Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 5 Jan 2006 01:37:27 -0800
From:      Jo Rhett <jrhett@svcolo.com>
To:        Peter Jeremy <PeterJeremy@optushome.com.au>
Cc:        stable@freebsd.org, current <current@freebsd.org>
Subject:   Re: FreeBSD Update is the binary update solution [Re: HEADS UP: Release schedule for 2006]
Message-ID:  <20060105093727.GK1358@svcolo.com>
In-Reply-To: <20051223045648.GH77268@cirb503493.alcatel.com.au>
References:  <43A266E5.3080103@samsco.org> <20051217215434.GB92180@svcolo.com> <20051217220807.GA28741@freebie.xs4all.nl> <43A492B6.6050305@t-hosting.hu> <20051217232856.GT77268@cirb503493.alcatel.com.au> <43A4B91D.8040304@samsco.org> <20051222211730.GK39174@svcolo.com> <20051223045648.GH77268@cirb503493.alcatel.com.au>

next in thread | previous in thread | raw e-mail | index | archive | help
> On Thu, 2005-Dec-22 13:17:30 -0800, Jo Rhett wrote:
> >But FreeBSD Update suffers from all of the same limitations that I've been
> >describing because of lack of integration with the Core OS.
> >
> >1. modified kernels are foobar
> >  ..yet are practically mandatory on production systems
> >
> >2. modified sources are foobar
> >  ..yet many common production situations require source compilation options
 
On Fri, Dec 23, 2005 at 03:56:48PM +1100, Peter Jeremy wrote:
> So you want to be able to make arbitrary changes you your source code
> and compilation options and then expect the FreeBSD project to provide
> a tool that will apply binary fixes to the resultant executables?
 
No.  I want a binary update mechanism.  Obviously if we have local
configuration options we'll have to compile our own binaries.  But doing
the work of tracking system updates currently requires us to build our own
patch tracking mechanism, and then re-write it for every new release.

If the core OS supported the necessary features for handling patches to the
core, then we could stop building all these tools ourselves and focus on
real work.

> I don't know that modified kernels are mandatory.  A lot of work has
> been going on to reduce the need to re-compile the kernel for common
> situations.  Likewise, I don't know that "many common" and "require"
> go together - IMHO, 'desire' or 'would be nice' are more appropriate
> descriptions.  Would you care to provide some details of what you
> believe can be done to reduce your need to re-compile the OS.

It's not a recompile issue.  It's a tracking/update/backout issue.  I don't
mind recompiling, if I could somehow push the updates to all the right
systems and they would have it stored somewhere that they were patched...

> I'm not sure that FreeBSD Update is totally unusable for you.  If you
> have the situation where you have a modified FreeBSD that you need to
> roll out to a large number of hosts, you just need to have your own
> FreeBSD Update server - you test the changes in your environment and
> then roll them out as you require.

I've looked it over, and our current mechanism works better than
freebsd-update at the moment.  Always subject to change.
 
> I don't run jails so I'm not familiar with the problems here.  Maybe you'd
> like to explain the problems you run into.
 
It's really just the same problem.  Some mechanism to find out what is, and
store what has been done.

-- 
Jo Rhett
senior geek
SVcolo : Silicon Valley Colocation



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20060105093727.GK1358>