Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 24 Jun 2012 09:34:43 -0700
From:      Kevin Oberman <kob6558@gmail.com>
To:        "J. Hellenthal" <jhellenthal@dataix.net>
Cc:        Adam Vande More <amvandemore@gmail.com>, Greg Byshenk <freebsd@byshenk.net>, "freebsd-stable@freebsd.org" <freebsd-stable@freebsd.org>, "Leonardo M. Ram?" <martinrame@yahoo.com>
Subject:   Re: fsck_ufs running too often
Message-ID:  <CAN6yY1vVq7sB6uU%2BegsNRRXpoAciOcGprEnGF7SLiX0yYKia%2Bw@mail.gmail.com>
In-Reply-To: <20120624160631.GA80121@DataIX.net>
References:  <1340401637.32116.YahooMailNeo@web113519.mail.gq1.yahoo.com> <3729A720-2C8F-4C84-B05A-201394C40D63@gothic.net.au> <20120624013624.GG24842@portland1.byshenk.net> <20120624025451.GA17721@DataIX.net> <CA%2BtpaK2q_9=cp6tQ__H9U9UDvbvmocF3sgc7jx_yYjRqALv01A@mail.gmail.com> <20120624160631.GA80121@DataIX.net>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, Jun 24, 2012 at 9:06 AM, J. Hellenthal <jhellenthal@dataix.net> wrote:
>
>
> On Sat, Jun 23, 2012 at 10:15:19PM -0500, Adam Vande More wrote:
>> On Sat, Jun 23, 2012 at 9:54 PM, Jason Hellenthal <jhellenthal@dataix.net>wrote:
>> >
>> > At one point it was proven that background fsck was not benefitial.
>>
>>
>> Where can we find this "proof"?
>>
>
> It was in the lists amongst many conversations.

I think the term 'proof' is being abused here. Proof is rigorous,
demonstrable and precise. It is not a non-specific coalescence of bit
and pieces in a mail thread.

My experience with background fsck is that its benefit is entirely
dependent on  the requirements of the system. A few things are
entirely clear and easily demonstrated. (Note, this is an assertion
and not proof.)
1. Background fsck takes considerably longer to run than fsck
2. fsck places a very substantial I/O load on the system, primarily on
the disk undergoing the check
3. Other system activity that requires I/O to the disk being checked will slow
4. Background fsck is limited and, when it cannot correct some
problems can impact system stability, especially in the event of power
failure
5. For systems with large file systems it will make the system
available for use MUCH sooner than would be the case if a standard
fsck was used

As with almost all system performance issues, background fsck requires
looking at its benefits and costs. For some, it is a big win. For
others it is a losing proposition because a server is rendered
effectively useless for an extended period.

Full disclosure: I have disabled background fsck on most of the
systems for which I am responsible, but not all.
-- 
R. Kevin Oberman, Network Engineer
E-mail: kob6558@gmail.com



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAN6yY1vVq7sB6uU%2BegsNRRXpoAciOcGprEnGF7SLiX0yYKia%2Bw>