gareth bsd at lordcow.org
Mon May 15 07:07:26 PDT 2006

On Mon 2006-05-15 (15:46), Oliver Brandmueller wrote:
> OK, I was not clear enough: During normal operations what's on the disk 
> and the view of the system to the filesystem are not necessarily the 
> same - this is especially true for open files. No matter how long it 
> takes for fsck to run, a r/w opened file will almost ever be in an 
> inconsistent state.
> Again: fsck is not for r/w mounted filesystems (except with -B for 
> filesystems that support it - namely ONLY UFS2!). If you use fsck in 
> traditional mode you will get unexpected results. And although fsck 
> tries very hard to keep you from breaking things badly, you have a good 
> chance to damage your filesystems if you use it improperly (no, I won't 
> try if -y or -f will force a check on a mounted partition - I still need 
> my filesystems).
> Again, very loud and clear: DON'T DO THIS.
> After telling you, fsck is not suuposed to be run on r/w mounted 
> filesystems in that way, I guess we can agree there's no room for a 
> discussion like that, OK? I mean, you don't use the hammer for screws,
> do you?

ok, i understand that it's not meant to be run on a mounted FS. i just used
to do that on my (linux) FS's non-evasively as a rough check that everything
was still ok. thanx, i see now that it can give unexpected results, i was
just initially interested in why it was showing this now, whereas before it
wasn't (in a roughly correlated way with the increase in power failures here).

More information about the freebsd-stable mailing list