On Mon, 30 Jan 2006 13:30:35 -0500 (EST)
der Mouse <mouse@Rodents.Montreal.QC.CA> wrote:

> Then, for reasons not very relevant, I created three files (with dd,
> reading from a tape device).  Two of them were 250G, one slightly
> smaller (somewhere in the 220G to 230G range).  Then I deleted them.
> rm deleted them all in less than second.  This ran counter to my
> experience dealing with large files, to the point that it made me
> suspicious.  So I ran df.  I still had a pre-rm df output on my screen,
> and the difference in free space before and after was on the order of a
> few tens of megabytes, nowhere near .7 terabytes.
>
> This made me very suspicious, so I unmounted the filesystem - nothing
> was using it at the time - and ran fsck.  fsck found lots of errors.

I have been following the thread, interested in the outcome.

This indeed seems broken, not that I can tell where the problem could
be.  I am curious though, was this done with soft dependencies enabled? 
In any case, with softdep, a file deletion could appear spontaneous, but
df summary would still normally be updated and a few seconds after the
operation the changes would normally be applied...

Even before the softdep queue is flushed, unmounting the file system
should cause the softdep managed metadata to be flushed, as well as a file
system sync, which seems to occur properly with the smaller file systems
I have here...  I'm wondering if softdep could be part of the code paths
in your tests.

Then possibly to eliminate the write buffer cache code paths, I guess
that another test could be done with the file system mounted
synchroneously, possibly...  Unless you're already confident that neither
softdep nor the buffer cache could be involved in the bug of course

I unfortunately don't recall if you're now trying with 3.0+ or -current,
the last thread seemed to be "2.0 and >2TB filesystems", unless I missed
yet another thread between these two (I just did a quick scan of my local
archive).  It would definitely be interesting to know if this also occurs
with a test -current system.

Thanks,
Matt

-- 
Note: Please only reply on the list, other mail is blocked by default.
Private messages from your address can be allowed by first asking.