#86 Issue closed: Inodes in fs creation reset to default¶
Labels: bug
wdpoorte opened issue at 2012-05-30 11:35:¶
During recovery, the number of inodes in a file system is reset to the default instead of using the original value.
gdha commented at 2012-06-01 12:44:¶
this depends on file system type in use and after the mkfs we cannot
tunefs it anymore. I find it pretty hard to decide if this is a bug or
not. Missing feature perhaps?
Btw by extending the LVOL you increase the inodes too. Is this not a
valid work-around??
wdpoorte commented at 2012-06-01 13:13:¶
Hi,
I don't think this is a valid workaround. I am in particular concerned
about file systems with a lot of small files. For these it might be
necessary to change the bytes/inode ratio (-i option in mkfs.ext3) to
prevent a half full file system with all inodes consumed.
Werner.
dagwieers commented at 2012-06-02 23:59:¶
@wdpoorte I agree we should in fact record the max number of inodes of each filesystem in the layout.conf, and restore that during the creation of the filesystem. That's the only safe option we have. That is why I marked your report as a bug. However I think we have higher priority issues to fix at this time, but we welcome pull-requests or discussions about this functionality.
wdpoorte commented at 2012-06-04 15:46:¶
I'll consider creating a pull request - a good opportunity to learn
git.
Then I can also create a pull request for the lvmvol dash issue if
needed.
dagwieers commented at 2012-06-05 15:40:¶
@jhoekx @wdpoorte I wonder if some of the other possible options to mkfs.ext2 should be recreated in the same manner ? Possible options are: fragment-size, blocks-per-group, number-of-groups, inode-size, journal-options, features
I don't mind if we decide that all or some of these are on a request basis, but I want to make sure that we at least looked at them and make a conscious decision.
[Export of Github issue for rear/rear.]