk on my system, I decided to create reiserfs on a 64Mb USB memory stick.
As I have had prior exposure to solaris-ufs, I thought, having a journal or not is a mount time option. But it turned out that, whether I eventually mount it with log or nolog, mkreiserfs sets aside space for log. So with default journal size, I was left with less than 50% of the stick capacity for my use. I thought this is too much and then re-read the man page, and figured there is this '-s' option which sets journal size. So I recreated the FS with '-s 513' (as per the man page min is 513), but this time mount fails to mount the stick saying its not 'reiserfs' partition; which has come as a surprise
as mkreisefs didn't complain at the time of FS creation.
I would have loved to see an option to mkreiserfs to avoid journal. This could have been small disk friendly...
So I didn't have a choice but to settle with less than 50% of disk space for my files.
debugreiserfs(1) is a definite hackers love. Very nice command if you are curious about where (to the granularity of 'byte') exactly is data/metadata.
As the disk space left is very less, reiserfs didn't bother to use internal nodes to store the tree. root block itself is a leaf node, where the directory entries and stat data is available. Whereas I wanted to see the 'formatted internal nodes' in action as I populate the FS.
Looks like my only choice is to get a bigger stick!!! or are there any work arounds ??
( Sep 05 2005, 05:21:44 AM PDT ) Permalink Comments [2]
