[6546] in Release_7.7_team
Re: Move early-linux into proposed?
daemon@ATHENA.MIT.EDU (Jonathon Weiss)
Thu Dec 3 22:49:32 2009
Message-Id: <200912040349.nB43nOEX013263@vorpal-blade.mit.edu>
From: Jonathon Weiss <jweiss@MIT.EDU>
To: Jonathan Reed <jdreed@MIT.EDU>
cc: Evan Broder <broder@MIT.EDU>, release-team@MIT.EDU
In-reply-to: Your message of "Thu, 03 Dec 2009 21:58:52 EST."
<A6A2F69E-26EF-4FC7-BC9D-BFDD2F4E1EF7@mit.edu>
Date: Thu, 03 Dec 2009 22:49:24 -0500
X-Spam-Flag: NO
X-Spam-Score: 0.00
I'm a little concerned by "should be able to be re-installed at any
time". That is a significant change to the contract for early, which
has always been the final phase of testing (after beta). Frankly, my
understanding is that it used to be that it was extremely rare (I
think it happened, but not more than once or twice) for a beta machine
to have to be re-installed under athena 9 (and earlier). OTOH, In the
Athena 9 model we essentially never did early testing on patch
releases, only on full releases (having to do with the fact that early
got their packs from the athena AFS cell, just like public). If you
really think these machines need to be able to be re-installed at any
time, perhaps we should move them into the beta, or better alpha,
cluster.
All of that siad, I think internal to IS&T testers who I assume are
more likely to report bugs (and in at least some cases better able to
write a coherent bug report) may be more valuable to you than random
cluster workstations. OTOH, maybe I'm just cynical in thinking that
users will just go to the next machine over and not report problems on
the test machines.
Jonathon
> I'm not opposed to this idea, but I'd like to come up with some
> signage we can deploy in the clusters before we actually turn it on.
> Also, removing private workstations is probably good, particularly if
> we're changing the meaning of the early-linux cluster. However, if
> the private workstations are within IS&T (which they probably are), we
> should consider sending mail to the moira contacts explaining what
> we're doing, and giving them the opportunity to continue participating
> in the early-linux cluster (with the caveat that the machines should
> not be their primary workstations and should be able to be reinstalled
> at any time).
>
> -Jon
>
> On Dec 3, 2009, at 3:58 PM, Evan Broder wrote:
>
> > Can we move the early-linux cluster into -proposed instead of
> > production?
> >
> > Right now we have a very small handful of machines pulling updates
> > from -proposed. Many of that very small handful aren't even regularly
> > used, and fewer are cluster machines. I think it would be useful to
> > expose -proposed to a wider array of machines, hopefully so that we
> > can more effectively catch bugs before they get pushed to production.
> > Also, this would allow us to tell users to go check a particular
> > machine to see if we've successfully fixed their bug.
> >
> > You can see the list of machines in early-linux by running `qy -s -f
> > machine gmcm '*' early-linux` (I won't spew the output at you).
> > Currently there are a handful of private machines that we should
> > possibly remove, and only a handful of cluster machines. I think we
> > should add to early-linux roughly the set of machines that
> > participated in the Debathena Beta deployment (so a few in m12 and a
> > few in m56).
> >
> > This obviously introduces the danger of needing a runaround if we
> > incapacitate the machines, but uh...hopefully we can avoid that? In
> > any case, it's pretty easy to hit w20, m12, and m56 if we need to.
> >
> > Opinions?
> >
> > - Evan
>
