[2714] in Release_7.7_team
Re: In case further clarification is necessary.
daemon@ATHENA.MIT.EDU (Garry Zacheiss)
Fri Apr 20 18:04:35 2001
Message-Id: <200104202204.SAA17421@leo.mit.edu>
To: Bill Cattey <wdc@MIT.EDU>
cc: ghudson@MIT.EDU, amb@MIT.EDU, release-team@MIT.EDU, afs-contacts@MIT.EDU
In-Reply-To: Your message of "Fri, 20 Apr 2001 21:25:59 -0000."
<0us_ZbZz0001BDpcsf@mit.edu>
Date: Fri, 20 Apr 2001 18:04:30 -0400
From: Garry Zacheiss <zacheiss@MIT.EDU>
>> WHAT? This looks to me like a bonafide Release 9.0 show stopper.
I'll point out that we do already have a Redhat 7.0 based alpha
release with working AFS. What Andrew meant is that the most recent
kernel available for Redhat 7.0 is 2.2.19, which does have some AFS
problems. We're using 2.2.17-14 for the alpha release, which works just
fine with our current AFS. The reason that an upgrade to 2.2.19 is
desirable is that it fixes a local root exploit in the kernel.
These are what I consider our options to be:
1.) Continue to use Redhat 6.2 for 9.0. This isn't really an option;
the people who've been clamoring for Redhat 7.x since it first came out
will have us tarred and feathered.
2.) Use Redhat 7.0 with 2.2.17-14 kernel. This is what we're doing
now. It works fine; it just means we can't upgrade the kernel
until the AFS work is done to make 2.2.19 work.
3.) Make AFS work with Redhat 7.0/Linux kernel 2.2.19. This is probably
less work than using a 2.4 kernel, but will still require us to go
to OpenAFS.
I believe either option 2 or 3 locks us into using the 2.2 kernel series
for the lifetime of the 9.0 release.
4.) Use OpenAFS with Redhat 7.1/Linux kernel 2.4.2-2. There were
patches posted to openafs-devel earlier today that make this
possible, although I haven't tested them yet.
5.) Convince IBM to support the Linux 2.4 kernel series on Redhat 7.1.
I believe that 4. would be ideal if we can make it happen, but that
2 is a tenable fallback plan, as the work needed to use 2.2.19 is likely
to occur before we have a public release of Athena 9.0.
Garry
