Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 8 Aug 2006 20:01:09 +0100
From:      Freminlins <freminlins@gmail.com>
To:        "Atom Powers" <atom.powers@gmail.com>
Cc:        freebsd-questions@freebsd.org
Subject:   Re: Large File System?
Message-ID:  <eeef1a4c0608081201w3dc7c040m8743c69a52e78c84@mail.gmail.com>
In-Reply-To: <df9ac37c0608080907r10f9f41fhba9f50f5f354a307@mail.gmail.com>
References:  <df9ac37c0608071513n7fdfc928r9d1aa3b9e8edfaec@mail.gmail.com> <ygfzmegm97g.fsf@dominion.borderworlds.dk> <df9ac37c0608071636j2f28bc6en66e985d528e8d85f@mail.gmail.com> <eeef1a4c0608080117r22d05716x1e623dd6ee2cf573@mail.gmail.com> <df9ac37c0608080907r10f9f41fhba9f50f5f354a307@mail.gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On 08/08/06, Atom Powers <atom.powers@gmail.com> wrote:

> What exactly does a journaling file system give you? As I understand
> it, it doesn't prevent corruption and it doesn't help you fix the
> corruption when it occurs.


As answered by Dan Nelson. It saves time (sometimes a lot) in the event of
an unclean shutdown/

File server. It's a Promise VTrak M300p, SCSI attached storage.
> Frankly I'm more worried about the system crashing than the storage
> device ( UPS, battery pack for the RAID controller, redundant power
> supplies ).


Yes, I had all that. It is of absolutely no use in the event of an unclean
shutdown (on FreeBSD). If the file system itself is dirty, it will need to
fsckd. The bigger the file system, the longer it takes (generall). That is
what journalling saves you.

To give you some indication of what this means in real life, I'll refer
(again, sorry) to a power outage we suffered in our colo. This is FreeBSD on
modern hardware:

Jul 23 17:52:05 weeble kernel: WARNING: /var was not properly dismounted
Jul 23 17:55:52 weeble fsck: /dev/aacd0s1f: 1352 files, 956469 used,
13988364 free (1484 frags, 1748360
blocks, 0.0% fragmentation)

I've snipped out the logs in between. But that's nearly 4 minutes to get
itself sorted out. That file system has only 1.9GB of data. Our Solaris
boxes came up straight away.


2.5TB file system, used mostly for archival storage. Of the initial
> 2.0TB of data, I expect only about 2-3% to change on a weekly basis.
>
> I have another 1.5TB Fiber Channel cabinet that I plan on using to
> store roaming profiles ( for MS Win uers) and home drives (for *nix
> users).


Depending on the size of the files, you may wish to use a different block
size.  Also I've found that on large file systems you may wish to change is
the "minfree" setting. You can do this either when running newfs, or tunefs.
The default setting is 8%. On 2.5TB file systems that's a lot. If you don't
know what this means check out the man pages.

That is about the only "management" you will need to do.

Perfection is just a word I use occasionally with mustard.
> --Atom Powers--
>


Frem.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?eeef1a4c0608081201w3dc7c040m8743c69a52e78c84>