Change of historical behavior

Tom "spot" Callaway tcallawa at redhat.com
Sun Apr 29 13:25:29 UTC 2007


On Wed, 2007-04-25 at 18:19 -0600, Dax Kelson wrote:
> Just a FYI...
> 
> While updating some courseware I noticed a change in decades old UNIX
> behavior on RHEL5/FC6. The behavior may not be super important, but I
> wouldn't be surprised if somebody, somewhere gets bit by it.
> 
> The change is as follows.
> 
> Previously with disk quotas enabled if your hard block quota was reached
> (or you were over your soft and the time had expired) but your
> file/inode quota was not reached, you could still create empty files.
> 
> This is no longer the case. Observe:
> 
> $ quota
> Disk quotas for user guru (uid 500): 
>      Filesystem  blocks   quota   limit   grace   files   quota   limit   grace
>       /dev/sda8    2048*      0    2048              11       0       0        
> $ touch /tmp/newfile
> touch: cannot touch `/tmp/newfile': Disk quota exceeded
> 
> My initial suspicion was that this is a side effect of RHEL5/FC6
> mounting all filesystems with the acl and user_xattr options (done via
> the default mount options field in the filesystem's super block),
> however with testing this appears not to be case.
> 
> This came up because the lab exercise I was updating had a step along
> the lines of "...now that you are 'over' quota run the command "touch
> anewfile". Can you explain why that still works?...".

Not to ask stupid questions, but should that have ever worked? I'd think
an outright rejection would be far useful than creating empty files
innocently.

Maybe this is a release notes item for F7?

~spot




More information about the fedora-devel-list mailing list