chflags on zfs (sappnd)

[George Mamalakis]

On 17/11/2010 21:25, Markus Gebert wrote:
> On 17.11.2010, at 20:00, Andriy Gapon wrote:
>
>> on 17/11/2010 18:38 Markus Gebert said the following:
>>> On 17.11.2010, at 11:49, George Mamalakis wrote:
>>>
>>>> Hi everbody,
>>>>
>>>> from http://wiki.freebsd.org/ZFS I understand that chflags are supported by zfs. But if I have a file with sappnd on a zfs filesystem, I am unable to execute a command like this:
>>>>
>>>> # touch lili
>>>> # chflags sappnd lili
>>>> # ls -lrto lili
>>>> -rw-r--r--  1 root  wheel  sappnd 5 Nov 17 12:38 lili
>>>> # echo "lala">>  lili
>>>> # echo "lala">>  lili
>>>> -su: echo: write error: Operation not permitted
>>>>
>>>> So, the first time it worked, but it stops working on any consequent time (when the file is no more empty).
>>>> I found a bug report on:
>>>>
>>>> http://www.freebsd.org/cgi/query-pr.cgi?pr=149495
>>> And:
>>>
>>> http://www.freebsd.org/cgi/query-pr.cgi?pr=151082&cat=kern
>>>
>>>
>>>> where a patch is suggested. Nevertheless, even though my sources are newer than the suggested patch, my source tree does not contain it.
>>>> Do we know anything more about it?
>>> The fix was committed to CURRENT on Oct 8th (r213634). The commit message talks about MFC after a week, however to me it looks like the fix hasn't made it to 8-STABLE yet.
>>>
>>> Also, there seems to be a related commit r213673 which essentially reverts r213634 and has a more general approach to handling ioflags with zfs. This commit was on Oct 10th, again 1 week MFC grace period, again not in 8-STABLE yet, at least to my knowlegde.
>>>
>>> Maybe MFC was simply forgotten, maybe there's a good reason to delay it, I don't know. We're using my patch (the one mentioned in the PRs) for now and append-only works as intended. Applying the changes in r213673 to 8-STABLE might be an option too, if you're considering patching to get the append flag working.
>>>
>>> Anyway, hopefully one of these fixes gets MFCd to 8-STABLE soon.
>>>
>> Good analysis, but did you forget to CC the committer(s)?
>> It is known that sometimes the committers do need a gentle (or not so) nudging :-)
> Right :-) mm@ and pjd@ now CCed.
>
>
> Markus
>
>

cheers guys :-)

-- 
George Mamalakis

IT Officer
Electrical and Computer Engineer (Aristotle Un. of Thessaloniki),
MSc (Imperial College of London)

Department of Electrical and Computer Engineering
Faculty of Engineering
Aristotle University of Thessaloniki

phone number : +30 (2310) 994379