Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 17 May 2010 12:41:17 +0300
From:      Nikos Vassiliadis <nvass9573@gmx.com>
To:        crwhipp@gmail.com
Cc:        freebsd-questions@freebsd.org
Subject:   Re: File system
Message-ID:  <4BF10F3D.2070207@gmx.com>
In-Reply-To: <1a7012fe7affe8caf4263d4d2c385614.squirrel@whipp.no-ip.org>
References:  <u2z768631271005081836k26590481qcaab03601799448d@mail.gmail.com>	<4BE84825.9060005@gmx.com> <1a7012fe7affe8caf4263d4d2c385614.squirrel@whipp.no-ip.org>

next in thread | previous in thread | raw e-mail | index | archive | help
Craig Whipp wrote:
> On Mon, May 10, 2010 10:53 am, Nikos Vassiliadis wrote:
>> Ansar Mohammed wrote:
>>> Hello All,
>>> I have a FreeBSD VM running. Whenever I reboot the VM without a clean
>>> shutdown it boots into single user mode and I have to run fsck.
>>>
>>> When I run fsck, the file system clearly has issues.
>>>
>>> Is there any way to have FreeBSD run on a better file system that wont
>>> crap
>>> out on me everytime I do and unclean shutdown?
>> I am really surprised no one proposed geom journaling. With gjournal,
>> I never had to do a manual full fsck and have had plenty of unclean
>> shutdowns. I also occasionally do fsck the filesystem and there were
>> no errors ever found. It definitely adds the ease factor I am looking
>> for in a journaling sollution in the case of an unclean shutdown...
>>
> 
> Correct me if I'm wrong, but since this FreeBSD install is running inside
> of a VM, in addition to any of the precautions suggested here to get data
> written or journaled to the disk as safely as possible, isn't there still
> the issue of whether the VM actuall commits these writes to the physical
> disk?

I guess the time needed for some data to be committed to stable
storage will be bigger in a VM environment. But that's always the
case, be it a VM, or a disk controller. There will be always some
data in-flight, some delay and a cache which will hold your data before
they arrive to stable storage. gjournal will replay all write attempts
(metadata and data) before the failure, so you should be relatively
sure that all writes are done correctly. I think Ansar just want to
avoid fsck and gjournal provides that. To ensure real data integrity
one should use something else, perhaps ZFS and not a journaling fs.

PS: I didn't see your message in time...

Nikos



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4BF10F3D.2070207>