NANDFS eats itself up

[Boris Astardzhiev]

Sorry for that, I haven't noticed it. I hope that this will do.

The attachment - http://justpaste.it/1kk5


On Wed, Nov 28, 2012 at 11:06 PM, Ronald Klop
<ronald-freebsd8 at klop.yi.org>wrote:

> On Wed, 28 Nov 2012 16:36:11 +0100, Boris Astardzhiev <
> boris.astardzhiev at gmail.com> wrote:
>
>  If I do that I might be unable to set back the verbose level to 0 since
>> the
>> output is very very NOISY. This means that I will have to start again from
>> the beginning if I need to reproduce it. Nevertheless here you go. Check
>> the attachment (OUTPUTNAND.txt.bz2).
>>
>
> Your attachment is stripped by the mailinglist.
>
> Ronald.
>
>
>
>  On Wed, Nov 28, 2012 at 5:13 PM, Mateusz Guzik <mjguzik at gmail.com> wrote:
>>
>>  On Wed, Nov 28, 2012 at 05:06:12PM +0200, Boris Astardzhiev wrote:
>>> > Yes, I do. I've made this to repeat via a script. The interval between
>>> each
>>> > SCP transfer is 100s. I thought it would give just enough time to the
>>> fs
>>> > for reclaiming its space back.
>>> >
>>> > Now I stopped the transfers and doubled the vfs.nandfs.cleaner_segments
>>> to
>>> > 10. The fs has no files in it but it reports "~9.6MB" of used space. 4+
>>> > hours later it has NOT changed at all it still reports "~9.6MB" Used.
>>> >
>>>
>>> Now this indeed sounds like a bug. Can you enable debug like this:
>>> # sysctl vfs.nandfs.verbose=0xffffffff
>>>
>>> The kernel will start printing a lot of debugging information. With
>>> default configuration of syslog this will end up in /var/log/messages.
>>> Capture something like 30 seconds of output and post it somewhere.
>>>
>>> --
>>> Mateusz Guzik <mjguzik gmail.com>
>>>
>>