quotas problem on 4.11/UFS
bde at zeta.org.au
Fri Mar 24 13:48:11 UTC 2006
On Fri, 24 Mar 2006, Nicolas KOWALSKI wrote:
> Bruce Evans <bde at zeta.org.au> writes:
>> IIRC, quota maps are sparse, with uid N mapped to offset
>> N*sizeof(somestruct). With 32-bit uids, N can be as large as
>> 4294967295, so the file size wants to be (this+1)*sizeof(somestruct) =
>> some not very large multiple of 4GB. The large file sizes for this
>> might even work, without wasting much disk space, since files can be
>> sparse too, but there may be overflow problems at 4G.
> Many thanks for this explanation.
> However, the filesystem does not have any files belonging to some user
> with a large or -2 uid. I checked it, and even "quotacheck -v" does
> not show anything like that; our current uids goes from 0 to 6000 max,
> and only these appear in the repquota result.
Perhaps it had them but there were none when you checked. I think the
quota file doesn't shrink if slots at the end of int become unused.
Files with a uid of -2 are created on nfs clients if root is not mapped
and root creates a file. I see quite a lot of them due to having a
world-writeable /c/tmp directory and using it as root on the client.
More information about the freebsd-fs