Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 15 Sep 2018 12:37:46 -0700
From:      Lee Brown <leeb@ratnaling.org>
To:        freebsd-geom@freebsd.org
Subject:   Re: geli - why do I need a keyfile
Message-ID:  <CAFPNf59nWkyodPnnCifytshCVafGNg3JTnYDcHgL6GvZ=sXmVQ@mail.gmail.com>
In-Reply-To: <20180915201819.50ac10a3@gumby.homeunix.com>
References:  <CAFPNf588xqRYZRoCACr2n_NyfMsMvrXPR4_DjWy4evBY_1HaAQ@mail.gmail.com> <20180915201819.50ac10a3@gumby.homeunix.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, Sep 15, 2018 at 12:18 PM, RW via freebsd-geom <
freebsd-geom@freebsd.org> wrote:

> On Fri, 14 Sep 2018 17:55:58 -0700
> Lee Brown wrote:
>
> > I want to create a geli provider as authentication only, no password,
> > no encryption.  I do:
> ...
> > Instead:
> > # echo " " > /tmp/key
> > solves that issue, but I still don't get why I even need a key file
> > with -e NULL?
>
> Because HMAC itself needs an encrypted secret key, otherwise anyone
> could write to the device without it being detectable.
>
> Without a securely entered passphase, or a passfile on removable media,
> HMAC doesn't provide any authentication, it just detects bitrot and
> naive attempts to modify the filesystem.
>
> Thanks for the explanation, in retrospect I should have read up on HMAC.

That's precisely my use-case data integrity verification only.  I'm
building a RAID1 gmirror on top of 2 geli providers, so if a disk rots it's
detected.  Now I just need to test how the gmirror reacts when the
underlying geli faults.

Much appreciated -- lee



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAFPNf59nWkyodPnnCifytshCVafGNg3JTnYDcHgL6GvZ=sXmVQ>